We Ranked Every AI Tool for Product Managers — So You Don’t Have To

Aakash Gupta 0:00 Today, we're breaking down the top product management tools of 2025, and we'll be ranking them all tier ranking style. Everybody is talking about Claude Code right now.

Anshumanni Rudra 0:09 There are times where I'd have 6 terminal windows open on my large screen, and it'll be Claude Code running in all six of them, doing different things on different parts of my directory structure.

Aakash Gupta 0:18 If a PM is trying to figure out their roadmap, what would be the best way to test out these tools?

Anshumanni Rudra 0:24 Pick the tools that specifically will help you in your workflow, uh, and then experiment with them, then figure out if they actually work for you and they're improving your overall productivity.

Aakash Gupta 0:32 Today, I'm with Anshumanni Rudra, one of the most legendary product leaders in India and Singapore. He was the VP of Product at Hotstar. Now he is a Group Product Manager at Google in charge of all of APAC Payments. Lovable has crossed 120 million ARR in less than a year. What's the grade here we're giving for Lovable?

Anshumanni Rudra 0:51 Lovable is a solid A. The only reason I won't give it an S is everything made using Lovable starts to feel the same.

Aakash Gupta 0:58 What is the single best AI tool for PMs?

Anshumanni Rudra 1:01 I think it's-

Aakash Gupta 1:05 Really quickly, I think a crazy stat is that more than 50% of you listening are not subscribed. If you can subscribe on YouTube, follow on Apple or Spotify podcasts, my commitment to you is that we'll continue to make this content better and better. And now on to today's episode. Anshumanni, thanks so much for being on the podcast.

Anshumanni Rudra 1:26 Uh, my pleasure. It's, it's, uh, it's, it's been, uh, a long time pending, but, uh, it's good to finally be on your podcast. I'm a big fan.

Aakash Gupta 1:34 Yes. When it comes to product leaders in your region of the world, when I ask people who we need to have on the podcast, they always mention your name, so the honor is mine. I have been seeing you tweet, post on LinkedIn about all sorts of AI tools, so I thought there would be no better person in the world to create a tier ranking of AI tools.

Anshumanni Rudra 1:57 Yep.

Aakash Gupta 1:57 Do you think we can do that?

Anshumanni Rudra 1:59 Absolutely. Absolutely. Let's

Anshumanni Rudra 2:01 dig in.

Aakash Gupta 2:01 Awesome. So the type of tool that everybody is obsessed with these days is AI Agents. So I brought together some of the AI Agent tools here. There's n8n, there's Make.com, there's Lindy, there's Airtable, there's Zapier, there's Relay. Let's go ahead and work through each of these and-

Anshumanni Rudra 2:20 Yeah

Aakash Gupta 2:20 ... take a look at them and give them a ranking. How would you describe n8n for folks?

Anshumanni Rudra 2:26 I think the way to think about n8n would be, um, you know, you have a chain of things to do. You have a, you have a set of tasks to do, and you want to put all of them together in a very visual flow. So, you know, the way I think about it is if, if you, if you have kids and you've ever taught them Scratch programming, right? Uh, it's, it's very much like that. So building agents in that way, in this sort of really interesting flow way of doing things, right, like A leads to B leads to C, uh, I think that's, that, that's a really smart way of people thinking about their processes and people thinking about how they are gonna automate and doing the work. And I think n8n does a really good job of that.

Aakash Gupta 3:02 Likewise. I think it's gotta be the most powerful agent-building tool out there, but I think it has a little bit of a technical barrier. It's really geared towards more of an engineering audience or a very technically proficient audience. So where are we putting this on the tier list ranking for product managers?

Anshumanni Rudra 3:23 I think for PMs who have a very strong technical background, uh, they might enjoy using a tool like this. But for a lot of other PMs who just want to quickly prototype stuff and might want to just test things out and see if this kind of an agent works for them, uh, this might not be a great first tool to pick up. Right. So, uh-

Aakash Gupta 3:43 So are we thinking it's like a B?

Anshumanni Rudra 3:45 I'd say, yeah, a B, a higher grade B. Yeah.

Aakash Gupta 3:48 [laughs]

Anshumanni Rudra 3:49 Yeah. A B+ in my opinion.

Aakash Gupta 3:50 All right. Yeah, it's a B+, but it's stuck in that B range.

Anshumanni Rudra 3:55 Yeah.

Aakash Gupta 3:55 All right. So Make.com is making a lot of hype right now. This is really another version, kind of like n8n. I would say it's slightly easier to use. It's, uh, got as many integrations, which is really powerful. A lot of enterprise are using it. What's your take on any-- on Make.com versus n8n?

Anshumanni Rudra 4:17 Uh, I've heard good things. You're absolutely right. I think in terms of integrations, I think which data sources are available to you, which of your, uh, products are available to you, uh, to add is gonna be a key part. And I think integrations will make it work, but I haven't used it heavily.

Aakash Gupta 4:33 Yeah. I think it's just that one tier below n8n. And the reason that it does that is just because it's still technical, um, but it's just not as powerful as n8n. N8n has even more integrations and functionality. So we'll go ahead and give Make.com a C. So Lindy.ai is one of those tools that I pay for myself. I've talked about it in other podcasts, that this is amazing. I have an email responder, a meeting prep assistant. I turn podcasts into blog posts. I have an email negotiator, and I'm gonna be creating more agents. What's really cool is they recently just released, um, not just a flow editor, but an agent builder where you can prompt it via AI. So I think Lindy is the version of n8n that every PM really wants because it's really easy to use and you can just prompt an AI. So I would say Lindy is either an S-tier or an A-tier tool. Where do you think it lands?

Anshumanni Rudra 5:31 Yeah. I think anything which allows you to build agents and using the power of generative AI, right? So you can actually chat with it and then figure out what should your agent look and feel like, as opposed to figuring out the bones of it, feels like it's more powerful.

Aakash Gupta 5:46 100%. So Lindy, we're gonna give it our first S-tier ranking. We're not gonna give out too many of those.

Anshumanni Rudra 5:52 Yeah.

Aakash Gupta 5:53 All right. So next up we have Airtable. Have you been using-- What have you heard about Airtable?

Anshumanni Rudra 5:59 I've used Airtable pre-AI, uh, which is, uh, which... And it was always, uh, a standout tool back then. I'm, I'm hearing great things about it in its new AI transformation, for sure.

Aakash Gupta 6:11 Yeah. I think that, uh, Airtable AI, they have a bunch of very specific things built for product managers, and it leverages a lot of the power of Airtable. So I think it's an incredibly good tool. Um, it has like literally tens of thousands of integrations. There's so many real companies actually building on top of Airtable for production-grade workflows that I feel like it has to be an A grade.

Anshumanni Rudra 6:37 Mm-hmm. Mm-hmm.

Aakash Gupta 6:38 Airtable gets the A.

Anshumanni Rudra 6:40 Mm-hmm.

Aakash Gupta 6:42 On to the last one, Relay.app. So this was actually built by a former director of product at Google, Jacob Bank. He was leading Gmail for a long time, and it is a very easy-to-use agent builder. It's not super advanced. It doesn't have the AI to talk to. It doesn't have as many integrations as N8N, but it's really easy to use and you can build like a 40-agent marketing team, which is actually what he's done straight in the tool. So where does this land?

Anshumanni Rudra 7:11 I think probably B.

Aakash Gupta 7:11 I feel like it's one tier below-

Anshumanni Rudra 7:13 Yeah

Aakash Gupta 7:13 ... Airtable and Zapier. All right, so we are gonna give Relay in the B category. All right, so that g- concludes our AI agents. So what's the best AI agent tool for PMs? I think we both agree it's Lindy.

Anshumanni Rudra 7:26 Yes.

Aakash Gupta 7:26 Awesome. So let's move into AI

Aakash Gupta 7:28 prototyping. Obviously, everybody has been talking about AI prototyping. Lovable has crossed 120 million ARR in less than a year since founding. What's, what's the grade here we're giving for Lovable?

