Run a private AI chatbot that answers from your own documents

Answers questions using your own knowledge base, running entirely on your own servers with no data sent outside.

How the work actually flows

It branches. Exactly one path is taken.

Pattern: Exclusive Choice (4) · Simple Merge (5)

flowchart TD trig>"question submitted in chat"]:::trig s0["receive question"]:::task s1["classify message type"]:::svc s2["search your documents"]:::svc s3["gather and dedupe passages"]:::task s4["write sourced answer"]:::svc trig --> s0 s0 --> s1 s2 --> s3 s3 --> s4 gx{"× small talk or real question"}:::gate s1 --> gx p00["reply casually"]:::task gx -->|"small talk"| p00 p10["search documents"]:::task gx -->|"real question"| p10 p11["write sourced answer"]:::task p10 --> p11 jn{"○ send reply"}:::gate p00 --> jn p11 --> jn jn --> s2 out[/"sourced conversational answer"/]:::out pay{{"private accurate document answers"}}:::pay s4 --> out out --> pay classDef task fill:#e7f6fe,stroke:#34b8f0,color:#2c2a29 classDef svc fill:#f6f8fa,stroke:#7c8795,color:#2c2a29 classDef mi fill:#e7f6fe,stroke:#0079a8,color:#2c2a29,stroke-width:2px classDef human fill:#fff,stroke:#0079a8,color:#0079a8 classDef store fill:#f6f8fa,stroke:#0079a8,color:#2c2a29 classDef trig fill:#00a4eb,stroke:#0079a8,color:#fff,font-weight:bold classDef trigtime fill:#00a4eb,stroke:#0079a8,color:#fff,font-weight:bold classDef trigdata fill:#8ad4f5,stroke:#0079a8,color:#06314c,font-weight:bold classDef gate fill:#fff,stroke:#e8a23d,color:#6b4708,font-weight:bold classDef out fill:#1f9d6b,stroke:#167a53,color:#fff,font-weight:bold classDef pay fill:#06314c,stroke:#021f33,color:#fff
Starts itA stepAn outside serviceOne path onlyPaths rejoinResultPayoff
Build size
Advanced

A larger build with multiple systems, AI reasoning, and custom rules.

Business functions
AI Agents & Autonomous SystemsAI Chatbots & AssistantsKnowledge Base & RAGSpreadsheet & Database Ops
Connects
OllamaPostgreSQL

The problem it solves

You want an AI assistant that can answer questions from your internal documents, but sending sensitive company data to a public AI service isn't an option. Off-the-shelf chatbots also don't know your specific business or files.

Who it fits

Businesses with sensitive or internal documents who need AI answers without using an external cloud AI service.

How it works

  1. A question comes in through the chat
  2. A small AI model decides if it's small talk or a real question
  3. Real questions are matched against your documents using local AI search
  4. The most relevant passages are gathered and deduplicated
  5. A second AI writes a sourced answer with a follow-up question
What you get

Answers pulled straight from your own documents

Your team gets accurate answers sourced from your own internal documents, with everything running on your own servers so nothing leaves your systems.

What you get

A sourced, conversational answer to your question, pulled directly from your own documents.

What you need

Your own server running Ollama and a PostgreSQL database, no external API required.

We can build this. But should you?

The hard question is not how to build it. It is whether this is the right thing to build first.

That is what a Fractional Chief AI Officer figures out with you, before anyone writes a line of code.

Let's Talk Strategy

Related automations

Back to the AI Playbook