SEERFISH /

The future of your startup.
Simulation, not speculation.

Seerfish is a swarm intelligence engine for startups.
Upload your deck. Hundreds of agents simulate the future.
See what three years from now looks like, today.
→ Run my deckFirst run free.→ Scout startups here$49.99/mo · 5 unlock credits.

Premise why simulation

“I don’t understand the phone at all, but I do understand consumer behavior.”
· Warren Buffett

As the story goes, Buffett invested in Apple after his director lost an iPhone in a taxi and said it felt like losing a piece of his soul.

This kind of emergent insight is what Seerfish simulates. Consumers, investors, team members and competitors are all modeled in the simulation. They debate, build on each other’s points, and get flashes of insight just like real life. Something that one LLM call or Excel modeling cannot reproduce.

Leaderboard live runs

live · updated hourly
SEERFISH LEADERBOARD
top 10 by Seerfish 3-yr ARR · this week
#locstagearr (3y)
scanning…
Top 10 by Seerfish 3-year ARR estimate. Big claims with the math to back them.
All the startups this week, not just the top 10. Free. Weekly. Read by investors.
Runs by Sunday, lands in Tuesday’s briefing.

Inside a run what 113 agents produced

specimen · synthesis · run 0481

Bootstrapped pre-seed. Marketplace and workforce play. Deck claimed $9.5M ARR by year three. Below: what the swarm said about a representative sample deck.

seerfish 3-yr arr
$3.93M
18mo $1.13M·5yr $8.29M
weighted-median triangulation
gap
2.4×
deck vs reality
light reaches half as far as expected
deck claim · 36mo
₹79Cr
($9.5M USD)
founder projection · pre-dive
confidence 89%·P(deck claim) 22%
agents 113debate turns 455rounds 40
dive deeper · click to expand
How the swarm got to $3.93Mfive pillars · weighted median
OPERATOR
$4.8M25%n=14 · ceiling $12M
DOMAIN EXPERT
$1.4M25%n=12 · 9.5% reach $1M
CUSTOMER
$1.3M20%n=27 · $120/mo WTP
INVESTOR
$4.8M15%n=10 · 100% fundable
FOUNDER
$9.5M15%n=2 · founder conviction
Operators and domain experts anchor the floor. Founder and investor priors lift the ceiling. The headline is the weighted median, not the mean.
Where 113 agents landeddistribution · 5 outliers at $9M+
4
17
14
2
12
23
5
$0.6M$3.5M$6.6M$9.5M
customer-heavy · 0–1M
domain experts · 1–3M
investors / operators · 4–9M
optimistic tail · 9M+
Only 5 of 113 agents estimate at $9M+. The mass clusters between $2.9M and $6.4M, where operators and investors anchor.
Top diagnosis it’s reliability, not demandcited by 31 agents

“The main risk is not demand. It is whether the company can keep service reliably working on bad days, when providers drop out, replacements are needed, or plans change.”

evidence that would shift this
  • Week-by-week coverage fill rates, including time-to-replace and success rate by coverage area.
  • Attendance, backup coverage, and escalation-resolution SLAs across the last 8 to 12 weeks.
  • Refund and complaint recovery logs that show the service stays dependable after a bad day.
  • Cohort retention for customers that used a backup versus customers that did not.
Four levers that move the scorewhat the swarm modeled
#1◉ OPERATOR
›

Solve "supply consistency". Attendance, backup coverage, and provider quality across expanding coverage areas. Hiring, training throughput, and unit economics must stay disciplined as the service scales.

score impact+$1.19M
#2◉ INVESTOR
›

10/10 investors say fundable at this stage. 4/10 would write a check. Top objection: "quality." Address mandate fit or find investors aligned with stage and sector.

score impact+$0.71M
#3◉ DOMAIN EXPERT
›

Domain experts anchor the score at $1.4M based on a 10% base rate for reaching $1M. Comparable companies had early traction by year 3. Add customer data, repeat purchase rates, or LOIs to prove the surviving cohort.

score impact+$0.35M
#4◉ CUSTOMER
›

$120/mo willingness to pay with 60% buy intent. Increase conversion by addressing alternatives customers currently use and demonstrating clear switching value.

score impact+$0.25M
The pessimist chorussix experts · lower tail · three visible
◉ DOMAIN EXPERT
$599K

“I believe the company can grow, but the odds of reaching the deck target are low because the category is operationally difficult and the base rates are weak.”

◉ DOMAIN EXPERT
$1.0M

“I believe the company can grow, but the category is operationally hard and reaching near-$10M ARR in 36 months looks unlikely without strong retention and execution.”