Anshumanni Rudra 7:43 Lovable is a solid A. The only reason I won't give it an S is everything made using Lovable starts to feel the same, starts to look the same. You know, it's-- everything uses the Shade CN library. Uh, you know, if, if you use the default Tailwind CSS React front end, it, it, it all sort of starts... Every app that you build with it looks very beautiful but looks similar, right? And, uh, I think that's a huge opportunity there for someone to actually build really good UX libraries, uh, UI component libraries. Uh, you know, Shade CN folks have done such a phenomenal job. Uh, you know, it's, it's the most standout, uh, you know, component library, but we need more. We need a lot more.

Aakash Gupta 8:28 Yeah. I found that Lovable, when I've been testing it head to head against a lot of these tools, it's good, but it's not amazing. Like, it gets the job done of AI prototyping, but it doesn't necessarily make me feel like it's built with incredible craft or something like that. People saw my dashboard. I have live Lovable sites where I've bought domains for them.

Anshumanni Rudra 8:49 Mm-hmm.

Aakash Gupta 8:49 I've tried all sorts of things, and ultimately I feel like this lands in the B category. It's AI prototyping tool, but it feels like it's probably just mainly a wrapper on Claude Sonic. What do you think?

Anshumanni Rudra 8:58 Yeah. I think B would be fine because if you think about it, what it doesn't do is, for one, it doesn't give you a very good structure to work with, right? So, uh, something that maybe would stand out perhaps for Replit, for example, is that it helps you build a plan before you actually start building. Lovable just pushes out the first thing with the very first prompt that you give it, right? And it sort of looks decent. And it, it, it's never-- It doesn't do a good job of building backends as well.

Aakash Gupta 9:25 Yeah. It's not very good at backend yet. Hopefully, the team and Elena Verna will improve on that.

Anshumanni Rudra 9:31 Yeah.

Aakash Gupta 9:31 All right, so the next tool that everybody's been talking about, this one also passed 100 million ARR recently, is Bolt.

Anshumanni Rudra 9:40 Bolt.

Aakash Gupta 9:40 Um, built-in import from Figma-

Anshumanni Rudra 9:42 Figma

Aakash Gupta 9:42 ... GitHub. I have plenty of Bolt projects. In fact, I just recently launched, uh, my cohort, and that was through Bolt.

Anshumanni Rudra 9:52 Excellent.

Aakash Gupta 9:52 So it works. You can live build things. It's pretty fast. Sometimes it runs into issues. What's the grade we're giving Bolt? How does it compare to a Lovable?

Anshumanni Rudra 10:02 I think it's definitely higher than Lovable, uh, mostly because I think it also does a very good job of building more complex apps and more complex end-to-end products. Does a better job of helping you think through what the backend of your product would be. Um, if you're building pure sort of, you know, solo webs-- uh, solo apps, which only you're going to use and multiple people are not going to use and, you know, it's, it's local storage, Lovable is great, right? But if the moment you want to sort of build something which multiple people need to use, which will have, you know, a, a proper storage system, then you need, you need something like a Bolt or even better.

Aakash Gupta 10:39 Totally. So I think Bolt hits that A-tier category.

Anshumanni Rudra 10:43 Yep.

Aakash Gupta 10:44 The next tool that everybody's been talking about is Magic Patterns, and I've been using this one personally a lot. Also have launched an app through Magic Patterns. You know, it can clone LinkedIn, it can recreate a screenshot. The coolest thing about Magic Patterns is that it is only front end, so it's really, really fast. So it's purpose-built for AI prototyping. It's not built for vibe coding. But I do think that's a bit of a deficiency, having vibe coding. So I'm inclined to put this in the B category. Does that sound right?

Anshumanni Rudra 11:17 That sounds about right. Uh, I think it's good to have tools which are very specific to a problem statement that like, "Hey, build great front ends," and that's about it. And that's good. That's a good problem to solve, right? I don't think everyone should try and say that, "Hey, we, we'll help you build end-to-end products," because sometimes you need a really good front end, and then you can go ahead and, you know, use another tool or use something else to build the back end. So that's absolutely fine.

Aakash Gupta 11:42 Okay. The next app that everybody's been talking about is v0. I've done a lot of bake offs where I, like, run all of these AI prototyping tools against each other, and what I love about v0 is after Magic Patterns, it's always the one to finish.

Anshumanni Rudra 11:57 Yeah.

Aakash Gupta 11:57 And it actually does backend, and it's backed by Vercel infrastructure, so it's pretty easy to just-

Anshumanni Rudra 12:03 Yeah

Aakash Gupta 12:03 ... publish-

Anshumanni Rudra 12:04 Correct

Aakash Gupta 12:04 ... something altogether.

Anshumanni Rudra 12:06 Yeah.

Aakash Gupta 12:06 So it's very important.

Anshumanni Rudra 12:07 It's very easy. Yeah. It's very easy to deploy products on Vercel, right? Uh, so anything that you make with v0 and you actually want to put it in the hands of a few users, even if it's just friends and family members, it's-- I, I think v0 does a much better job.

Aakash Gupta 12:21 Yeah. So I think v0 probably lands in that Bolt A-tier category, right?

Anshumanni Rudra 12:25 Yeah.

Aakash Gupta 12:27 It's a A-tier tool. And then base44. So this is the tool-- This is the mythical tool, the one-man, $80 million exit. Actually insane that he did this, Meir Shlomov. What is our take on base44?

Anshumanni Rudra 12:45 I haven't used it much. I'll defer to you.

Aakash Gupta 12:48 So I've used it, uh, quite a bit. I love it. I actually think it's, like, one of the best at back-end and actually creating an app. So I'm inclined to put this in the A-tier category as well.

Anshumanni Rudra 12:59 I concur. Yeah.

Aakash Gupta 13:01 So the hot question then is we have three tools in the A-tier category-

Anshumanni Rudra 13:06 Yeah

Aakash Gupta 13:06 ... from this AI prototyping. We put Bolt, v0, and base44. If we had to crown a winner, what is the best AI prototyping tool right now?

Anshumanni Rudra 13:17 I think right now, I would say, for me at least, Bolt is slightly ahead, right, by a few points, uh, mostly because it has a better structure. You know, when you start off, it by default thinks about both the front-end and the back-end in a better way. Uh, allows you to build a little bit more advanced, uh, apps from the get-go. So in my book, I would say Bolt is a little higher for prototyping.

Aakash Gupta 13:44 Okay. I think it's neck and neck. We have a difference of opinion for the first time. Anshumanni giving Bolt the best AI prototyping tool title. I will give it to v0, just because in a lot of these head-to-head bake-offs that I have, Bolt runs into, like, errors-

Anshumanni Rudra 13:57 Yeah

Aakash Gupta 13:57 ... sometimes, and doesn't seem to be able to dig itself out of those errors fast. v0, it's like, almost never runs into errors. I never have to hit, like, fix this. I did a demo video. The Bolt CEO was on the podcast, and we probably spent like thirty minutes just debugging-

Anshumanni Rudra 14:12 Oh

Aakash Gupta 14:12 ... and I don't think that happens as much with v0.

Anshumanni Rudra 14:15 That's true.

Aakash Gupta 14:17 All right. So Replit. What would you tell us about Replit?

Anshumanni Rudra 14:22 I think for me, of all the tools that I've used to build sort of end-to-end products, uh, apart from actually using an IDE or using something on my desktop itself, but web-based tools, I think Replit takes the cake for me. Uh, it is-- I think the first critical thing, especially as a PM, is that you want a little bit of a structure before you get into building anything. So the moment you start giving Replit prompts, it says, "Hey, let's figure out what we are going to build," before it actually starts writing any piece of code. And I think that one single thing is probably what makes Replit far stronger than every other tool, which is that it jams with you. It comes up with a proper plan and only then gets into the execution phase. Um, what I'm also able to do now with Replit is that it has multi-agent facility, so you can actually work on different aspects of your product. You can, you know, while you are bug fixing on one end, your o-other agent could be going through your repository and making sure all front-end bugs are solved. Uh, you could have a third agent working on just making sure that all your, uh, you know, documentation is in order. So I, I genuinely enjoy that a lot. I've built a whole bunch of tools that, uh, you know, for, for my own sake, which I've shared with, uh, uh, with a few of my friends. Uh, and I think Replit, for me, is, is, is my go-to for building end-to-end products which I can quickly deploy. Lots of integrations, really, uh, easy to deploy, uh, your built products as well.

