Summary of Noah Shinn - Building Instinct: The Personal Agent - [Invest Like the Best, EP.493]

Overview of Invest Like the Best EP. 493: Noah Shinn on Building Instinct

In this episode, Noah Shinn, founder of Instinct, explains his vision for a new kind of personal AI agent: a “superhuman person” with a phone, computer, email address, and calendar that can act on your behalf across everyday tasks. The conversation focuses on how Instinct works, why Noah believes the future of software is an agent-first interface, how trust and safety are being built in, and how this could reshape everything from scheduling and travel to commerce and consumer behavior.

What Instinct Is

Instinct is positioned less as an app and more as a new experience for interacting with AI.

Core idea

• A personal assistant that can text, call, email, browse, book, cancel, and coordinate
• Designed to feel like a socially aware operator, not a prompt box
• Built to handle both simple tasks and complex, multi-step workflows end to end

Interaction model

• Users can interact via:
  • text
  • voice
  • phone calls
  • email
• The product intentionally avoids adding unnecessary UI layers
• Noah argues the future is “no app” or minimal interface, with AI acting through existing channels

Key Use Cases and Examples

Noah shared several examples of how users are already using Instinct in surprisingly powerful ways.

Personal organization and life admin

• Finding time on calendars between trusted contacts
• Coordinating schedules with spouses, friends, and colleagues
• Managing inboxes and follow-ups automatically

Shopping and wardrobe planning

• Users scan their wardrobe and body measurements
• Instinct suggests outfits for the week
• It can also browse and buy clothing online based on preferences

Financial and subscription management

• Scans bank transactions and subscriptions
• Identifies unused services
• Cancels subscriptions end to end
• Surfaces savings back to the user

Travel and reservations

• Books trips, hotels, flights, and ride-share coordination
• Can optimize around user preferences like airline, seat type, hotel style, and calendar timing
• Noah sees travel as one of the biggest immediate categories for agents

Social planning and logistics

• Coordinating group outings
• Finding mutually available times
• Selecting concerts, restaurants, or activities based on shared interests
• Even routing an Uber to multiple friends efficiently

The “Trusted Person Network”

A major theme of the episode was Instinct’s network of trusted connections.

How it works

• Users only connect with people they trust
• Different relationships can have different permissions
• Example: a spouse may get broad access, while a colleague may only see work calendar or limited inbox context

Why it matters

• Makes agent-to-agent coordination possible
• Helps create a new kind of social infrastructure
• Trust violations can be surfaced and may affect the relationship within the network

Noah’s view

He sees this as more than a feature—it’s a new type of social graph with weighted permissions, not just binary connections.

Trust, Privacy, and Safety

Noah spent significant time discussing how users come to trust agents and what protections Instinct uses.

Trust builds over time

• It often takes weeks for users to fully trust the product
• Noah treats that as healthy and expected
• A key proxy for trust is when users begin sharing:
  • a credit card
  • email access
  • other sensitive accounts

Retention signal

• Noah said that once users share at least one sensitive piece of information, retention rises sharply
• This is evidence that trust and usefulness reinforce one another

Safety architecture

Instinct uses multiple layers of protection:

• Firewalls to filter incoming malicious content
• Independent monitoring systems that approve or block actions before execution
• Proactive scrutiny of outputs and tool calls to prevent hallucinated or unsafe actions

Core principle

• The user should always remain in control of their data
• Data access should be shareable and retractable at any time

Business Model and Alignment

A big topic was how Instinct makes money without misaligning incentives.

Noah’s concern

He does not want Instinct to influence users toward purchases or behaviors that are not in their interest.

Why this matters

• He contrasted Instinct with ad-driven consumer platforms
• His view: if the agent is smarter and more socially aware than the user, a monetization model that pushes behavior could be dangerous

Likely model

• A transaction take rate on value delivered through the platform
• Similar to:
  • Apple Pay
  • Shopify
  • Amazon
  • travel/OTA marketplaces
• The idea is to exchange distribution and convenience for a small cut of completed transactions

Why this could work

• A large share of platform activity is in travel
• Travel and booking categories already support meaningful take rates
• Noah believes the company can expand this model across many categories as the interface becomes more central to digital life

How Instinct Could Reshape the Internet

Noah’s larger thesis is that agents will reorder how people interact with software, commerce, and services.

Short- and medium-term changes

• Reservations become dynamic and always-on
• Travel becomes proactive and preference-aware
• Ride-sharing and food delivery get ordered before the user explicitly remembers to do it
• More purchases happen with less friction, not less often

Long-term changes

• The internet itself may be “rewritten”
• Many apps may collapse into one adaptive interface
• Users will shift from:
  • searching
  • clicking
  • filling forms
  • repeating preferences
    to simply stating goals and intents

Business impact

Noah argues:

• Businesses that benefit from more transaction volume will thrive
• Businesses that rely on user attention, friction, or inertia may be disrupted
• A major theme is less time in the app, more value from the service

Scaling, Compute, and Operational Challenges

A large part of the interview focused on the hard operational realities of building a fast-growing agent.

Growth

• Instinct is invite-only
• Early growth was driven almost entirely by word of mouth
• Noah described growth compounding around 10% day over day
• Users reportedly started trading invites and even reselling them

Compute challenge

• Growth is not just a user problem; it’s a compute forecasting problem
• Compute demand grows faster than traditional consumer apps because the agent is proactive and works in the background
• The company must buy capacity months ahead, risking being:
  • under-provisioned
  • or overpaying for unused capacity

Efficiency strategy

• Noah said the team is heavily focused on inference optimization
• By customizing workloads and using batch work where latency is less important, they can reduce costs dramatically
• The goal is to keep the product affordable, with Noah expressing a long-term ambition to keep it free

Product Philosophy: Understandability Over Capability

One of the strongest recurring ideas was that Instinct should feel understandable, not just powerful.

What matters most

• Users should be able to predict what the product will do
• The experience should feel calm, clear, and intuitive
• The team tests not just outputs, but:
  • tone
  • message shape
  • cognitive load
  • emotional feel

Why this matters

Noah believes many AI products over-focus on capability and under-focus on:

• usability
• trust
• emotional comfort
• readable, low-friction interactions

Competitive Landscape and Market Shape

Noah sees the market as open, but moving quickly.

His view of competition

• He respects other products like Muse
• But he spends relatively little time worrying about direct competitors
• He believes adoption is still early and uneven globally

Market structure

• He thinks the category may become winner-take-most, or at least very concentrated
• But the transition will likely involve collaboration with existing platforms rather than pure replacement

Relationship with incumbents

• Instinct aims to work with companies like Uber Eats, travel providers, and booking platforms
• The ideal outcome is better conversion and more transaction volume for everyone involved

Key Takeaways

• Instinct is building an AI agent that acts like a trusted, socially aware assistant.
• The future interface may be much simpler than today’s apps—possibly just voice, text, and proactive action.
• Trust is built gradually, and safety systems are central to the product.
• The business model is likely to come from transaction take rates, not ads or behavior manipulation.
• The most disruptive impact may be in categories where apps currently monetize user friction and attention.
• Scaling the company depends as much on compute planning and workflow optimization as on user growth.

Notable Insight

“We all have really been waiting for AI to act for us.”

That line captures the episode’s core thesis: users already understand the promise of AI; the challenge now is delivering it in a way that feels safe, useful, and natural enough to become part of everyday life.