Anzhu: Housing Decisions, Mapped

What the company appears to be: Anzhu is a self-reported Codex Skill that enables users to make family housing decisions through an interactive map. The product uses GPT-5.6 for conversational reasoning and structured planning, with deterministic local rendering and state management.

What changed: The author shifted from an enterprise AI agent model (that would require sharing sensitive data) to a local-first Skill architecture that keeps planning state on the user's machine while using external AI for flexible reasoning.

Single most important open question: Does Anzhu have any commercial traction, revenue, or customer adoption beyond the single developer's prototype?

The description is self-reported and unverified. No evidence of revenue, customers, or usage exists in this analysis. The author states that the project was submitted to a hackathon, but no further commercial evidence is provided.

The description states that Anzhu is:

• A Codex Skill
• An interactive housing-planning workbench
• A local-first application with deterministic state management
• Built using Codex and GPT-5.6 for reasoning, with a Next.js frontend
• Designed to help users make family housing decisions through conversation and visual planning

The author describes it as a "local-first Skill" that separates flexible AI reasoning from deterministic execution, keeping sensitive data local while allowing the model to reason about complex family constraints.

The description states:

• The product was initially imagined as a "fully hosted real-estate agent"
• This evolved into a "local-first Skill" that lets users choose their host model
• The positioning shifted from requiring users to hand over sensitive data to keeping planning state local
• The author claims this approach makes the model's creativity usable over time through guardrails

The claim evolution shows a shift from an enterprise AI agent model to a privacy-preserving, deterministic local-first architecture.

Not evidenced. The description does not specify:

• Who the target customers are beyond "family housing decisions"
• What demographic or geographic segments they serve
• Whether there's a specific customer persona or ideal customer profile

Not evidenced. The description does not contain:

• Any information about pricing models
• Revenue streams
• Monetization strategies
• Customer acquisition costs
• Unit economics

The description states:

• Built with: codex, gpt-5.6, maplibre-gl-js, next.js, node.js, react, typescript, zustand
• Architecture separates flexible intelligence (Codex/GPT-5.6) from deterministic execution (local workbench)
• Uses a strict Skill contract as boundary between AI and state management
• Includes automated tests (85 passing regression tests)
• Has session isolation with unique visual session IDs per conversation
• Employs guardrails to make output auditable
• Supports voice walkthroughs synchronized with map markers

Not evidenced. The description does not contain:

• Any evidence of revenue or customers
• Usage metrics or adoption data
• Product-market fit indicators
• Growth trends or user engagement
• Market traction beyond the hackathon submission

Not evidenced. The description does not contain:

• Information about competitors
• Market positioning relative to existing solutions
• Competitive advantages or differentiators
• Industry landscape or market size

Inferences based on self-reported information:

• Single developer team: Only one member listed (zhulijin1991 zhulijin)
• Hackathon prototype: Submitted to OpenAI 2026 hackathon, suggesting it's a prototype
• No commercial evidence: No revenue, customers or traction data provided
• Unproven market demand: The author states they don't know if there's demand for this specific approach
• Technical complexity risk: The architecture requires balancing AI flexibility with deterministic state management
• Privacy vs. utility tradeoff: The local-first approach may limit the product's utility compared to cloud-based solutions

1. What is your actual customer acquisition strategy beyond the hackathon?
2. How do you plan to monetize this product if it remains a developer tool?
3. What specific family housing decisions are users currently making with this tool?
4. How do you plan to scale from one developer to a full team?
5. What evidence do you have that there's market demand for this approach?
6. How will you handle the technical complexity of maintaining deterministic state while using flexible AI models?
7. What is your go-to-market strategy for reaching families with housing decisions?