Skip to content
Aakash GuptaAakash Gupta

Zapier's CEO Will Grade Your AI PM Skills Live

Zapier CEO Wade Foster wrote the AI fluency rubric that kicked off a wave of companies grading PMs on how they use AI. In this episode, he grades 3 PM artifacts live, from a PRD written with ChatGPT to a full agent stack with a nightly triage swarm, and explains why only about one PM in 30-40 should earn a transformative rating. Then he opens up his own robot staff, including the scribe that cut two hours of end-of-day cleanup to 15 minutes. Full Writeup: https://www.news.aakashg.com/p/zapier-ai-fluency-rubric-for-pms Transcript: https://www.aakashg.com/how-zapiers-ceo-grades-pms-on-ai-fluency/ Timestamps 00:00 - Intro 02:39 - How Zapier measures AI fluency 06:53 - Where most PMs should sit on the curve 09:17 - Ads 11:23 - How one PM built Zapier's company brain 13:22 - Grading PRD one, written with ChatGPT 24:43 - Slop cannon or turbo brain 29:55 - Grading PRD two, prototype plus customer evidence 38:03 - Grading PRD three, the full PM agent stack 46:07 - Why not every PM should build the factory 48:17 - Wade's robot staff, from morning brief to scribe 57:06 - Why Wade vibe coded a CEO CRM 1:01:56 - Why transformative is an n of one 1:03:45 - Zapier since 2021, headless SaaS and new competitors 1:12:55 - Outro 🏆 Thanks to our sponsors Arize - Ship AI agents and features faster with fewer regressions - https://arize.com/?utm_source=aakashgupta&utm_medium=newsletter&utm_campaign=arize_sponsor_ai Get a full year of Arize’s paid plan, along with 8 other AI tools, at bundle.aakashg.com Key Takeaways 1. Basic AI use is the floor, not the bar - Summarizing, drafting, looking things up and generating a basic PRD are table stakes at Zapier in 2026. Everything above that line needs structured, reusable systems for specs, prototypes and research. 2. Adaptive is where the curve should sit - Zapier wants most PMs at adaptive most of the time, with one transformative example to point to. Living at transformative full time means you are tuning your systems instead of shipping. 3. Transformative means an n of one - Wade's bar is that you are the first person doing a new thing and figuring it out as you go. On a team of 30 to 40 PMs, that is roughly one rating a half, and Wade grades himself as adaptive in most places. 4. You should be ahead of your CEO inside your own discipline - Wade should be behind his PMs on PM-specific AI fluency. When he spots a technique that is obvious to him and new to the PM, he gets nervous. 5. Never stop at the PRD - A document alone reads old-fashioned in 2026. Build the prototype with a coding agent, even though production code is not the PM's job, because seeing is believing. 6. Evidence got cheap, so there is no excuse - Sift Gong calls, Zendesk tickets, subreddits, LinkedIn and X before asking anyone for buy-in. The buy-in conversation now happens after you have proof, not before you start. 7. Judgment separates a turbo brain from a slop cannon - AI plus judgment looks like watching a new species get invented. AI without judgment generates so much noise that it is harder to work with than a teammate who uses no AI at all. 8. Label the effort level when you share AI-assisted work - Say whether you skimmed it once or stand behind every sentence. Handing over unreviewed output just moves the judgment onto the reader. 9. The job shifts from doing tasks to auditing the factory - When agents handle spec, prototype, eval and first-draft code, your job is to ask why the machine produced a bad batch. Where to put the human gates is the design question nobody has solved yet. 10. The new no-code is code, and the user is now the agent - Zapier's bet is deterministic execution in the cloud, so workflows run reliably without burning tokens or keeping a laptop open. If your software cannot run headless, you are swimming against the current. 👨‍💻 Where to find Wade Foster LinkedIn: https://www.linkedin.com/in/wadefoster X: https://x.com/wadefoster 👨‍💻 Where to find Aakash Twitter: https://x.com/aakashgupta LinkedIn: https://www.linkedin.com/in/aagupta/ Newsletter: https://news.aakashg.com #AIFluency #ProductManagement 🧠 About Product Growth The world's largest podcast focused solely on product + growth, with over 200K+ listeners. 🔔 Subscribe and turn on notifications.

Aakash GuptahostWade Fosterguest
Sep 14, 20261h 14mWatch on YouTube ↗

CHAPTERS

  1. 0:00 – 2:52

    AI fluency is becoming a baseline expectation for PMs

    Aakash frames the episode around a new reality: PMs are increasingly being evaluated on AI fluency, not as a nice-to-have but as a core competency. Wade Foster joins to explain how Zapier thinks about AI expectations and where most PMs fall short.

