Creating practical AI

I build AI apps for the construction industry. Specs, codes, schedules, transcripts, etc. Apply AI to your big data. I bring 40 years of problem solving experience to the task. Autodesk Revit and AutoCAD were my previous specialties.

I help contractors, engineers, and project managers get clarity from complexity. See https://aifab.xyz/hays/ for a list of apps made for a client.

I’ve created tools that parse 6,000+ page construction specifications, analyze multi-thousand-line project schedules, integrate with Autodesk Construction Cloud, and made chatbots using kind any of documents as their knowledge bases. Other examples are a legal chatbot and law assistant using 69000 pages of state laws as its knowledge base, and author emulation apps using all of the authors’ published books.

Simple tech stack for most apps: Python for logic, Streamlit for the web UI.

Secret sauce: For large chatbot knowledge bases, I use a hybrid context retrieval approach. Semantic search combined with LLM-generated keyword search, then re-ranked to get the top context for the user prompts to the LLMs. This enables getting unusual industry jargon, abbreviations, and other specific text from the knowledge base.

Contact: design@truevis.com. See: X, GitHub, LinkedIn.