Part one
Problems people feel today
Social media makes people anxious and performative
People are scared to post, toxicity hurts mood, and the feed is engineered to hook. The bet: build for the silent majority and for wellbeing, not engagement.
Money makes people anxious, so they avoid it
43% of Gen Z avoid their banking app; fintech's top complaint is fear, not features. The bet: lead with opportunity and action, not the number.
Staying informed without being consumed
The feed keeps you scrolling, venues have dead screens, and there is no calm way to keep up or react. The bet: news you do something with, or glance at, not doomscroll.
Connecting with real people is slow, and it exposes you
Starting a group takes forever and costs your number; your presence is static; families split across platforms. The bet: instant, private, no-contact-sharing connection.
Part two
The agent economy
Barely begun, and the biggest opportunity. Today's agents reach consumers as chatbots, not as generative interfaces, and they are barely proactive. That gap is the opportunity.
It has not started: identity, control and payments for agents are all unbuilt at consumer scale, and being early is the bet.
Chatbots, not generative interfaces: agents are still a text box you prompt. Agents that build their own interface and act without being asked is exactly what Agentic Desk does.
Agent-to-agent payments badly need a consumer version: the plumbing (x402, stablecoins) exists, but there is no human-friendly front door.
The stack that could be one platform bet
Identity, owner control and the payments layer may be a single platform play, with agents as content and a storefront on top.
Part three
Paying and programmable value
Conditional money, group money, per-call machine money, money on a habit: none of it is native to consumers or the web, and stablecoins change the economics. The bet: programmable, instant, stablecoin-settled value made simple.
Part four
Cross-cutting patterns and opportunities
The patterns that show up across many ideas, and where the opportunity sits.
Anxiety is the real driver, not features
In both social and money, the barrier is emotional avoidance, not a missing feature. Whoever designs for the avoider wins demand nobody else is serving.
Behaviour change with skin in the game
Dashboards and habit apps only track and shame. Putting something on the line actually moves behaviour, especially when someone else is in your corner.
Proof of real, in a synthetic web
AI-generated identities and content are flooding in. Value accrues to whatever can prove realness: a real human, a real meeting, an agent that is labelled and not impersonating.
Escaping the phone number
The phone number is a broken primitive for identity and contact. Decoupling it, with personas and no number, is a structural moat incumbents cannot copy.
No install, instant, web-first
The app download is the friction that kills adoption. Browser-first, QR-and-link entry, working in one tap for people who have nothing installed.
Ambient, lean-back, and physical screens
Not everything should be a scroll. Glanceable ambient surfaces, typography as interface, and the dead screens in venues as an out-of-home channel with location-relevant promos.
Hyperlocal and the real world
Place, footfall, venues, events and in-person presence are the moat AI cannot fake, and the bridge between digital and physical is underbuilt.
New income for individuals
A through-line: give people a way to earn from what they already have (assets, knowledge, agents, resources, spare capacity), instantly.
Trust and auditability of AI
As AI does more, people need to trust the output: sources, confidence, cost, a second model checking the first, and honest "not advice".
Value in the data that is not in APIs
The valuable signals (reviews, availability, footfall, satellite) are not in any API; only scraping and agent research reach them. Democratising alt-data is a real edge.
Ephemerality by design
Things that end or fade: it starves trolls, cuts liability, and removes the pressure of a permanent record. A deliberate architecture, not a feature.
Brands as the customer, inside conversations
A recurring revenue pattern: brands and venues pay to live inside chats, rooms and screens, as participants not interruptions.
Regulation as tailwind and constraint
Rules cut both ways: a dated mandate can be a tailwind, and a behavioural rule can be a design constraint that keeps you honest.
Part five
The strategic axes
How to choose between them.
Timing. The agent economy has not started. Some bets are early and huge (the agent stack); others solve problems people feel right now (money and social anxiety, instant rooms).
Meta vs weirdly private. Go big and neutral (Skews, Feed.me, When Then Pay), or go deliberately private and niche (Gazette, Rooms, The Chill Network, Absent Minded).
Ride the rails vs build the rail. For agents and payments the giants build the rails; the edge is the layer on top (Endpoint's routing, The Fixer's control). Do not build the rail.
Works today vs bet on arrival. When Then Pay, Chill Money, Skews and Rooms solve problems now; the agent-economy ideas bet the shift arrives soon.
Consumer-viral vs B2B infrastructure. Viral consumer (Rooms, Skews, Mint, Gazette) versus B2B and infrastructure (KYA, Endpoint, Agentic Desk). Agentic Desk is the exception: B2B, but with a data moat and a finance edge.