$6B
Roughly the 2025 US online-dating market.
External benchmark · US online dating, 2025 estimate · Business of Apps, Apr 2026; Match Group, 2025 annual report · Market estimate; not product traction evidence.
The companion that knows you — and knows when to introduce you.
Clayon learns who you are through real conversation, and remembers it privately — no profiles to perform, no public display.
When something is actually worth connecting over, Clayon quietly finds the person who shares it, and introduces you both, only when you both say yes.
01 — The problem we start from
We spend a staggering amount of our lives on connections that go nowhere.
01 · The friction
Hours scrolling and swiping on strangers. Networking events that lead to nothing. Cold messages that never get a real answer. First calls where, five minutes in, you both realize this was never going to fit.
02 · The repetition
We repeat the work of explaining ourselves from scratch across dating, jobs, roommates, tutors, freelancers, dinners, and rides. Every app starts cold, connects badly, and makes us rebuild ourselves from zero.
03 · The cost
We do all of it alone.
04 · The shortcut
The well-connected have always had a shortcut: a wingman. Someone who listens, remembers what you are looking for and what you cannot say out loud, keeps an eye out, and makes the introduction count.
The companion
Clayon is that wingman, for everyone, for everything — one companion that actually knows you, that you never rebuild.
02 — The connection economy
The connection economy is booming. The connections aren’t.
We have never spent more to meet people — and never felt less well-connected.
$17.8B
LinkedIn revenue in fiscal 2025.
External benchmark · LinkedIn fiscal 2025 revenue · LinkedIn / Axis Intelligence, 2026 · Company-platform context; not product traction evidence.
1.3 billion
Registered LinkedIn members.
External benchmark · Registered LinkedIn members, 2026 reference · LinkedIn / Axis Intelligence, 2026 · Registrations are not active relationships or product traction evidence.
46%
US adults who have used dating apps and report a negative experience.
External benchmark · US adults who have used dating apps · Pew Research, Feb 2023 · Self-reported experience; not a safety conclusion or product evidence.
48%
US adults who have used dating apps and report facing harassment.
External benchmark · US adults who have used dating apps · Pew Research, Feb 2023 · Self-reported experience; not a safety conclusion or product evidence.
1 in 3
US job postings estimated to be a ghost job that is never filled.
External benchmark · US job postings, Nov 2025 estimate · HR Dive and CNBC, Nov 2025; Business Economics, 2025 · Estimate; definitions vary; not product traction evidence.
So in the end
The product is the browse, not the connection.
The pattern
Why are existing solutions designed this way?
Across every vertical, the same pattern appears: enormous demand to connect, enormous spend to help people connect, and broad dissatisfaction with the result. The apps optimize for engagement and browse volume — not for the right connection — because the right connection ends the search, and ending the search ends the revenue.
Marketplace model
Monetize the browse, the inventory, and the lead.
Public profiles, ranking, and time-on-app keep the search going. The product is the browse, not the connection.
Wingman model
End the search with one good, reciprocal connection.
A wingman understands your real reasons, boundaries, and timing privately — exactly what existing apps make you broadcast.
How do you let a pair or group discover reciprocal fit without exposing themselves before they choose to?
03 — So how does it work?
Two systems, one grammar.
The relationship system
A companion worth returning to even if it never introduces you to anyone.
It earns a correctable, private understanding through conversation. It remembers with provenance and uncertainty. You can see, correct, restrict, or forget what it knows.
It listens, remembers what mattered, notices patterns, helps you think, and follows through — without pretending to feel, chasing dependency, replacing people, or claiming to be therapy.
The connection system
Reciprocal fit without a public profile.
When you want something, the connection system turns the intention into a purpose-scoped projection, evaluates reciprocal fit privately, and walks both people through staged, separate decisions.
It explains a proposal without exposing private evidence. It abstains when evidence or consent is absent — and abstention is a success, not a failure.
04 — What we are building
An AI platform you talk to every day — like a friend who is always there, pays attention, and happens to know a lot of people.
It is not a marketplace you browse. It is a relationship you keep. The speed is yours to set; the privacy is not.
- 01
It gets to know you.
Ordinary conversation builds a private, correctable understanding. No public profile, no cold start, no rebuilding yourself from zero.
- 02
It watches for opportunities.
It quietly keeps an eye out for a collaborator, neighbor, tutor, teammate, customer, or friend whose situation genuinely fits yours now.
- 03
It helps you connect.
When a fit is real and mutual, both people approve exactly what is shared before a conversation opens.
Six lives. Six different kinds of fit.
A companion starts with the life you are already living.
Choose a person. See what they tell Clayon, what Clayon helps them understand, and how a carefully bounded connection becomes possible.
Aisha · Friendship through recurring repair
A place where someone comes back.
What’s happening
Retirement took away Aisha’s classroom—the place where people returned at the same time, week after week. Coffee meetups, gallery groups, and a friend-finding app gave her pleasant evenings, but no place in someone’s ordinary week. Those formats rewarded spontaneity. Aisha connects while people make or repair something together.
