SpaceHub

Company: SpaceHub

Self-reported basis: The description is entirely self-reported by the authors, unverified, and submitted as part of a Devpost hackathon entry. No external corroboration exists.

SpaceHub is described as a community facility-management platform enhanced with generative AI and transactional safety mechanisms. It allows users to book facilities via chat, with AI interpreting natural language into structured intent, while backend systems (PostgreSQL, state machines) validate and execute operations deterministically. The system includes two core components: GenState UI for safe UI rendering and TransactGuard for zero-hallucination transaction previews.

What Changed: The project appears to be a proof-of-concept or prototype built in a hackathon context. It introduces novel safety patterns around AI interaction with backend systems, particularly through shadow transactions and deterministic UI rendering.

Single Most Important Open Question: Is there any evidence of real-world usage, customer feedback, or product-market fit beyond the authors' own description?

The description states that SpaceHub is a community facility-management platform, enhanced by two technical components:

1. GenState UI: A system using OpenAI Structured Outputs and Zod to parse user intent into typed requests, which are then mapped through a deterministic finite-state machine (FSM) to a fixed allowlist of Vue components.
2. TransactGuard: A mechanism that runs real database mutations inside isolated PostgreSQL shadow transactions, captures results, and rolls back changes before showing previews for actions like bookings or cancellations.

The system is described as not generating UI or outcomes directly from the LLM, but instead using AI only to interpret language into validated intent, with backend systems handling all operational decisions.

Inference: The product appears to be a chat-based booking interface for shared-use facilities, designed to avoid hallucinations in critical workflows like payments and scheduling.

The description states that SpaceHub was built around the constraint:

“AI can make that experience conversational—but only if it never invents availability, prices, refunds, or successful transactions.”

This suggests a positioning shift from typical generative UI platforms — where LLMs are often the decision-makers — to one where LLMs act as translation layers, and backend systems control outcomes.

The authors claim:

• They reversed the usual Generative UI trust model.
• AI parses language, while the database dictates outcome and UI state.
• The system avoids “probably correct” in favor of deterministic behavior.

Inference: SpaceHub positions itself as a safe, deterministic AI-assisted workflow tool, especially for domains where hallucinations are unacceptable (e.g., scheduling, commerce, healthcare).

The description states that SpaceHub is built for community facility management, such as:

• Court bookings
• Gym access
• Studio rentals
• Boardroom reservations

It is implied to be a B2C or B2B SaaS platform targeting organizations managing shared-use spaces.

No explicit customer segments, personas, or buyer types are mentioned. The authors do not describe specific user roles (e.g., facility managers, end-users, admins) or how they would interact with the system beyond chat-based booking.

Inference: The ICP likely includes small to mid-sized organizations managing shared-use facilities, but no evidence of target customer definition exists.

There is no evidence in the description of:

• Revenue model
• Pricing structure
• Monetization strategy
• Customer acquisition or retention tactics

The authors do not describe how SpaceHub would be sold, licensed, or offered to users beyond its hackathon prototype.

Inference: The business model remains unspecified, and no commercial traction or monetization signals are evident.

Key technical elements described:

• GenState UI uses OpenAI Structured Outputs + Zod for intent parsing.
• TransactGuard runs mutations in PostgreSQL shadow transactions at serializable isolation.
• Uses Vue components, PostgreSQL, Node.js, Express.js, Supabase Realtime, and Tailwind CSS.
• Implements state machines, Redis locks, QStash scheduling, and PayMongo/mock checkout for safe commits.

The system is described as:

• Non-interactive historical UI
• Deterministic backend FSM mapping
• Shadow transaction previews
• Real-time UX via FullCalendar + Supabase Broadcast

Inference: The architecture shows a strong focus on safety, determinism, and transactional integrity, especially in AI-assisted workflows.

There is no evidence of:

• Revenue
• Customers or user adoption
• Product-market fit
• Real-world usage
• Product iterations or versioning beyond the hackathon prototype

The authors describe this as a hackathon project, and no data on traction, growth, or retention is provided.

Inference: The product is at an early stage (prototype), with no demonstrated traction or maturity.

There is no mention of:

• Direct competitors
• Market size or TAM
• Competitive advantages
• Prior art or similar tools in the marketplace

The description does not reference any existing platforms for facility booking, AI-assisted scheduling, or transactional safety in generative UIs.

Inference: No competitive positioning is evident. The project may be in a novel niche, but no market context is provided.

1. No commercial traction or revenue evidence: The product is described as a hackathon prototype with no real-world usage.
2. Unproven scalability: Shadow transactions and deterministic UIs are novel, but no evidence of performance or scalability under load.
3. Limited customer insight: No user personas, feedback, or buyer journey described.
4. High technical complexity without validation: The system is built on several complex components (state machines, shadow DB transactions), but no evidence of successful deployment or testing.
5. No monetization path: No indication of how the product would be sold or priced.

Inference: The project is highly experimental, and its commercial viability remains unproven.

1. What was the actual user feedback during the hackathon? Did anyone use it beyond prototyping?
2. How does the system handle edge cases or ambiguous inputs that don’t match known patterns?
3. Are there any real-world constraints or assumptions in your architecture that might not scale?
4. What is the long-term vision for GenState UI and TransactGuard as open-source tools?
5. Have you considered how this would integrate with existing facility management systems or enterprise platforms?