Aakash Gupta 15:48 Awesome. So Replit we really like. The question is, is it an S-tier or is it an

Aakash Gupta 15:54 A-tier? Today's episode is brought to you by Miro. Let me ask you something. How many tools are you juggling just to get a single project across the finish line? One for brainstorming, another for planning, something else for tracking tickets. That's where Miro comes in. It becomes an all-in-one collaboration workspace. Whether you're consolidating user research from several interviews, developing and synthesizing product briefs or a wireframe, or project managing development, Miro brings everyone into the same space. It's fast, intuitive, and fully loaded with features like project templates, two-way Jira sync, and integration with software like Draw.io and PlantUML. Miro's AI features can be used to synthesize elements in a board to develop a ready-to-review product requirements document in seconds. If you're tired of tab overload and scattered workflows, try Miro. Head to miro.com and see why over ninety million users choose Miro to guide from idea to outcome. Today's episode is brought to you by Vanta. As a founder, you're moving fast toward product market fit, your next round, or your first big enterprise deal. But with AI accelerating how quickly startups build and ship, security expectations are higher earlier than ever. Getting security and compliance right can unlock growth or stall it if you wait too long. With deep integrations and automated workflows built for fast-moving teams, Vanta gets you audit-ready fast and keeps you secure with continuous monitoring as your models, infra, and customers evolve. Fast-growing startups like LangChain, Writer, and Cursor trusted Vanta to build a scalable foundation from the start. So go to vanta.com/aakash. That's V-A-N-T-A.com/A-A-K-A-S-H to save one thousand dollars and join over ten thousand ambitious companies already scaling with Vanta.

Anshumanni Rudra 17:44 You have to remember that Replit is overall, even before the whole AI sort of usage came in, Replit was already a very, very strong sort of web-based IDE. So Replit has been doing this really well for a long time. So I think there, there's a bunch of accrued benefits that they have in understanding what developers are thinking about. I would-- I think for me it's, uh, it, it's probably

Anshumanni Rudra 18:05 the highest. Uh, uh, in all the other prototyping tools, I'd probably put it right there at an S.

Aakash Gupta 18:11 All right. So Replit is the real AI prototyping tool winner. We had put it in the vibe coding category, but it spans the AI prototyping category as well.

Anshumanni Rudra 18:20 Yeah.

Aakash Gupta 18:20 It has a very highly agentic agent, I would say.

Anshumanni Rudra 18:23 Yes.

Aakash Gupta 18:23 It can go out and plan and just work for hours and hours, which is something that really does set it apartReplit, we love it. We give it really high reviews. On to Windsurf. So Windsurf, a lot of people ask me, "Is Windsurf even still a thing?" So I like Windsurf. We did a tutorial podcast on Windsurf. It is really good. It's not quite at that Cursor level, I think. Cursor just has slightly more attention for detail, and ever since Windsurf got acquired by the Cognition team, which makes Devin AI, back in July, it seems like Windsurf kind of lost its momentum and its top leader. So I'm not gonna give Windsurf a super high grade. I'm gonna give Windsurf a C. Does that sound right?

Anshumanni Rudra 19:09 Sounds about right.

Aakash Gupta 19:11 All right. So there's Windsurf. The next one, I mean, it's, it's the hottest thing in product management circles right now, Claude Code. Everybody's talking about Claude Code right now. I think that Claude Code is really one of the best tools out there. I upgraded from Claude Pro to Claude Max just because of Claude Code. Because you can put in your entire context, like all of your PRDs, all your meeting notes, all your descriptions. It'll handle all that context. And you-- it's very highly agentic, so you can have it go in and do things, interact with different things-

Anshumanni Rudra 19:50 Yeah

Aakash Gupta 19:50 ... fetch different things, and it'll plan. So overall, Claude Code, I think it might be my single favorite AI tool. So for me, it would be an S-tier. Does that sound right?

Anshumanni Rudra 20:02 Oh, absolutely. Absolutely. No doubt. Cla- Claude Code at this point, in terms of just for writing code, uh, the way it functions, uh, and it's very strange for those of us who grew up in the '90s, who grew up a lot with terminal-based tools, right? And who's, you know, be-before, uh, uh, you know, web-based tools became a thing. Claude Code is like a-- It's such a refreshing thing to go back to, to that level of having control. Uh, and the fact that you're just chatting with it and it's, it's building this great context for you, it's going through your files. I have-- There are times where m-my son walks in and I'd have six terminal windows open on my large screen, a grid of six, and it'll be Claude Code running in all six of them, doing different things on different parts of my directory structure. It's beautiful. I love Claude Code.

Aakash Gupta 20:52 Claude Code is the, probably the best tool here. Um, so we might have to go back and, uh, be harsher on some of those other tools. We'll see.

Anshumanni Rudra 21:00 Yeah.

Aakash Gupta 21:00 So then we have Cursor. Cursor, you know, I actually use Claude Code inside of Cursor.

Anshumanni Rudra 21:07 Yep.

Aakash Gupta 21:07 Sounds like you use it in the terminal, which is-

Anshumanni Rudra 21:09 Yeah

Aakash Gupta 21:09 ... you're bolder than I.

Anshumanni Rudra 21:11 Yeah.

Aakash Gupta 21:11 I always [chuckles] need the IDE around it, but I think it's definitely a step above-

Anshumanni Rudra 21:17 Yeah. Oh, absolutely

Aakash Gupta 21:18 ... Windsurf, but it's a step below Claude Code.

Anshumanni Rudra 21:22 Um, absolutely. I think the, for me, Cursor, because it's, it's sort of, uh, it, you know, it's a fork of VS Code, it's always been a phenomenal IDE, right? There are some stylistic choices that they've made, right, which everyone else doesn't have, and I think I would love to do a large UXR study on why your agent should be on the right side of your screen as opposed to the left side of the screen. Pretty much every tool that we have covered so far, the agent is on the left side of the screen. It really bothers me, right?

Aakash Gupta 21:56 [laughs]

Anshumanni Rudra 21:56 Cursor is the only one where the agent is on the right side of the screen and all your chat s-sticks to the right side, and then you have the rest of the screen to sort of like really focus. Uh, and I know, of course, like on a, on a, on a, on a laptop screen, there is no right hand bias, right? Like, just like phones. But there is a right hand bias for a lot of us in the way we think about products and the way we use products, and I just wish everyone added it as a, as, as a, you know, as, as something that you could just toggle and say, "Hey, just swap these around. I want my preview window on the left side and I want my agent on the right side." And because Cursor has that, my usage of Cursor is actually way higher than others because I just mentally want to be-- have my chats on the right side.

Aakash Gupta 22:40 Okay. So Cursor is an A-tier tool. Seems like we are AI tool fans, 'cause we are clustering in the S and A tier. [laughs]

Anshumanni Rudra 22:47 We, we, we're getting a little bit of a nice, uh, you know, uh, bell-shaped curve going there. Too many on the S tier.

Aakash Gupta 22:53 Yeah.

Anshumanni Rudra 22:53 Heavy on the-- We are heavy on the front, uh, uh, on the top trades, but I think we, we, we are, we are acting as good profs. We are nice, uh, [chuckles] ...

Aakash Gupta 23:03 We're nice. Anybody can advertise with us in the future. [laughs]

Anshumanni Rudra 23:06 Yes.

Aakash Gupta 23:07 We have not-

Anshumanni Rudra 23:09 Being harsh

Aakash Gupta 23:09 ... said your product is terrible.

Anshumanni Rudra 23:11 Yeah.

Aakash Gupta 23:11 So Codex by ChatGPT is next. This is actually how they basically coded most of the recent agent kit that they just rela-launched.

Anshumanni Rudra 23:19 Hmm.

Aakash Gupta 23:20 It's very highly agentic, um, but it doesn't seem nearly as polished as Claude Code for my taste. I think that the agentic capabilities are nice, but Claude Code seems to be just more fun to use with the various verbs that it throws in there, so it gives you lots of regular dopamine hits or its token counter.

Anshumanni Rudra 23:39 Yeah.

Aakash Gupta 23:39 It feels like Claude Code has a little bit more tastefully been built.

