Opportunity report AI draft report Trending eval Listed publicly

Reading mode

Readable in one screen · decide continue or stop first

ConceptReges Town – a 3D city run by AI citizens, in the browser

Can a browser-based 3D city with visible AI citizen routines

Author · MicroValidator Last updated Who this is for Indie game developers and streamers who want a living, AI-driven city scene with

Core takeaway / TL;DR

The vision is too broad to validate as-is. The minimum experiment worth running

Suggested next step Scope down to one district and ~30 AI citizens with visible

The vision is too broad to validate as-is. The minimum experiment worth running

Next experiment

Scope down to one district and ~30 AI citizens with visible

Best thing to validate

A no-code, embeddable 'living city' widget could save indie devs and streamers w

Biggest risk

This is a 'cool demo, not a job' — the wow factor of watching AI citizens may no

Critical assumption (summary)

That someone will pay for a browser-based AI-citizen city scene rather than building it themselves or using a free alter

Show objections and continue/stop

Strongest objection

This is a demo with a wow factor, not a job. The novelty of watching AI citizens may wear off quickly, and existing tool

Continue if

If streamers show the strongest signal, pivot to a 'living city backdrop for streams' tool with scene presets and minima

Pivot if

If fewer than 10 waitlist signups and no payment intent appear after sharing on HN and indie-dev communities, stop — the

Show score rationale

Multi-agent LLM simulations (e.g., Generative Agents) are technically feasible but costly at scale.,Browser 3D rendering of ~30-50 agents is feasible on modern hardware.,API cost scales with agent count and decision frequency; a naive implementation may be uneconomical.

Fill the critical gaps (max 2)

“I don’t know” is fine — we give how to get evidence; we never invent interviews or payments.

Q: Which specific segment (streamer, indie game dev, educator) has a recurring, paid job that a browser-based AI-citizen ci

The vision is broad; without one named segment with a recurring job, there is no product, only a demo. Payment intent ca

How to get evidence: Run a landing page with three value-prop variants (streamer backdrop, dev testbed, classroom demo) and measure which variant gets the most signups. Then intervi

Q: Does watching AI citizens drive retention and payment, or is it a one-time novelty?

If the wow fades after one session, a subscription model fails. Retention is the difference between a demo and a product

How to get evidence: In the prototype, track repeat visits and time-on-page per session. If users return 3+ times in a week and ask for persistence or more control, that is a retent

02 Where the signal came from

Raw keyword

Reges Town – a 3D city run by AI citizens, in the browser

Discovery channel

Hacker News

Source note

Hacker News Show HN · 产品向标题线索，非需求证明

04 Business breakdown

Who needs this

Indie game developers and streamers who want a living, AI-driven city scene without building a multi-agent simulation and 3D rendering from scratch.

Job to be done

Embed a browser-based 3D city where AI citizens visibly work, shop, and socialize, so I can use it as a stream backdrop, game prototype, or demo — without writing simulation code.

Why they might pay

Saves weeks of building multi-agent simulation + 3D rendering. The embeddable widget delivers a visually compelling living world in minutes.

Alternatives

Text-based multi-agent sims (e.g., Generative Agents style), game engines with NPC AI scripting, watching others' demo videos, or building from scratch.

Differentiation wedge

One district, ~30 AI citizens with visible daily routines, time-speed control, and an embeddable iframe. No persistence, no city-wide simulation.

OPC scope

One district, ~30 AI citizens, visible daily routines (work, shop, socialize), time-speed control, embeddable iframe. No persistence, no city-wide simulation, no user accounts beyond waitlist.

Open questions

  • Which segment pays first — streamers, indie devs, or educators?
  • What is the retention driver — persistence, control, or visual variety?
  • What is the per-city hosting cost at 30-50 agents with LLM calls?

05 Search & acquisition

Primary channel:

Hacker News (original source), indie game dev communities, AI hobbyist forums, streamer tool directories.

