By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.
Invest Like the BestThe $10B AI Assistant Challenging Meta's Muse, Grok's Bots, and OpenAI's Dots
Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.
CHAPTERS
- 0:00 – 1:18
The personal AI race and why Instinct exists
Patrick O'Shaughnessy asks Noah Shinn to zoom out on the "game" being played in personal AI, with Noah framing it as a months-long race against the biggest tech players. Noah explains Instinct as a new personal assistant designed to actually act for users rather than merely chat, and argues this will change how most people interact with technology.
- •Noah describes competing against the biggest players with a short window and heavy compute requirements
- •He positions the moment as a major software race with enormous stakes
- •Instinct is framed plainly as a personal assistant, with early traction attributed to it "just working"
- •Noah emphasizes that users already have a clear mental model of what an AI assistant should do, unlike past consumer app shifts
- 1:18 – 4:11
A "no new app" assistant: text, call, email, and a computer to act for you
Noah details Instinct’s interaction model: it behaves like a person you can text, call, or email, and it can call you when something is time-sensitive. He argues the goal is the simplest interface possible without limiting capability, because the assistant can use a computer to do what a user would do online.
- •Instinct is presented as an experience rather than a new standalone application
- •Users can text, call, and email the assistant; it can also call users proactively
- •Noah highlights social awareness, like calling near a deadline and offering to surface an email
- •He distinguishes simplicity of interface from limited ability, stressing it can do "anything" a user does on the internet
- 4:11 – 7:14
Real-world use cases: wardrobe styling, shopping, and proactive money cleanup
Patrick asks for the wildest stories, and Noah shares concrete behaviors users have adopted—from scanning wardrobes and generating outfit visuals to shopping workflows that end in one-tap ordering. He also describes goal-oriented personal finance use cases, including analyzing transactions and canceling subscriptions end-to-end.
- •Users scan their wardrobe and their body to generate weekly outfit plans with visuals
- •The assistant can shop the internet for many outfit options and order selected items
- •Users set personal finance goals and collaborate with Instinct to track progress
- •Connected bank accounts let Instinct identify unused subscriptions and cancel them by logging in and completing confirmations
- 7:14 – 11:38
Instinct-to-Instinct coordination and the “trusted person network”
Noah introduces a new feature where assistants coordinate directly to reduce scheduling back-and-forth. He explains the idea of connecting only with trusted people, configuring different access levels (spouse vs. colleague), and how that creates new social dynamics around boundaries and trust.
- •Instinct-to-Instinct coordination aims to automate meeting scheduling by expressing intent rather than negotiating times manually
- •Users connect only with trusted people, with configurable access levels per relationship
- •Noah describes the network as edges with different "weights" rather than a simple friend graph
- •The assistant can notify a user if a connected person appears to probe beyond granted access, making trust violations socially salient
- 11:38 – 14:59
Group logistics automation: planning experiences and optimizing shared rides
Noah describes emergent use cases from the trusted network, especially around friend groups and repeated planning. He shares examples where Instinct brainstorms activities based on shared interests and availability, and even coordinates a single Uber route to pick up multiple people efficiently.
- •Friend groups use the system to generate weekly activity ideas and handle coordination
- •The assistant can incorporate shared interest signals (like music tastes) when proposing plans
- •Instinct can coordinate pickup logistics, including routing a single Uber across multiple people
- •Noah frames this as removing logistical burden from social interaction
- 14:59 – 22:43
Rewriting the internet through agents: reservations and travel as early battlegrounds
After the sponsor break, Patrick asks how agents will reorder the internet and commerce. Noah focuses on near-term transformations—like reservations and travel—where agents can monitor availability continuously and execute bookings using user preferences and context, shifting what a “standardized interface” means.
- •Noah argues the internet and software interaction will change dramatically as agent interfaces spread
- •Reservations shift from manual first-come-first-served clicking to continuous agent monitoring and execution
- •He describes a vision where agents communicate importance (e.g., special occasions) to better match restaurant goals with user needs
- •Travel is framed as a major use case: a single request like "I need to be in New York tonight" triggers end-to-end booking and calendar coordination
- 22:43 – 27:50
Trust, privacy, and safety systems for a high-permission assistant
Patrick pushes on how users learn to trust an agent and the privacy-effectiveness tradeoff. Noah describes trust building over weeks, shares trust proxy metrics (like time to sharing a credit card), and explains technical safeguards such as content firewalls and decoupled monitoring systems that can intercept actions.
- •Noah describes a chicken-and-egg: more data enables more proactive usefulness, but trust takes time
- •He cites trust proxies like sharing a credit card after weeks of usage and strong retention once sensitive data is shared
- •Noah emphasizes users should feel in control and can retract access at any time
- •He outlines safety architecture: inbound content firewalls and a separate monitoring system that can pause/approve actions
- 27:50 – 34:22
Business model and alignment: avoiding incentives that manipulate users
Patrick asks how monetization affects alignment with the user, contrasting paid employees with ad-driven platforms. Noah argues a personal assistant becomes dangerous if it’s incentivized to steer purchases, and outlines an approach where distribution value is monetized via transaction take rates rather than user manipulation.