Anshumanni Rudra 23:44 Absolutely. Absolutely. I think Claude's focus on code itself, on the fact that even Claude 4.5 or 4.1 are much better at analyzing code and thinking about code. I think Claude's focus on code is very, very clear, and which is why Claude Code also stands out, uh, as a coding agent.

Aakash Gupta 24:05 So I'd-- I'm inclined to give Codex a B. Is that fair?

Anshumanni Rudra 24:10 Yeah. I'd, I'd-- Yeah. Between-- It's a C+, B- kind of a thing for me.

Aakash Gupta 24:16 Yeah. It's right on that edge. Might even be a C tool. Um, yeah. Let's give it a C tool because we've given a lot of high grades. So especially for PMs, you know, it's not gonna be the most important thing, I think-

Anshumanni Rudra 24:29 Yeah

Aakash Gupta 24:29 ... for you.

Anshumanni Rudra 24:30 Yeah.

Aakash Gupta 24:31 It's maybe your developer has a preference for Codex for some reason or another, but for us-

Anshumanni Rudra 24:36 Yeah

Aakash Gupta 24:36 ... we'd say go with Claude Code or Cursor.

Anshumanni Rudra 24:38 Yep.

Aakash Gupta 24:39 So GitHub Copilot, this is what Cursor and Claude Code and Codex are taking all of the market share from. I think that, uh, in my experience using it, that's a-- the developers are smart. They're leaving for a reason. It's just not as good. It's very Cha- you can see this here, too. It's very GPT-focused. It's not enough Claude-focused, and Claude is just a better coding model.

Anshumanni Rudra 25:04 Mm-hmm.

Aakash Gupta 25:04 So I think it really suffers because of that.

Anshumanni Rudra 25:07 Yep. Yep.

Aakash Gupta 25:08 So I think this is our first D-tier tool. It just-- It's not useful for PMs.

Anshumanni Rudra 25:13 No.

Aakash Gupta 25:15 I did talk to a team. I actually gave a talk yesterday at a medical devices company. The only AI prototyping or vibe coding tool they had access to was GitHub Copilot.

Anshumanni Rudra 25:25 Oh, true.

Aakash Gupta 25:25 So for them, we worked through, like, how to use GitHub Copilot for all the AI prototyping and stuff. And so even though we're giving it a D-tier, an AI tool is better than no AI tool, right? [chuckles]

Anshumanni Rudra 25:36 Ab- absolutely.

Aakash Gupta 25:36 AI prototyping is better than not.

Anshumanni Rudra 25:37 Absolutely. Absolutely. It's, it's like, I mean, having a tool essentially by default, even if it's improving your efficiency by 10%, it's still a huge gain, right? But of course, some of the top-tier tools here are multipliers, are true multipliers, right? So you can improve your throughput three X, four X, 10 X.

Aakash Gupta 25:59 Yep. So we gave both Replit and Claude Code in this vibe coding category an S-tier rating.

Anshumanni Rudra 26:05 Hmm.

Aakash Gupta 26:06 If we had to crown one winner, what is the best vibe coding tool? Are we both in agreement it's Claude Code?

Anshumanni Rudra 26:12 It is, because even within Replit, my agent runs on Claude. [chuckles]

Aakash Gupta 26:16 Yes, always. So Claude Code is the winner. Let's move into the big LLMs and

Aakash Gupta 26:24 really need no introduction for anybody. But ChatGPT, of course, is the big LLM that everybody uses. And so what's our take on it lately? You know, we've got ChatGPT, we've got Pro, although I'm, I'm not paying for that right now. Instant. Um, they just released the agent mode. I personally use probably Deep Research the most out of everything because it goes and checks a couple hundred sources. Where-- What's our take on the latest for GPT-5? What would the grade be?

Anshumanni Rudra 26:57 I'm, I'm really glad that they've moved away from keeping the focus on which specific, uh, model you're using and just gone towards... Uh, I think that's a move between all of them, between Gemini, between Claude and everyone else, which is like, hey, thinking versus fast, right? Quickly sort of get people away from like, oh, three point five was better than three point one, or like four point one is better than four point five, um, because people started building preferences towards specific models. So I think one of those things I'm, I'm, I'm... That's one specific thing that I'm really happy about. I think now we are, we are hitting these spaces where as a PM, right, of course, on benchmarks, these things are crossing benchmarks. So like, oh, okay, this is gonna win an in-international maths olympiad, and this is gonna like get a gold in that. But from PM productivity perspective, are your GPTs, personal GPTs becoming better as a result of moving to a better model? And I think we are hitting marginal returns now in some of those spaces with each of these successive models. So I think it's between their fourth and fifth sort of levels with, uh, with GPT-5, uh, with ChatGPT-5. I, I don't see that large a difference from a productivity gain or the fact that there were ab- things that I was able to do which I'm not able to do anymore, right? Deep research is fantastic, but to be truly honest, there's nothing really to choose between deep research if you have access to all of them, right? Uh, and in fact, uh, and we'll talk about Manus a little later. If you can do deep research across multiple things simultaneously, so run multiple deep researches at the same time, which a few general-purpose tools allow you to do, that becomes way more powerful, uh, than running it one at a time.

Aakash Gupta 28:42 Yeah. So I'm inclined to only give ChatGPT a B. Like, everybody's heard of it, everybody uses it, but it's ultimately just-- right now it's just okay, right?

Anshumanni Rudra 28:51 Yeah. And it's, it's to be seen. You know, they, they released a really interesting report on sort of how the seven hundred million users are using it on a weekly basis. It'll be interesting to see now that this is a, this is a widely successful product which has reached 10% of the world, right? 10% of the world's population uses it weekly. How does the next ten or twenty percent come on board? And I think the current way the, the way the product is func- uh, built right now is not what's going to sort of cross that classic chasm of early adopters, and 10% is early adopters, let's be honest, right?

Aakash Gupta 29:26 AI is writing code faster than ever. But can your testing keep up? Testkube is the Kubernetes native platform that scales testing at the pace of AI-accelerated development. One dashboard, all your tools, full oversight. Run functional and load tests in minutes, not hours, across any framework, any environment. No vendor lock-in, no bottlenecks, just confidence that your AI-driven releases are tested, reliable, and ready to ship. Testkube, scale testing for the AI era. See more at testkube.io/aakash. That's T-E-S-T-K-U-B-E.I-O/A-A-K-A-S-H. Today's episode is brought to you by the experimentation platform Kameleoon. Nine out of ten companies that see themselves as industry leaders and expect to grow this year say experimentation is critical to their business. But most companies still fail at it. Why? Because most experiments require too much developer involvement. Kameleoon handles experimentation differently. It enables product and growth teams to create and test prototypes in minutes with prompt-based experimentation. You describe what you want, Kameleoon builds a variation of your webpage, lets you target a cohort of users, choose KPIs, and runs the experiment for you. Prompt-based experimentation makes what used to take days of developer time turn into minutes. Try prompt-based experimentation on your own web apps. Visit kameleoon.com/prompt to join the waitlist. That's K-A-M-E-L-E-O-O-N.com/promptToday's episode is brought to you by Dovetail, the real-time customer intelligence platform. If you're looking to accelerate innovation, prioritize revenue, and do more with less, Dovetail can be your solution. You can connect up all of your customer data so that customer understanding isn't just a quarterly ritual, but feedback is constant. That allows you to match ever-changing expectations and operate like the best teams. Teams like Vercel, Shopify, Whole Foods, Zapier, and Wize all use Dovetail as their AI customer intelligence platform to prioritize roadmaps, innovate, detect product friction, break down silos, and prevent churn. So if you wanna check out Dovetail, grab the link in the description and check out the latest new AI tool that PMs need to master. After ChatGPT, I think in usage and in popularity probably comes Perplexity next, I think. Honestly, it feels like a ChatGPT wrapper with some search that's really good in my mind. I think it can be good for specific things like competitive research and just basically replacing Google. So I'm inclined to put it similar to ChatGPT as a B. Where would you put Perplexity?

Anshumanni Rudra 32:16 My own usage of Perplexity has actually gone down quite drastically. Uh, the early days were very interesting, uh, in terms of how I was using it as a search, and especially their-- they had this wonderful thing where, you know, you'd search for something, and then they'd give you more options to do a deeper search, and you could go down a rabbit hole, um, and really explore a topic. AI mode does that for me right now, so, you know, it's-- my usage of Perplexity has dipped quite drastically.

