Self-hostable AI workspace

One workspace for every question your data can answer.

RunaxAI brings project documents, SQL data, live web research, browser tasks, memory, and streaming tool execution into one cited assistant experience.

file types ingested

tool families

observability planes

150

backend tests passing

Open RunaxAI Sign in

What does our Q3 report say about churn, and how does it compare to industry benchmarks?

document_search

Searching "Q3 churn analysis" across 847 documents…

web_search

Fetching SaaS churn benchmarks 2024…

RunaxAI

Q3 Report.pdf · p.14

profitwell.com

Ask a follow-up question…

Platform

The homepage now needs to describe the full application, not just a chat box.

RunaxAI is a product surface for knowledge work: projects, documents, tools, memories, sessions, auth, deployment, monitoring, and evaluation are all part of the current system.

Project workspaces

Create a project, upload PDFs, CSVs, Word docs, Markdown, or text files, then chat against that project with source-backed answers.

General chat

Use document search, SQL queries, web search, and browser tasks from one conversation.

Memory

Durable facts from conversations and manual memory entries stay available across sessions.

Tool planner

Calls are streamed inline, deduplicated, cached when safe, and coordinated across sequential and parallel tools.

Data access

Ask database questions in plain English while the app keeps auth, ownership, and rate limits in the request path.

Workflows

Built for repeat work, not demo prompts.

The app has real paths for ingestion, search, streaming response state, memory management, protected sessions, and source inspection.

PDF, CSV, DOCX, MD, TXT

Document intelligence

Upload files, let the worker ingest and embed them, then ask the project assistant for cited answers.

Tools stream as they run

Operational research

Combine live web search, browser tasks, database queries, and project evidence in one response.

Manual and extracted facts

Persistent context

Memory captures stable facts about projects, preferences, and background so follow-up work starts with context.

Metrics, traces, logs, evals

Production feedback

Prometheus, Grafana, Loki, Tempo, and evaluation scripts make quality and behavior observable.

Production

The boring pieces are already part of the product.

RunaxAI includes the operational layers that make an AI workspace maintainable: ownership checks, secure auth cookies, background ingestion, metrics, logs, traces, and evaluation scripts.

Application controls

Email/password and Google OAuth

HTTP-only JWT cookies

Protected chat and project routes

Rate limiting with auth-aware subjects

Runtime stack

Next.js app router

FastAPI SSE backend

PostgreSQL ownership model

Redis sessions, cache, jobs, memory

MinIO document storage

Pinecone vector retrieval

Prometheus, Loki, Tempo, Grafana

Pytest and retrieval evaluation harness

Ready when you are

Start with a chat. Grow into a workspace.

Open RunaxAI

Self-host locally or deploy with the included stack

Upload and ask

Project files become cited answers.

Research live

Web and browser tools fill the gaps.

Remember context

Durable facts stay available.

Measure behavior

Dashboards and evals expose quality.