How Clayon helps
As Aisha talks to Clayon, it notices the same pattern across teaching, ceramics, repair, and neighbor projects. It reflects that pattern back and lets her correct it before any search begins. Once she approves it, Clayon turns it into a small, matchable intention: every two weeks, daytime, a quiet public place, mutual repair, and no pressure for instant emotional disclosure. Her loneliness stays private.
What happens next
Clayon finds Minh because he independently asked for the same rhythm and refused one-way repair requests. It shows each of them the shared fit, asks for interest separately, and reveals only the details they both approve. They meet at a public library. The deck stays broken. The hour works.
What stays private
Her loneliness and insecurity remain private.
Reference journey from the Clayon specification—not product traction.
05 — How we compete
The relationship layer sits before the marketplace.
It focuses on the moment marketplaces skip: understanding the person and staging safe, reciprocal disclosure. Trusted partner integrations can work across platforms with pairwise identifiers, no people search, no profile export, and consent for every connection.
Not a general-purpose AI assistant
Understanding that can open a human connection.
General assistants end with an answer or a completed task. This system combines longitudinal, purpose-bound, owner-controlled understanding with a pathway to human connection.
Not a marketplace
No browse, inventory, or people on display.
There are no public profiles, one-sided search, ranking, or engagement-driven browse. Where marketplaces already work well, the relationship layer routes to them.
Not a simulated companion
The relationship supports life beyond the product.
The companion supports human connection and your own thinking. It never makes attachment or dependency the destination and never substitutes itself for people.
Not a coaching or wellness service
Reciprocal connection, not one-to-many service.
Coaching services provide professional support. They do not create reciprocal, consent-gated connections between participants.
06 — Business model
Pricing hypotheses for three customer surfaces.
Research price bands · USD · July 2026
These are test prices for willingness-to-pay and cost-to-serve research. They are not validated prices, current offers, revenue, or forecasts.
Companion response service
$0 or $15–$25/mo
Core: $0. Extended: $15–$25 per month or $150–$250 per year.
The first research anchor is $19 monthly or $190 annually. Overtime pricing is not established. Every tier keeps the same implemented capabilities, model policy, memory, safety, consent, and trust rights; payment changes only the disclosed ordinary-response window and response target.
No per-message charge. Exact service windows and response targets must be measured before any price becomes a live offer.
07 — How Clayon comes online
A 24-month proof sequence.
Timeframes are planning, not commitments. The month-15 checkpoint is decided on evidence, not runway pressure.
- 01
Phase 0 · approximately months 1–6
Earn the relationship.
Ship one continuous private companion with inspectable, correctable memory. Establish a calm public presence and validate with a founder-approved alpha of 8–12 consenting adults. No matching yet.
- 02
Phase 1 · approximately months 6–12
Prove one matching lane end to end.
Build intentions, matching, proposal, reveal, and connection for one lane; synthetic proof first, then one real consented connection. Begin with the cofounder/collaborator lane and hold the line on no browse and no scores.
- 03
Phase 2 · approximately months 12–15
Operable trust and the first paying customer.
Add opt-in proactive contact, notifications, deletion and revocation under load, operations, analytics, a first paid individual tier, an organization pilot, and a partner lane.
At month 15: continue, narrow toward sustainability, or stop — on evidence.
- 04
Phase 3 · approximately months 15–24
Repeat and expand.
Messaging, organizations, mobile, and a second lane follow only after the first is proven. Growth and team size remain tied to evidence gates.
08 — Who’s building Clayon
A small senior core of system thinkers, with governed AI agents as execution leverage.
Seven capabilities, staffed by roughly five people plus AI leverage. The proof round is capped at five full-time people.
- 01
Product / founder
Relationship thesis, trust contract, roadmap, evidence gates, and go-to-market.
- 02
Full-stack / platform engineering
API, persistence, generated client, Temporal, Mastra, and consent grammar.
- 03
ML engineering
Matching, retrieval, embeddings, evaluation, and governed AI orchestration.
- 04
Mobile engineering
Authenticated native companion, accessibility, offline, and device boundaries.
- 05
DevOps / infrastructure
Dokploy, PostgreSQL, Temporal, R2, CI/CD, backups, observability, and least privilege.
- 06
AI security and trust
Threat modeling, prompt-injection defense, agent identity, red-team, and incident response.
- 07
Quality and evaluation
Synthetic evaluation, human scenario review, accessibility, and evidence-state discipline.
09 — We need your help
Help us prove the first real, reciprocally useful, safe connection that would not have happened otherwise.
We are looking for aligned partners and backers — investors, design partners, advisors, and people who would use Clayon — to support a focused 24-month proof.
I’m interested — how can I help?
Our team
The people building Clayon.
A small core of people who own whole outcomes, with governed AI agents as execution leverage.
Curly Nguyen
Arina Kudaiar
John Bourdon
Nguyen Dong Ho