Aakash Gupta 32:44 Yeah. So is Perplexity a C-tier?

Anshumanni Rudra 32:47 Yeah. For me right now, it's at a C.

Aakash Gupta 32:50 Yeah. Perplexity is at a C-tier. Sorry, guys. We still love you, Arvind. Um, Claude Sonnet, so I think it's obviously an amazing code model. We've already talked a lot about Claude Code. For me, it's the workhorse LLM. It's what I use for all of my writing, all my ChatGPT projects have become Claude projects now. So I think it's quite powerful, and the nice thing is, like, unlike Claude Code, you can just approach it in an easy interface. So I feel like it's something like an A. Is that right?

Anshumanni Rudra 33:19 Yeah, I think it's a... Because of the power of the underlying model, I think Claude for me is, is the de facto, uh, model that I use now. Uh, across, across all workloads, pretty much every agent is defaulted to using 4.5 or 4.1.

Aakash Gupta 33:38 Yeah.

Anshumanni Rudra 33:38 And then for directly using Claude, right, makes a lot more sense.

Aakash Gupta 33:44 Yep. So it's very, very powerful for PMs. You should set up a lot of different ChatGPT projects and be using... or Claude projects and be using them. Uh, the next-

Anshumanni Rudra 33:55 One thing that I, one thing that I genuinely like about Claude is the artifacts bit. So Claude allows you to build artifacts and share it with other people, and I think quick, quickly figuring out something, uh, and then turning it into an artifact and then sharing it with someone, which is like a deployed app, but within Claude, uh, within Claude surface, I think it's a very powerful, uh, uh, feature that they've built. I genuinely like artifacts a lot.

Aakash Gupta 34:20 Yeah. I think Groq is super overrated. I don't know how it does so well in the evals. It feels to me like, I don't know, maybe they're gaming the evals or something like that, and the just different stuff that Elon is always tweeting is just so weird. You know, what is our thoughts on Groq?

Anshumanni Rudra 34:41 I think its context window seems to be really, really small, right? And what it knows about you. So my usage of Groq always has been, I've been on Twitter for a fairly long time, and just trying to figure out what I have said on Twitter or how I have interacted with people itself feels, uh, feels missing. That's the data that they have available, right? And so Groq should actually do a really good job of telling you your own persona on Twitter and the things that you've done, things you've tweeted. It does a poor job of even doing things like, "Tell me the five posts of mine that have done really well over the last 10 years," and it just could go back a few years, and that feels like a lost opportunity given that you have all this, all these tweets to have trained the model on, right, and have it available in its context window.

Aakash Gupta 35:30 Yeah. Right.

Anshumanni Rudra 35:30 So yeah, my, my usage of it is very low, and as a PM, I don't know what I'd use it for.

Aakash Gupta 35:34 Yeah. So I think it gets that D-tier. Sorry, Elon. Then we have the Gemini models. They have underneath them NanoBanana and Veo 3 as well, so I think that does give them a higher rating than otherwise would. I mean, I think NanoBanana is by far the best-

Anshumanni Rudra 35:52 Beautiful

Aakash Gupta 35:52 ... image model, and since they nerfed Sora 2, I think Veo 3 is the best video model. So I'm inclined to say PMs don't actually have to create many ads, images or video, but it's really good, so it's something like a B. Does that sound right?

Anshumanni Rudra 36:06 Yeah, but PMs have kids, and kids love NanoBanana. [laughs] No, but I-

Aakash Gupta 36:10 The kids

Anshumanni Rudra 36:10 ... I think overall, and, and I think Gemini's deep research still is, for me, uh, the strongest of all the, the deep research modes.

Aakash Gupta 36:19 Mm. So it's pretty good. So is B okay, or does it need to go up to A?

Anshumanni Rudra 36:24 I'd actually, uh, we have... I'd, I'd still, my personal preference, despite working at Google, is Claude, so I'd still say Claude is ahead of Gemini a little bit.

Aakash Gupta 36:34 Yep. Yep. It is. Microsoft Copilot, I am not a fan. I feel like it's just a neutered version of ChatGPT, but it's the only version of ChatGPT a lot of big enterprise companies give access to their employees. So it's kind of like GitHub Copilot. If you have access, use it, otherwise avoid.

Anshumanni Rudra 36:54 Yeah. Yeah. Haven't used it much.

Aakash Gupta 36:57 Not a great tool. You're not missing too much. And then Manus.Um, Manus is one of the most... People are worried about Manus because it's Chinese, but it's one of the most agentic LLMs. What do we think about Manus?

Anshumanni Rudra 37:13 In their highest tier, what Manus allows you to do is, uh, essentially run multiple searches simultaneously. So imagine if you wanted a deep dive into every single one of Shakespeare's plays. Right? Now, a single deep research would be, "Hey, go and research Hamlet, and then write me, like, a 2,000-word repo- report on this. I want a character analysis. I want to know which pop culture, music videos, and books, and television shows have been inspired by it." But what if you wanted to simultaneously do it for every single Shakespeare book ever written, every single Shakespeare play ever written? Right? So Manus is actually, uh... It, it's fantastic because it's actually able to run multiple searches simultaneously, so it's-- it'll thread them out and say, "I'm gonna run 24 different searches. I'm gonna use a lot more of your tokens. So, you know, be, be careful about this. This is gonna cost you these many credits." Right? So Manus has a credit-based system, and it'll very clearly say, "Only run this if you have enough credits and you are comfortable running this." So I find Manus to be incredibly useful. I, I, I use it for a lot of deep searches. Uh, it generates these really interesting repo- reports for me. Um, it does some certain things fe- It, it's really good at coding. It's really good at image generation. Uh, if... As in, as a, a free use agentic tool, it's, it's really good because if you wanna sort of, like, say, "Hey, there are these specific websites on which I will go, and there are these specific functions that I'll do every day," you can set up regular tasks. So, you know, every day, morning, 6:00 AM, it'll actually go, uh, it, it, you know, it's, it's like a cron job that it'll run every morning, 6:00 AM. It'll do that specific thing, extract certain data, which is only on a website. Sadly, you can't use Zapier or anything else to pull it. It'll copy-paste that, put it in a doc, and, you know, have it ready for you at, at, at a certain time, or it could, in fact, e-email it to you, right? So for a lot of these general purpose things, I think Manus is actually a really fantastic tool. Uh, very strong team based out of Singapore, uh, doing some great work.

Aakash Gupta 39:13 Okay. Based out of Singapore. So shout out to Singapore, where you are as well. Um, let us decide then, does Manus get an A or a B?

Anshumanni Rudra 39:26 I think if we look at all the As so far, right? Uh, hmm. Manus is a strong B+ for me, right?

Aakash Gupta 39:36 B+.

Anshumanni Rudra 39:36 As much, as much as I love it, uh, I'd still give it a B+.

Aakash Gupta 39:40 [chuckles] We'll move it all the way to the left-

Anshumanni Rudra 39:42 Yes

Aakash Gupta 39:42 ... on the B category. [laughs]

Anshumanni Rudra 39:45 Yes. Because now we-

Aakash Gupta 39:46 Very powerful tool.

Anshumanni Rudra 39:46 Yeah. But we have to now grade based on what we have already graded, so relative grading has started. Ah, the worst thing that happened at college. [laughs]

Aakash Gupta 39:54 Just the one exception, I suppose. So we went through all the LLMs. I think we both agree Claude is the best LLM for product managers.

Anshumanni Rudra 40:02 Yes.

Aakash Gupta 40:03 All right. So now we're gonna go into the data insights category. I think, um, a lot of people are sleeping on this category in my personal experience. I'll ask people, "Oh, do you-- are you using AI experimentation? Are you using AI agents?" Not so many people are, but they're actually really powerful. Amplitude AI agents, they can literally look in the data. They can say, "Oh, this might be a good experiment for you." Then they can actually go create that experiment. Then you can just have your designer and engineer review it, and you can launch it, and it'll even tell you what results to look at for that experiment.

