Show HN: Pmspace.ai – I built a tool to let you chat with your construction docs

I'm the founder of pmspace.ai. For over a decade, I worked in the architecture and construction world, and I was constantly frustrated by one thing: information chaos.

Every project generates a mountain of documents—hundreds of drawing sheets, 800-page specification books, endless reports, emails, and change orders. The critical data needed to make a decision was always somewhere, but finding it felt impossible. Using Ctrl+F across dozens of PDFs was a nightmare, and the ambiguity often led to expensive mistakes on the job site.

One night, after wasting an entire afternoon looking for a specific fire-resistance rating, I thought, "Why can't I just ask my project a question and get a direct answer with a source?"

That question led me to build pmspace.ai. I started with the feature I desperately needed myself: an intelligent document processing engine. You can upload all of your project's unstructured documents, and the system uses embeddings and LLMs to create a unified, queryable knowledge base. You can perform semantic searches ("find all walls with a 2-hour fire rating") or have a conversation with a specific document ("summarize the requirements for structural steel in section 05 12 00").

This document intelligence is the core of a larger platform we're building to be a true single source of truth for project teams. It connects directly to our workflow automation, BIM viewer, and other modules, all designed specifically for the AEC industry. Unlike generic AI wrappers, it's built to understand the complex, cross-referenced nature of construction documents.

We're a small, bootstrapped team. Our business model is a standard SaaS subscription, with pricing based on the number of users and data storage. There's a 14-day free trial with no credit card required so you can test it out properly.

This is still early, and there is so much more to build. I would be incredibly grateful for your feedback. I'd love for you to try uploading a complex document and see how it performs. What works? What's broken? What's the most frustrating information management problem you face in your own work?

Thanks for taking a look!