    • PMs are already being graded on AI fluency at leading companies
    • Wade Foster authored a widely shared AI fluency rubric
    • AI can create leverage—but can also produce ‘slop’ without judgment
    • Episode goal: show concrete artifacts and grade them live
  2. 2:52 – 4:23

    Zapier’s AI fluency rubric v2: unacceptable → capable → adaptive → transformative

    Wade walks through Zapier’s updated AI fluency rubric (v2), explaining the four performance bands and what behaviors map to each. He emphasizes that the rubric evolves quickly as model capabilities and workplace expectations change.

    • Four levels: unacceptable, capable, adaptive, transformative
    • Capabilities like summarization and basic PRDs are now table stakes
    • Rubric is role-specific; PM section focuses on reusable structured workflows
    • Rubrics must be continually updated as the space moves fast
  3. 4:23 – 6:53

    What ‘adaptive’ and ‘transformative’ PM work looks like in practice

    Wade describes how higher-performing PMs use AI for structured spec/prototype generation, scalable insight discovery, and rapid solution exploration. At the top end, PMs start building systems—agent pipelines and ‘product factories’—not just shipping individual features.

    • Adaptive: structured approaches, reusable workflows, rapid solution generation
    • AI expands PM reach into adjacent skills (data, marketing, sales)
    • Transformative: systems-building and agent pipelines for product delivery
    • Concept of software/product factories: systems that build systems
  4. 6:53 – 8:41

    Where most PMs should sit on the curve (and why ‘transformative’ isn’t daily)

    Aakash asks about desired distribution across the rubric. Wade explains most PMs should operate in the ‘adaptive’ zone most of the time, occasionally reaching for transformative work via experiments, because living in ‘transformative’ mode can trade off with execution.

    • Most PMs should be adaptive most of the time
    • Transformative work is occasional experimentation, not a constant state
    • Individuals can have gaps and spikes across categories
    • Performance is situational; it’s not a permanent label
  5. 8:41 – 13:22

    Transformative half-example: capability jumps + a ‘company brain’ phase shift

    Wade recalls a recent shift where new models dramatically increased reasoning and coding ability, changing what “good” looked like overnight. Zapier saw a product leader build a shared ‘company brain’ from the ‘second brain’ concept, giving the product org a noticeable step-change in speed and quality.

    • Model capability leaps changed daily work (less manual coding)
    • Second brains evolved into a shared company brain concept
    • Centralized context (strategy, design system, coaching files) boosts productivity
    • A system-level upgrade can ‘jet pack’ an entire product org
  6. 13:22 – 23:47

    Grading PRD #1 (ChatGPT-written): good idea, but ‘old-fashioned’ in 2026

    Aakash presents a PRD for a Zapier IDE harness for Claude Code. Wade likes the product intuition but critiques the artifact as insufficient for 2026: a PRD alone is no longer persuasive without a prototype and customer evidence, both of which are now fast to produce with AI.

    • Positive: interesting concept and plausible Zapier advantage
    • Critique: ‘why stop at the PRD?’—build a prototype
    • Critique: add customer evidence from tickets, calls, social, etc.
    • AI enables deeper validation before seeking buy-in
  7. 23:47 – 30:11

    AI slop vs judgment: ‘delegate the work, not the accountability’

    They discuss why AI fluency can’t be separated from product judgment. Wade endorses the ‘slop cannon vs turbo brain’ framing: AI amplifies output, but without taste and responsibility it can flood teams with low-quality work and shift evaluation burden upward.

    • Great output requires combining AI leverage with human taste/judgment
    • Bad judgment + AI = high-volume low-signal ‘slop’
    • Workplace norm: be transparent about effort level and review depth
    • Core principle: you can delegate work, not accountability
  8. 30:11 – 38:14

    Grading PRD #2: prototype + evidence moves the work to ‘adaptive’

    Aakash brings a stronger artifact: a prototype UI plus supporting evidence from calls, tickets, communities, and interviews. Wade rates it solidly ‘adaptive,’ praising the improved proof and validation, while pushing for faster iteration cycles and earlier design-partner momentum.