Anshumanni Rudra 40:38 Wow.

Aakash Gupta 40:39 So they've built some highly agentic stuff. Um, I think it's pretty good. I think that they have a lot more finesse to do in terms of the quality bar that they hit on their regular product versus their AI agents. It's not exactly there yet. So I'm inclined to give it, like, a B. It feels like a Gemini-style grading where it's good, but there could be more that we could get out of it. Does that sound right?

Anshumanni Rudra 41:02 Sounds good. Sounds about right.

Aakash Gupta 41:04 All right. So we are gonna give Amplitude our ever-growing B category. The next tool we have is Kameleoon. A lot of people may not have even heard of this. I hadn't heard of it until very recently, and that's because they came up with this pretty cool innovation, which is prompt-based experimentation. Build experiments in minutes by chatting with AI. And a lot of big companies are using it, literally, like Toyota and Fossil and stuff. So on their website, they can just chat, and it'll fully code the variants, and then you just have review again with your design and engineer, and you can launch an experiment, and it'll analyze it for you. Their prompt-based experimentation in terms of quality is quite a bit higher than Amplitude. It's not as agentic, so I'm inclined to give it probably a B as well. Does that sound right?

Anshumanni Rudra 41:55 Yep. Yep.

Aakash Gupta 41:58 So Kameleoon, very powerful tool. I think you should be trying, if you're not, to use a tool like one of those. The other big alternative that people talk about these days is Optimizely. And Optimizely just released-- I think a couple months ago, I said, "Hey, Optimizely doesn't have AI-powered testing." And their SVP commented on my LinkedIn post. He said, "Wait a few days."

Anshumanni Rudra 42:22 Coming soon. [laughs]

Aakash Gupta 42:23 And so they finally released it. And, um, it's very much marketing-focused, I would say, compared to PM-focused. So an Amplitude or a Kameleoon, I think very, very good for tests within your app. I think Optimizely best for tests on your homepage, which typically marketing owns. And I think that their AI agent isn't quite as developed as those other two, so I'm tempted to put it in that lower category of C. Does that sound right?

Anshumanni Rudra 42:53 Yep.

Aakash Gupta 42:55 All right. Pretty cool tool, though. We're not saying don't use it. If that's what you have access to, guys, use it. [laughs] Statsig. This just got purchased by OpenAI, actually, for $1.1 billion. Company out of Seattle.They're really-- They don't have that many AI features built into their product per se, but they're really good at analyzing AI experimentation. In fact, that's why OpenAI uses it to analyze all of their experimentation and other companies like Braze and Notion and Ancestry. They have some interesting views out there, um, for product managers. Like you can go look into a single surface area and see what are all the experiments I ran on the checkout page over the last two years, and how much did that move metrics. They even have a way to look at like across all your pages, how many of those pages, what impact did, did you get on different pages, checkout page versus homepage versus conver- uh, payment funnel, different steps of the payment funnel. So because it's not super AI powered, but it's good for AI, I'm tempted to put it in the C category. Does that sound right?

Anshumanni Rudra 44:04 Yep. Yep.

Aakash Gupta 44:06 So if they build a little bit more AI functionality, they could bump up, but for now... So we went through the data and analysis. We put two of them, Amplitude and Kameleoon in the B. We put Optimizely and Statsig in the C. If I had to choose one, I think Kameleoon is the best, so I would rate that as the best Amplitude, um, experimentation tool. Does that sound right?

Anshumanni Rudra 44:27 Sounds about right. Yeah. And I think it also tells us that there's a, there's a lot more growth possible, right? There are the S&A tools are still to be built or some of these tools will improve over time and, you know, rise higher in the ranking.

Aakash Gupta 44:40 Yeah. There's a lot of opportunity in this space. Vibe experimentation is gonna blow up. I think probably five percent of you are doing it right now. I think fifty percent of you will be doing it in a year or two. So the next tool that we have, this next category, I think a lot of people don't even know that

Aakash Gupta 44:55 this is a category of AI tools, which is discovery and research, right?

Aakash Gupta 45:00 Discovery is the most important thing we do as a PM. So what this tool Unwrap does is it's gonna collect all of your data from everywhere where you talk to customers, Slack, social media, even some of your ana-analytics. Perplexity is a big fan of this tool, actually, one of their named customers. And then it will generate like, "Okay, these are the features you should build based on feedback." And if you build a feature and you wanna ask for feedback, you can go see, "Oh, here are the hundred times customers have talked to us about it." Um, so I think it's a very powerful tool. It's a little bit new in the startup game than its competitor, which we're gonna talk about coming up, Interpret and Dovetail. So I'm inclined to give it a C. Does that sound right?

Anshumanni Rudra 45:45 Yep. Yep. Still some way to go.

Aakash Gupta 45:48 Unwrap gets a C. Okay. The next tool, this tool is much more, um, I would say, rolled out than Unwrap, where Unwrap is very B2C focused. Interpret is very B2B SaaS focused, and they have a lot more big clients now, whether that's Canva or Strava or Perplexity or Atlassian. Uh, looks like Perplexity is using both interestingly enough. Uh, but Interpret, um, even better, I would say at connecting to things like App Store reviews and G2 and Capterra, which anybody in B2B SaaS, you know, that's where people are talking to. You can see, yep, my boss at Apollo.io back then gave one of their testimonials. Um, I would say that it's an awesome tool. Not enough people are using it. I'm tempted to put it in the A category because it literally tells you what features to build and how much ARR they can add.

Anshumanni Rudra 46:42 That's excellent. I'm actually surprised there aren't more tools in this category simply because back in the machine learning days, right, sentiment analysis was probably the first problem that people used to solve. What are people saying about our product? So, you know, going through your Play Store listings and, uh, you know, all your social media channels, et cetera. It used to be the problem to solve. And I, and I don't think it's a solved problem yet. So surprised that there are so few tools which are, which are doing it, given that the surfaces and channels have actually gone up where customers interact with you.

Aakash Gupta 47:15 Yeah. I think you can even hook in your text messages if you text message customers like I used to.

Anshumanni Rudra 47:20 Yeah.

Aakash Gupta 47:20 And then we have Dovetail. So Dovetail used to be a user research platform. That's where most people actually experienced it. But they've just pivoted into competing with Unwrap and Interpret now as they realize that this customer intelligence category is so big. And so I think that-- You see Notion using both products. A lot of these companies are using both products. Um, but I think it's very powerful. It doesn't get as far as Interpret. It's just not as developed in terms of here's the ARR or here's the revenue you can expect, and it doesn't connect into your CRM as deeply. It doesn't connect into your analytics platform as deeply. But it's very powerful, so I'm tempted to give it a C just like Unwrap.

Anshumanni Rudra 48:00 Yep.

Aakash Gupta 48:02 All right. So in terms of the best discovery and research tool for product teams, we give Interpret the really, the high ranking. That's what you should be using. Okay. Now let's go into the browsers.

Aakash Gupta 48:17 And what would you say about Comet?

Anshumanni Rudra 48:21 I think Comet started out with a lot of promise, and now they've rolled it out to everyone. Uh, I think usage would go up, uh, and they'd get more insights and more feedback. But I think the-- there's a lot more to do. You know, when I, when I think about what a AI-first browser should be, uh, Comet probably-- Like the top 10 things that I would come up with, Comet probably does two or three right now. So there's a, there's a, there's a long way to go, right? I think AI-driven browsers and AI-first browsing experience is, is the general purpose agent of all general purpose agents, right? Because it should essentially be able to do pretty much everything for you. Uh, they also don't have as much context about users, which is gonna be a very big deal. A browser essentially is an extension of you, and without having that kind of context, it's gonna be very difficult to, uh, do things effectively for users.See, this is the, this is the classic ask versus do split in usage, right? Perplexity was very good with asking. You asked it questions, it gave you answers. Comet is about doing, right? Browsing is about doing things. So the ask versus do dichotomy, I think it's, it's a very tough one. It's n-not like a lot of people are getting it right, but there's a, there's some way to go. As it stands today, I think it'll be a C.

Aakash Gupta 49:38 A C. All right. Yeah, maybe even leaning into the D category.

Anshumanni Rudra 49:43 Yeah.

