01 / 12
01 / 12

AIMM

Dating apps match profiles.
We model human interaction.

A new compatibility engine built on how people actually function together — not who they say they want.
Seed Round — $2M Stage — Pre-product Category — Matchmaking / Relational AI 2026
02 / 12
The Problem

The incentive is broken.
The model is wrong.

Dating apps optimise for retention, not relationships. They surface volume — not fit. The swipe mechanic rewards photos and first impressions, and punishes self-awareness. Users who know what they want are the worst customers.

Worse: stated preferences are poor predictors of actual attraction. People describe an ideal partner and then consistently choose someone different. No existing platform detects or resolves this gap.

40%+ Users significantly frustrated with match quality
3–4× Platforms tried before giving up
≠ Stated preferences vs. real attraction patterns
What every app gives you
  • High match volume
  • Preference-based filtering
  • Photo-first ranking
  • Engagement loops that reward staying
→
What users actually experience
  • Low connection quality
  • Repeated mismatch cycles
  • Exhaustion, churn, return
  • No understanding of why it failed
03 / 12
The Insight

Compatibility is not similarity.
It is fit.

Two people with identical values, interests, and backgrounds can be structurally incompatible. Two people who appear different on paper can function exceptionally well together. Compatibility is about how two systems interact — not how similar they look.

Must match
Alignment
shared foundation
values life direction non-negotiables
Must balance
Complementarity
dynamic tension
emotional expression decision style energy
Must sustain
Tolerance
friction capacity
handling difference repair capacity resilience
Must evolve
Adaptability
long-term viability
growth orientation change tolerance
◈
A structured, reflective person and a spontaneous, expressive one may be a strong pairing — if their emotional tolerance is high and their values align. No preference filter catches this. No swipe detects it.
04 / 12
The Solution

We build a model of
how you actually function.

Not a preferences list. Not a personality test score. AIMM constructs a living model of each person across three interlocked layers — capturing who they are relationally, how they respond instinctively, and what genuinely draws them in.

This model is the input to matching. It is never shown to the other person. It is used only to compute fit.

Relational Layer — how you connect
communication style attachment pattern emotional needs conflict behaviour
Cognitive & Instinct Layer — how you decide
decision-making style problem-solving reaction under pressure instinctive responses
Attraction Layer — what draws you in
calibrated visual preferences presence & presence signals real vs. stated attraction
The model updates continuously. Messaging behaviour, interaction patterns, and feedback signals all refine it over time.
05 / 12
How the model is built

Four signals.
No forms. No self-report bias.

Self-reported data is unreliable. People describe who they want to be, not who they are. AIMM extracts each layer through behaviour, reaction, and inference — not through what the user volunteers.

01 🧠

Adaptive Dialogue

Conversational depth-probing with contradiction detection. What you avoid saying is as informative as what you say.

→ relational model
02 ⚡

Scenario Response

Real-world situations with no right answer. Reveals instinctive behaviour under social and emotional pressure.

→ instinct model
03 ◎

Visual Calibration

Reaction to generated images extracts real attraction patterns. Bypasses the gap between stated and actual preference.

→ attraction model
04 ◈

Presence Reading

User photos are analysed for style, presentation intent, and physical presence — not just appearance.

→ presence model
06 / 12
The Matching Engine

The engine doesn't compare
profiles. It computes fit.

Most matching systems compute similarity — how close two profiles are to each other. AIMM runs Fit(A, B): a weighted function that asks whether these two people will balance, sustain, and grow together over time.

High confidence on a dimension locks it. Low confidence releases it — so the system doesn't narrow the pool based on noise. Each match is returned with a structured explanation, not a percentage.

◈
The moat is the function itself. Every interaction — messaging latency, reply depth, ghosting events — updates the model. The engine becomes more accurate with every user, every week.
Fit(A, B) =
complementaritybalance what differs
+alignmentshare what must match
+tolerancesustain what clashes
+adaptabilityevolve together
psych_layer Values, attachment pattern, communication model highest weight
attract_layer Calibrated attraction — confidence-weighted blind on low-conf
context_layer Life stage, proximity, schedule alignment filter pass
behav_layer Messaging, engagement, feedback — continuous update adaptive
07 / 12
What the user receives

Not a score.
A structured case for the match.