◉ DOMAIN EXPERT
$1.3M

“I believe the category is operationally hard and trust-heavy, so scaling cleanly across cities is difficult despite some early demand.”

Three more visible to Scout subscribers. Talk to any agent in the simulation.

How the swarm thinks

five things

Five things that make the simulation real.

open roots
Built on OASIS and MiroFish. Two open-source frameworks for multi-agent simulation. We orchestrate. We don’t fork.
role-archetype personas
Before the agents run, we research the public landscape. What operators in this category say. What investors at this stage are funding. What customers in this segment are buying. Personas are constructed from these public role archetypes. Not impersonations of named individuals.
personas, not prompts
Each agent has a persona derived from your deck and your market. Not a generic role-play. The customer who churns has a reason. The investor who passes has a thesis.
threaded debate
Agents post. Other agents reply. The platform ranks what each agent sees. By round forty, the population converges through real interaction. Not parallel estimation.
persistent memory
Each agent remembers across rounds. Round 23 builds on round 14. Round 40 reflects the whole simulation. Ask any agent later and they recall what they said and why.

Pricing three doors. one free.

cancel anytime
founder · free
One run.
Ever.
  • Simulation
  • Score, levers, diagnosis, distribution
  • Leaderboard entry
  • Public, anonymous, or hidden
founder · $20/mo
Two runs a month.
Featured in the briefing.
  • Everything in Free
  • Two runs a month (no rollover)
  • +2 runs for $20
  • Receive VC messages
  • Featured in the weekly briefing
scout · $49/mo
Five founders, in full.
Message and intro.
  • Free briefing
  • Triage the leaderboard
  • country, stage, ARR, Index
  • 5 full reads/month (no rollover)
  • +5 reads for $49
  • Message founders you’ve read
  • Request intros
The briefing is free for any VC with an email. Country and ARR only.
Cancel anytime. No credit card to start.

Questions worth answering

six

Six questions worth answering. The rest live in the docs.

[ HOW IS THE SEERFISH INDEX CALCULATED? ]
Index = simulation² ÷ deck claim. Match the swarm and earn full credit. Double-claim and lose half. Higher is better. The index rewards founders who claim big and have the math to back it, not founders who claim safe.
[ WHAT DOES A SCOUT "READ" ACTUALLY DO? ]
A read unlocks one founder’s full simulation. Score breakdown, top diagnosis, lever rankings, distribution, agent transcripts. Once read, the founder stays unlocked for that Scout permanently. Reading also enables messaging that founder and requesting an intro. Five per month. No rollover. Reload at $49.
[ WHAT IF MY COMPETITOR SEES MY ENTRY? ]
Founders pick visibility. Public on the leaderboard with country and stage only. Anonymous (still on the leaderboard, no identifying fields). Or hidden (briefing only, intros via direct message). The founder controls what appears. Every week, every run.
[ IS THIS REAL OR JUST AN LLM IN A TRENCHCOAT? ]
The agents post on a Reddit-style platform across forty rounds, reply to each other, and update their positions. Each persona is seeded from real-world stakeholder research, not generic role-play. The score is computed from what the agents do, not from a single model call. Run a deck twice and the convergence holds. That is what makes it different from one LLM in a trenchcoat.
[ WHAT IF THE SIMULATION PRODUCES A NUMBER I DISAGREE WITH? ]
Then the gap is the conversation. Look at the levers. Each one shows what the deck claimed, what the agents simulated, and what the score moves to if you change it. Update the deck. Re-run. Watch the number move. Most founders re-run nine times before bed. Re-runs are unlimited under the founder subscription.
[ WHY A HUNDRED AGENTS AND NOT A THOUSAND? ]
A hundred is enough for the disagreement to be honest. More agents add cost without adding signal. The bottleneck on prediction quality is not agent count. It is persona depth and ground-truth research. Both are where we spend the cycles.

Next up Seerfish product

coming next

Most product teams use AI to write PRDs faster.
That’s not transformation.

Seerfish Product reimagines how product work happens. Not just writing PRDs. Hundreds of agents simulating the future. Pricing, features, repositioning. See the future today.

No spam. One email when it ships.

Run yours back at the surface

Run yours.

Tuesday morning at 9, the next briefing fires.

Be on it.

→ Upload my deck
First run free.
Warren Buffett, Berkshire Hathaway, and Apple are not affiliated with Seerfish. References are illustrative.
Seerfish · Simulation, not speculation