Aakash Gupta 49:43 It's, like, something we don't really recommend. Yeah. It's a D. Okay.

Anshumanni Rudra 49:48 Yeah.

Aakash Gupta 49:49 And then we have DIA, which I don't know if a lot of people have even heard of, but it's pretty cool. Um, it's pretty powerful. So DIA allows you to chat with your tabs. Uh, it's created by the people who created the Arc Browser, Browser Company. They were recently acquired for six hundred million dollars by Atlassian. It's really good for writing with context. It's really good for learning. It's good for planning. Even shopping it can help you with. So there's a lot of things I think that I like about it. It's privacy first, just like everything with the Browser Company, which I think is sometimes a worry with Comet. So I'm tempted to give it that C category. It's powerful. There are some good use cases, but it's not gonna transform your life.

Anshumanni Rudra 50:32 Yeah. I think the cross-tab functionality and being able to chat with your tabs itself, I think, is a huge plus right there.

Aakash Gupta 50:38 Yep. Very powerful. Okay. So what is the best

Aakash Gupta 50:44 AI browser for PMs? We say DIA. Now on to roadmapping. So every PM knows they have to build roadmaps. It's part of life. Uh, one of the roadmapping tools that's out there is Jira Product Discovery. So they're gonna help you with AI, you know, create tickets. They're gonna help you with AI connect your user insights to what's the actual tasks you're gonna create. So it's a really good tool if you have a very developer-heavy team, I think-

Anshumanni Rudra 51:15 Yeah

Aakash Gupta 51:15 ... where you're trying to say, "Okay, how do we better integrate where my developers like to work with my product management tools?" It's moderately AI powered, but it's mainly just regular functionality. There's nothing, like, hugely wow about it, but it's useful. So I think it ultimately just falls in that C category.

Anshumanni Rudra 51:36 Yeah. It's, it's that very classic product management though. Uh, one of my favorite anecdotes of, uh, one of my favorite PMs I worked with, uh, uh, one of my mentors never used to write PRDs. He just used to create a hot list of tickets, and he'd come in in the morning, and he'd generate like 12 tickets, 14 tickets, and he'll assign it to different people, and it's like he'd break down the entire thing. It's almost like a very API driven brain that he had, and I think a tool like this would be his favorite thing to use.

Aakash Gupta 52:04 Love it. So the next tool we have is Pendo. Uh, Pendo is now like a two-and-a-half million dollar product management platform used by a lot of the big companies out there, you know, the Octas, HubSpots of the world.

Anshumanni Rudra 52:17 Mm-hmm.

Aakash Gupta 52:17 Their AI functionality, they're building a lot of it. I don't-- I think they haven't totally released it all quite yet.

Anshumanni Rudra 52:23 Mm-hmm.

Aakash Gupta 52:23 But they have agent analytics that I think is interesting, which, uh, actually all of your-- across all your AI agents, it'll show you how much people are using and how much they're spending, 'cause sometimes they go out and overspend.

Anshumanni Rudra 52:34 Right.

Aakash Gupta 52:34 And then it has, like, AI analytics features because its core sort of thing that it did, right, was, like, it built analytics. So you can ask it things like, "Create a dashboard for me," or, "What analytics should I be looking at?" I think-- It doesn't-- It seems like they're more going for a platform play and they haven't put a lot-

Anshumanni Rudra 52:52 Mm-hmm

Aakash Gupta 52:52 ... of polish into this feature, and so I'm tempted to give it that D tier.

Anshumanni Rudra 52:56 Yeah. Although that AI usage feature is gonna be very useful. Uh, I'm hearing a lot from early stage, mid-stage startups that their AI bill at the end of the month completely sort of, you know, cascaded into, uh, i-into a territory where they could not control it because, you know, the more, the more, uh, tokens you get, uh, the higher your bills. Um, so I think just that feature itself, I think would be very useful for younger teams. Uh, but yeah, I think, uh, overall the product needs a lot more.

Aakash Gupta 53:28 Yeah. They need to keep working. So the next tool in this space is Productboard. Again, another one of those billion-dollar product management platforms, really big amongst legacy companies. They've started to release their AI features. They actually have put a lot of finesse and time into this. They've been building these AI features for years. So in terms of how do-- what customer research should I be focused on? How do I connect customer research to the items in my roadmap? I think it's one of the most powerful tools out there. Um, so I'm tempted to give it that B category ranking very similar to Amplitude. Does that sound right?

Anshumanni Rudra 54:04 Makes sense.

Aakash Gupta 54:06 So what is the best tool for AI roadmapping according to us? Productboard.

Anshumanni Rudra 54:11 Productboard.

Aakash Gupta 54:13 All right, so the next category

Aakash Gupta 54:15 is documents. PMs have to write so many documents. So Notion AI, obviously everybody's used it, everybody's tried it.

Anshumanni Rudra 54:24 Yeah.

Aakash Gupta 54:25 But you're not a huge fan.

Anshumanni Rudra 54:27 Oh, no. Let, let me, let me post-qualify that, right? I'm-- I've never been a fan of their app. Their mobile apps to me have seemed like a lost opportunity given that a lot of my work specifically happens on the go. I love their desktop products. I love their, uh, web-based products, but, uh, the mobile app has always been a sore point for me. The AI stuff is fantastic though. I think Notion has done some great work here in becoming a AI first, uh, content tool. I still am sort of old school, so I still would give a double thumbs up to Obsidian.

Aakash Gupta 55:01 Yeah.

Anshumanni Rudra 55:01 It still acts as my second brain. Yeah.

Aakash Gupta 55:03 All right. So we're gonna go much faster with these up tools. We're gonna give it a C rating. Does that sound right?

Anshumanni Rudra 55:09 Yeah.

Aakash Gupta 55:10 All right. So everybody's heard about Jasper. Probably only a few of you have used it. It's a AI tool built for marketers. I think in the re-- era of Claude and PMs not needing to do marketingIt's not really a useful tool, and it's a D-tier. Does that sound right?

Anshumanni Rudra 55:27 Yeah. Not a tool focused on PMs.

Aakash Gupta 55:28 We have another tool s- we have another tool similar to that, which is copy.ai. Very, very popular on your GTM team. They're starting to even do a- agentic translation and localization and other things like that. But I think in the era of a p- all-purpose tool like Claude, it's not the most useful thing you could be spending your time on as a PM, and it probably, again, lands in that D-tier. Is that right?

Anshumanni Rudra 55:54 Yep.

Aakash Gupta 55:56 Not the most useful thing for your time. Okay. This is gonna be a controversial one. I'm curious what you think about it. Grammarly.

Anshumanni Rudra 56:03 Ah. Again, the old warhorse. It's probably integrated into everyone's browser, um, you know. I don't like the idea of, you know, that green G on the side of every text box anymore. It used to have very high utility, uh, but now, given that by default, a lot of our writing, uh, is done in an agentic way, uh, the-- you know, if, if you use Google Docs, then you're already sort of editing within the Docs, uh, and using Gemini for it. If you're using Claude for most of your writing, then, you know, a, a lot of this editing and snippets are already in place. So its utility right now, I think how it deals, uh, with the modern world is gonna be tough.

Aakash Gupta 56:55 I think we gotta give it that D rating, unfortunately. We don't love this category for PMs. [chuckles]

Anshumanni Rudra 56:59 Yeah.

Aakash Gupta 57:01 All right. Then Gamma AI, by far the leading AI presentation tool out there, so we gotta talk about it, 'cause PMs have to create a lot of slides.

Anshumanni Rudra 57:09 Yep.

Aakash Gupta 57:09 Um, my problem is that I can never really use what they put out there. It looks okay, but then when I look at the details, all of the sudden I'm just like, "Ugh." And then editing it is much harder than just editing in Google Slides-

Anshumanni Rudra 57:22 Yeah

Aakash Gupta 57:22 ... PowerPoint or Keynote. So I ultimately feel like it's not useful, and it's a D-tier.

Anshumanni Rudra 57:28 It's not. It's-- Especially when it generates images, it's very difficult to go and then sort of edit the images out. There are times where I feel like it's a lot more work in post-editing, uh, and actually getting to a finished product. I'd much rather build a simpler, less beautiful looking presentation, but for more effective presentation, right? Uh, so yeah, the functionality and reliability of the slides gets in the way of, uh, you know, the delight factor. The delight factor is high, but usability is not.