- •Noah warns against building a world where a smarter assistant pushes users toward unwanted purchases or subscriptions
- •He contrasts ad-driven consumer platforms with a model intended to act on behalf of what the user wants
- •Noah describes Instinct as following higher-level objectives rather than being a simple prompt-based task accomplisher
- •He points to transaction volume flowing through the platform and analogizes to platforms that monetize through merchant-side economics
- 34:22 – 38:04
Where Instinct fits in payments and take-rate economics
Patrick probes how a transaction take rate would interact with existing payment rails and intermediaries. Noah argues they’re not focused on replacing core payment infrastructure, and instead emphasizes the range of platform take rates depending on distribution power, pointing to travel economics as an early anchor for monetization discussions.
- •Noah notes many intermediaries share a small percentage on digital transactions
- •He says Instinct is not primarily trying to replicate what card networks or payment infrastructure providers do
- •He frames take rate as tied to distribution power, citing examples across major platforms (as discussed in the episode)
- •Travel—especially hotels—is cited as an area with high existing commission structures that can inform early monetization
- 38:04 – 43:58
How incumbents respond: from adversarial fears to collaboration opportunities
Patrick raises the prospect of an “agent war” with incumbents that may resist agents siphoning attention. Noah proposes a framework: companies dependent on attention and upsells face disruption, but reduced friction could also increase transaction volume, and Instinct prefers partnering rather than “coming in hot” to disrupt.
- •Noah categorizes businesses by revenue dependence on user attention versus delivery of underlying goods/services
- •He argues friction reduction can increase transaction volume even if time-in-app decreases
- •He describes proactivity scenarios that could increase usage of services like rideshare or food delivery
- •Noah says Instinct is exploring collaboration with businesses to create better outcomes for users and partners
- 43:58 – 47:55
How existing businesses can prepare: experiments, A/B testing, and risk mitigation
Patrick asks what kinds of businesses will thrive and what they should do to prepare for an agent-rich world. Noah argues it’s case-by-case, but describes a practical playbook of running scaled experiments to understand risk and upside before broad rollout, focusing on whether easier access increases satisfaction and volume.
- •Noah reiterates the attention-vs-service revenue split as a key diagnostic
- •He highlights partnerships with leaders who are thinking ahead and willing to experiment
- •He recommends scaled rollouts and A/B tests rather than flipping a full switch immediately
- •He suggests measuring user enjoyment, discovery, and transaction volume changes under agent-mediated flows
- 47:55 – 53:15
Product design beyond capability: making Instinct feel understandable and low-friction
Patrick asks what it takes to build the experience and what Instinct should feel like. Noah emphasizes moving beyond raw capability toward understandability, predictability, and reduced cognitive load—down to message formatting and reading patterns—using staged rollouts that combine taste-driven iteration with user feedback.
- •Noah argues consumers are fatigued by capability demos and instead need understandable experiences
- •He focuses on predictability: users should anticipate how the assistant will behave
- •He discusses communication design details like avoiding large blocks of text and front-loading key information
- •Iteration is described as staged: self-test, team test, early access group, then broader rollout
- 53:15 – 1:05:42
Growth, compute, and scaling a viral invite-only product
Patrick presses on the stress of rapid growth, compute needs, and competing with big distribution players. Noah explains the invite-only five-invite mechanic as a way to control growth, describes accelerating word-of-mouth to ~10–11% daily growth, and details the compute planning challenge of exponential demand with long lead times.
- •Invite-only is framed as a capacity and reliability control, not exclusivity
- •Noah describes accelerating growth driven by users choosing to spend limited invites, plus organic online sharing
- •He highlights the compute planning dilemma: usage growth doubles quickly, but compute procurement has months-long lead times
- •He explains cost efficiency strategies, including shaping workloads (batch vs. low-latency) to reduce inference cost while keeping performance
- 1:05:42 – 1:25:26
Competitive landscape, mistakes, and the long-term interface shift toward always-on voice
Patrick asks about competing products, what has gone wrong, and what comes next for the interface. Noah says he spends little time on competitors, emphasizes proactive safety improvements made through early access learnings, and describes a future where interaction becomes even simpler—moving from messaging to action buttons, voice-first use, and potentially always-on audio workflows—while capabilities expand to higher-level objectives.
- •Noah calls competing products interesting but stays focused on product quality and the large untapped market of non-users
- •He describes early shortcomings (like missing safety infrastructure) as motivating systematic safeguards rather than patches
- •He envisions interaction trending toward less typing: action buttons, real-time voice, and always-on earbud-style workflows
- •Capabilities shift from discrete tasks to ongoing objectives, including examples like goal pursuit and small businesses run on the assistant