    • Prototype makes the idea concrete and easier to evaluate
    • Customer evidence strengthens the case and clarifies bottlenecks
    • Rated ‘adaptive’: strong caliber for a modern PM org
    • Next-level push: compress timelines—why wait until Q4?
  9. 38:14 – 46:08

    Grading PRD #3: the full PM agent stack and ‘product factory’ loop

    Aakash presents an end-to-end agent architecture: skills library, orchestration sub-agents, tool/MCP layer, deterministic enforcement, and a memory layer. He describes a nightly ‘triage swarm’ that clusters signals, manages hypotheses, and triggers an auto-build pipeline from PRD → prototype → evals → code, with human gates. Wade calls this ‘transformative’ territory because it redesigns the operating system of product work.

    • Agent stack layers: skills, orchestration, tools/MCP, deterministic enforcement, memory
    • Nightly triage swarm: dedupe, cluster, tag, contradiction detection
    • Auto-build pipeline: PRD → review panel → prototype → evals → code draft
    • Factory mindset: humans audit systems, not execute every task
  10. 46:08 – 48:32

    Why not every PM should build the factory (org-level scaling of ‘transformative’)

    Wade explains that in larger orgs, you don’t want every PM reinventing the same agent factory. Instead, a small group builds the system while the rest adopt and pressure-test it; transformative ratings should be rare and reflect truly novel, first-of-kind contributions.

    • Small companies may require individuals to build full systems
    • At scale, centralize the product factory rather than duplicating it per PM
    • Most PMs should run the system; a few redesign it
    • Transformative is ‘N of one’—a very high bar
  11. 48:32 – 53:17

    Wade’s ‘robot staff’: personal agents that run on deterministic Zapier workflows

    Wade demos his personal agent dashboard, starting with a morning brief and explaining why Zapier is useful for running agents deterministically. He contrasts deterministic steps (code/API calls, low cost, reliable) with selective token usage for summarization and drafting.

    • Morning brief agent: calendar/inbox/to-dos summarized daily
    • Deterministic execution reduces cost and increases reliability
    • Agents can follow links and attachments for added context
    • AI can generate workflows; Wade builds in Cursor then deploys to Zapier
  12. 53:17 – 56:50

    From daily recap to meeting agendas: high-leverage executive workflows

    Wade highlights his most valuable agent, the ‘scribe’ daily recap, which extracts action items from email, Slack, and meeting transcripts and drafts follow-ups. He also describes an agenda-building agent for exec/board meetings that surfaces the most important issues and reduces agenda bias toward ‘pet topics.’

    • Scribe agent cuts end-of-day admin from ~2 hours to ~15 minutes
    • Drafts follow-up emails and captures action items from transcripts
    • Agenda agent pulls from meetings, Slack, email, metrics, and AI sessions
    • Improves judgment by forcing important topics onto the agenda
  13. 56:50 – 1:03:43

    Why Wade vibe-coded a CEO CRM (and why writing in a human voice is hardest)

    Wade shows a lightweight CRM he built for CEO-specific workflows, despite generally advising against vibe-coding CRMs. The system identifies enterprise accounts needing attention, drafts outreach emails in his voice, and helps him maintain relationships at scale; he notes that consistent human-style writing is the hardest part and requires iteration and analytics.

    • CRM flags accounts via signals like renewals and usage changes
    • Generates 10–20 drafted customer emails weekly with rationale
    • Main value: scalable CEO-to-customer relationship-building
    • Hardest challenge: getting agents to write naturally and in-voice
  14. 1:03:43 – 1:09:50

    Zapier’s business and strategy shift: ‘the new no-code is code’ + headless SaaS

    They pivot to Zapier’s business trajectory since the 2021 valuation peak; Wade notes they’ve grown and valuations matter less without frequent fundraising. Strategically, he argues that AI changes SaaS: the agent becomes the primary user, so SaaS must become ‘headless’ via APIs/MCP to let agents build/run workflows, while Zapier differentiates through deterministic execution and recoverability when automations fail.

    • Zapier has grown since 2021; valuation is less central without primary raises
    • ‘New no-code is code’: agents write/run code even if humans never see it
    • SaaS should expose capabilities via APIs/MCP; agent is the user (B2B)
    • Zapier focus: deterministic automation, lower token burn, reliability, recovery
  15. 1:09:50 – 1:14:10

    Competition in automation: strategic focus in a bigger market

    Aakash asks about newer competitors (Lindy, N8N, etc.) and how Zapier counter-positions. Wade explains that rising demand expands the market but increases competition, forcing sharper strategic choices about where to win; Zapier leans into integration depth, deterministic runs, and agent-assisted setup/maintenance.

    • Automation demand is higher than ever, attracting more competitors
    • Integration library remains a durable advantage (9,000+ tools)
    • Winning requires crisp strategy and customer-specific positioning
    • Differentiation: deterministic execution, easier setup, and resilience when workflows break

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.