Every other platform
87%
You're compatible — start chatting
No explanation. No context. No why.
AIMM
Why this pairing — the specific fit dimensions at play
Expected interaction dynamics over time
Where this pairing is strong
Where friction is likely — named, not hidden
The system learns from every interaction
💬
Messaging behaviour
Latency, depth, avoidance signals
◎
Engagement patterns
Who gets attention vs. who gets ghosted
↻
Explicit feedback
Post-date signal, continuation intent
⬆
Model update
Profile recalibrated — next match is sharper
08 / 12
Market

A $13B market with no
category winner yet.

The global dating market is $13.1B and growing. Every major platform is adding AI — but none have rearchitected around fit. They are making the same product faster. The fit-based, psychographic-first category has no owner at consumer scale.

TAM
Global online dating market
$13.1B
SAM
Premium AI dating, EN-speaking markets
$2.8B
SOM
AIMM Year 3 target — 0.74% of SAM
$20.78M ARR
Market size · $B · 2022–2030E
09 / 12
Business Model

Pay to be matched correctly.
Not to be seen.

Phase 1 is paid-only — no free tier. Every onboard requires LLM inference and image generation; a money-back guarantee handles acquisition risk. Phase 2 licenses the matching engine to third parties.

Premium
$49
/month · 55% of users
  • Psychographic profiling
  • Multi-modal calibration
  • Unlimited matches
  • Distance / radius filter
  • Stated income preference
  • 7-day money-back guarantee
Elite
$149
/month · 35% of users
  • All Premium features
  • Plaid income verification badge
  • GPS / address confirmation
  • National pool access
  • Priority match queue
  • Background check (basic)
Concierge
$499
/month · 10% of users
  • All Elite features
  • Passport / ID scan verified
  • Certified profile badge
  • HNW-exclusive pool
  • Dedicated matchmaker
  • International matching
Concierge vs. Kelleher International: $5,988/year vs $30,000–$45,000/year. The same calibre of intentional, curated matching — without the geographic constraint or the retainer.
Blended ARPU ~$85/mo LTV:CAC target 15:1 Gross margin ~78% Monthly churn target 6%
10 / 12
Competitive Advantage

Not a feature.
A new category.

Incumbents are adding AI to swipe products. AIMM replaces the swipe product entirely. The moat deepens with every user — proprietary interaction data, calibrated vectors, and behavioral history cannot be replicated by a chatbot bolted onto a feed.

◈
The aligned incentive: AIMM wins when users find a relationship and leave. No other platform can say this. This is the long-term vision — a relational intelligence system, not a dating app.
✦
The data flywheel: each user's profile and calibration history improves the model for everyone. The longer AIMM runs, the harder this is to replicate.
Traditional apps
Pay to be seen.
AIMM
Pay to be matched correctly.
✓ Psychographic LLM interview
✓ Contradiction detection
✓ Fit-based matching — not similarity
✓ Zero swipe mechanic
✓ Adaptive learning loop
11 / 12
Financials

$816K → $18.5M ARR
in three years.

ARR ($M) & Paying Users — 3-year trajectory
Year 1 — 2026
$816K ARR
800 paying users
ARPU~$85/mo
Concierge users80
LTV:CAC15:1
MarketsUS + UK
Year 2 — 2027
$4.54M ARR
4,200 paying users
ARPU~$90/mo
Concierge users420
Gross margin~78%
Markets+ CA, AU
Year 3 — 2028
$18.53M ARR
14,500 paying users + B2B
ARPU~$95/mo
Concierge users1,450
B2B ARR$2M
Markets+ EU, APAC
Gross margin ~78% LTV:CAC 15:1 CAC $45 Breakeven est. M+20
12 / 12
The Ask

$2M seed to prove
the model works.

18 months of runway to build the product, onboard the first paying cohort, and reach the metrics that justify a Series A.

Use of funds
Engineering & product$900K
AI/ML research$350K
Marketing & growth$350K
Operations & legal$250K
Runway reserve$150K
Series A targets (M+18)
ARR$4M+
Paying users3,000+
LTV:CAC≥ 10x
Match acceptance rate≥ 60%
First B2B API contractsigned
$2M
Seed ask
18mo
Runway
Series A
Exit condition
"We don't ask who you like.
We understand how you function — and who fits that system."