Click a keyword to copy. Tags: AI candidate / queried / thin sample.

Talking points

  • Living AI city in your browser — no simulation code needed.
  • Watch citizens work, shop, and socialize in a 3D city.
  • Embed a living world in your stream or prototype in minutes.

Minimum experiment 06 Minimum validation experiment

One falsifiable step before you continue or stop.

Valid signal = real task completion, reuse, or pay intent — with observe + continue/stop.

Build a minimal prototype: one district, ~30 AI citizens with visible daily routines (work, shop, socialize), time-speed control, running in the browser.

Observe

Does it run smoothly in a browser? What is the LLM call cost per minute at a reasonable tick rate?

Continue

Runs at a usable frame rate and API cost stays under a few dollars per day.

Stop

Performance is unusable or API cost exceeds $10/day for 30 agents.

Create a landing page with a recorded demo, a waitlist, and three value-prop variants (streamer backdrop, dev testbed, classroom demo).

Observe

Which variant gets the most signups? What do signups say in the optional comment box?

Continue

At least 20 waitlist signups in one week, with one variant clearly ahead.

Stop

Fewer than 10 signups and no variant stands out.

Share the prototype and landing page on Hacker News and indie-dev communities.

Observe

Comments, direct messages, and whether anyone asks for a specific use case or offers to pay.

Continue

At least 5 people name a specific use case or ask about pricing.

Stop

Feedback is mostly 'cool' with no named use case or payment intent.

Interview the first 10 waitlist signups (15 minutes each): what would you use this for weekly, and what would you pay?

Observe

Do they name a recurring job and a budget? Do they mention a current workaround?

Continue

5+ interviewees name a recurring use case and a budget range.

Stop

All interviewees say 'nice demo' but cannot name a weekly use or budget.

Offer an early-bird tier ($15-30/month, first city included) to the waitlist.

Observe

Any pre-payment or committed signup?

Continue

At least 1 pre-payment or 3 committed signups.

Stop

No one is willing to commit money or a firm yes.

FAQ

Should I keep building this idea?

The vision is too broad to validate as-is. The minimum experiment worth running

What is the smallest next experiment?

Scope down to one district and ~30 AI citizens with visible

What is the biggest risk right now?

This is a 'cool demo, not a job' — the wow factor of watching AI citizens may no

Who is this validation for?

Indie game developers and streamers who want a living, AI-driven city scene without building a multi-agent simulation and 3D rendering from scratch.

How should I read the score?

Current OPC feasibility score is 4/100. It is an AI-assisted executability check, not a market guarantee. Low scores mean “prove demand first,” not “never build.”

07 Evidence & boundaries

Demand evidence summary

No traceable demand evidence. The vision is broad; the next step is to find one segment with a recurring job and get them to name a use case and budget.

External record

None

User-stated

None

Model hypothesis

  • HN commenters may have expressed interest in AI-citizen simulations, but that only supports 'people are curious', never willingness to pay.
  • Some developers may want a browser-based 3D city as a testbed for multi-agent AI without building rendering themselves.
  • Streamers may want a visually compelling 'living background' that runs without heavy setup.

Known facts

  • The keyword came from Hacker News.
  • The vision is a browser-based 3D city run by AI citizens.
  • No product definition, user, or business model was provided.

Unverified assumptions

  • 'AI citizens' implies LLM-driven agents with visible behavior.
  • 'In the browser' implies WebGL/Three.js-style rendering.
  • Existing multi-agent text simulations exist and are free, so the 3D browser aspect is the differentiator, not the AI itself.

AI inference: Multi-agent LLM simulations (e.g., Generative Agents) are technically feasible but costly at scale.,Browser 3D rendering of ~30-50 agents is feasible on modern hardware.,API cost scales with agent count and decision frequency; a naive implementation may be uneconomical.

Advanced: one-click frontend skeleton

Turn this idea into a single-file HTML MVP (Tailwind CDN + core interaction mock).