Aakash Gupta 57:59 So really quickly, Granola,

Aakash Gupta 58:01 everybody's probably heard about it. It's the AI notepad for people in back-to-back meetings. You use it a lot for transcribing your meetings without actually having somebody participate in the meeting.

Anshumanni Rudra 58:11 Yeah.

Aakash Gupta 58:11 So it seems very powerful.

Anshumanni Rudra 58:13 It's very powerful.

Aakash Gupta 58:13 Is Granola A-tier, B-tier?

Anshumanni Rudra 58:17 Right now, at least for me, if you, if you're the type of person who has a lot of meetings, uh, and you constantly are looking for context as to who you were speaking to, Granola does a very good job of not just transcribing, but then transcribing it and turning it into something meaningful, and then you can chat with it. So you can chat with Granola across all the, all the transcribe, uh, transcriptions that it has across different chats, and it's-- then builds a much better context on you. Uh, so very powerful. I think for people who use Granola will swear by it. And it helps you generate-- Suppose you and I now are having our third meeting or fourth meeting. Because it knows that Aakash and Anshum have had three meetings before, it will generate input points that you're supposed to talk about this because you had taken these as action items last time. There are things that it does which are really powerful.

Aakash Gupta 59:05 Yep. Then we have these other meeting tools. We're going to just very quickly go through these. I think Fireflies is the worst of the bunch. It's pretty good, but it's not as good as its competitor, Otter.ai, which ends up landing itself in the C category.

Anshumanni Rudra 59:20 Yep.

Aakash Gupta 59:20 Fathom, same thing. It's not as good as Opera dot-- Otter.ai, so why use it? And then, uh, TLDV, similar sort of thing. Useful tool, but maybe not as good as Otter. Are we okay with those rankings?

Anshumanni Rudra 59:33 Yeah. Yeah.

Aakash Gupta 59:35 So by far the best meetings tool for PMs is Granola. Let's move into dictation next. So WhisperFlow is kind

Aakash Gupta 59:43 of the famous tool. There's also Super Whisper. I think you should be using one of these. [chuckles]

Anshumanni Rudra 59:48 I, I use WhisperFlow-

Aakash Gupta 59:49 To draft

Anshumanni Rudra 59:50 ... I use WhisperFlow a lot, to the point where I think now my-- I actually shudder at the idea of typing at times. I, I learned typing the old-fashioned way on a typewriter as a kid. Uh, so I, I can type. I actually type at sixty-five words a minute, but my typing has actually fallen since I've started using WhisperFlow. And my wife and son know when I'm especially working with a vibe coding tool, I always am dictating to it, and they can actually hear me shouting, "No, that's not the bug I asked you to solve. That's not how you fix it. Fix it this way." They're like: "Wait, you, you're just shouting at your machine?" So it, it, it's sort of gone into that territory. WhisperFlow actually makes, makes you mi-mildly feel like Tony Stark, uh, you know, uh, talking to his, uh, AI assistant and, and, you know, getting them to do the things in the right way.

Aakash Gupta 1:00:38 It's as close to Jarvis as you're gonna get. And then I think slightly better is Super Whisper. So it basically takes what WhisperFlow does, but even with more functionality. So I think we'll put that in the S-tier and WhisperFlow in the A-tier. Does that sound good?

Anshumanni Rudra 1:00:52 Yep. Haven't used Super, uh, Whisper, but yeah, I'll defer to you.

Aakash Gupta 1:00:55 The best dictation tool for you guys is Super Whisper. And then we're gonna go into AI video really fast. So

Aakash Gupta 1:01:03 there's Descript, there's Loom, there's Synthesia, and there's Kling. The way I would think about these is that as a PM, you don't have to actually build that much marketing material. So if you're building a podcast, Descript is great, but otherwise, I don't think it's that useful. As a PM, you do wanna eliminate async communication, but what does Loom AI get for you? Maybe just a little bit better transcription in emails, not everything.

Anshumanni Rudra 1:01:27 Mm.

Aakash Gupta 1:01:27 So I feel that's the C-tier. Synthesia is gonna give you an AI avatar that works, but do you really need to be focused on marketing as a PM? Maybe not. So that's D-tier. And then Kling, really good for video, but-Do you need to be focused on that as a PM? Maybe not, so D-tier. Do those all resonate?

Anshumanni Rudra 1:01:45 Yep. I think not necessarily a strong use case for PMs.

Aakash Gupta 1:01:49 Yep. So Loom is the tool for PMs working

Aakash Gupta 1:01:52 on AI video. Let's move into AI design. So Figma AI, Figma Make, pretty important tool, and it uses your design system. So I think it's quite good for AI prototyping, and I think it's probably something like that B-tier, where it's pretty powerful. And then you have UI Wizard, which is like a UI Wizard competitor to Figma. It's just not as polished as Figma. So I would put that in the C-tier. Do those resonate?

Anshumanni Rudra 1:02:17 That resonates because UI Wizard also is not used by your design team. Figma is, right? So you, you, you get the best and the latest design system to work on top of as a PM.

Aakash Gupta 1:02:27 Yep. And then we have Linear. They have their project management tool where you can call AI agents like Devin or Code Cursor, and it'll just go and code the task that you set up. It's a Jira competitor. I think it's powerful, but it's not hugely AI-based as of yet, so I'm gonna give it a C-tier. Does that sound right?

Anshumanni Rudra 1:02:47 Yep.

Aakash Gupta 1:02:48 And then it looks like Zapier. Uh, I think we talked about this a little bit. It's the workhorse, and so Zapier is probably a B-tier, right below the N8Ns of the world, right?

Anshumanni Rudra 1:03:00 Yeah.

Aakash Gupta 1:03:02 All right. So we have to crown a single

Aakash Gupta 1:03:05 AI tool. What is the single best AI tool for PMs? Is it Claude Code?

Anshumanni Rudra 1:03:09 Ooh. I think it's Claude Code.

Aakash Gupta 1:03:14 Yep. So Claude Code is the absolute best AI tool for PMs. After that, it's probably like a dictation tool like Super Whisper. Then it's probably an all-purpose writing and debugging tool like Replit, and then Lindy. Does that sound right?

Anshumanni Rudra 1:03:27 Yeah. Yep. Thought creation stuff.

Aakash Gupta 1:03:29 These are your top four tools to be an AI-powered product manager. So we just covered a ton of tools. If a PM is trying to figure

Aakash Gupta 1:03:42 out their roadmap, what would be the best way to test out these tools and learn new tools that come out in the future?

Anshumanni Rudra 1:03:49 I think have really important use cases in front of you. I think all PMs, what they need to do first is they need to analyze how they're spending their week, how they're spending their day, and then looking at things which take up a larger chunk of their time, and then figuring out, "Hey, for this specific problem, which is the tool that I need to be working on?" And there's enough and more information about tools, uh, in the world. But pick the tools that specifically will help you in your workflow, uh, and then experiment with them. Then figure out if, if they actually work for you and they're improving your overall productivity.

Aakash Gupta 1:04:22 You heard it from the man himself. He is a

Aakash Gupta 1:04:24 legend. We could have another podcast just talking about all of his growth stories, and we probably will if you guys comment below. As I previewed at the beginning of this episode, if you comment the order of the top four tools that we put and you DM me on LinkedIn, I will give you a free year of a paid subscription to the newsletter. Anshumanni, thank you so much for being here.

Anshumanni Rudra 1:04:46 Aakash, my pleasure. Thank you so much for having me.

Aakash Gupta 1:04:49 Bye, everyone.

Anshumanni Rudra 1:04:52 Bye.

Aakash Gupta 1:04:52 So if you wanna learn more about how to shift to this way of working, check out our full conversation on Apple or Spotify Podcasts. And if you want the actual documents that we showed, the tools and frameworks and public links, be sure to check out my newsletter post with all of the details. Finally, thank you so much for watching. It would really mean a lot if you could make sure you are subscribed on YouTube, following on Apple or Spotify Podcasts, and leave us a review on those platforms. That really helps grow the podcast and support our work so that we can do bigger and better productions. I'll see you in the next one.