Your company's data, ready for AI agents in days, not months
Vecten Open is an open-source data platform for the AI-agent era. An agent sets it up in your own cloud. Your people ask it questions from Claude or ChatGPT and see only what they're allowed to.

Two pieces. Use both, or only the one you need.
Vecten Core
For a company with no data warehouse yet. Core connects the systems you already run, keeps the data current and organizes it into one consistent picture, in your own cloud account.
About Vecten CoreVecten MCP
For a company that already has a warehouse, or has just built one with Core. People ask questions from Claude and other AI assistants. Each sees only what they're allowed to, and every answer comes with a receipt.
About Vecten MCPYour systems in, answers out. All inside your own cloud.
Core brings in the data from the systems you already run. MCP lets your people ask it questions from Claude or ChatGPT. An AI agent does the building and the upkeep, and a person approves every change.
The systems you already run
- CRM
- Finance
- Bookings
- HR
- Spreadsheets
- Any API or database
Your data warehouse
All your data in one place, with one definition of each number.
Vecten MCP
Lets people ask the data questions. Each sees only what they're allowed to.
About Vecten MCPNothing passes through Vecten. The data, the code and the record of every question stay in your account.
Your people and their AI assistants
- Claude
- ChatGPT
- Your own agents
A data platform used to take a team and most of a year. Now an agent builds it in days.
Most companies never got a data warehouse, because the first step was a senior hire or an expensive product. The ones that did often ended up waiting on one engineer. AI agents change both.
Getting a data platform
Before
- Hire a data engineer
- Buy the tools
- Build pipelines
- Model the data
- First dashboard
- Hire a data engineer
- Buy the tools
- Build pipelines
- Model the data
- First dashboard
6–12 months
With Vecten Open
- Tell an agent
- Approve each step
- Ask questions
- Tell an agent
- Approve each step
- Ask questions
Days
Adding a source or a metric
Before
- Raise a ticket
- Wait in the queue
- Engineer builds it
- Next release
- Raise a ticket
- Wait in the queue
- Engineer builds it
- Next release
Weeks
With Vecten Open
- Ask an agent
- Approve the change
- Ask an agent
- Approve the change
Same day
Answering a question
Before
- Email the data team
- Wait for a query
- Get a spreadsheet
- Ask a follow-up
- Email the data team
- Wait for a query
- Get a spreadsheet
- Ask a follow-up
Days
With Vecten Open
- Ask Claude or ChatGPT
- Answer with a receipt
- Ask Claude or ChatGPT
- Answer with a receipt
Seconds
Five promises. Why we built it.
We've spent 15 years building data platforms for private capital and healthcare. Vecten Open is everything we learned, rebuilt for a world where AI agents do much of the work. These five promises are what drove it.
Open source
Apache 2.0
Free to use, change and sell, with no licence key and no tracking. Everything a team needs to run it is open. A few enterprise extensions, such as single sign-on, are paid.
Read the licence (opens in a new tab)Malleable
Yours to reshape
We ship a working starting point, not a finished box. Whoever needs a new source or metric asks an agent and gets a change to review, so nobody waits on one engineer.
Agentic flow and malleable software (opens in a new tab)A data platform
One definition of each number
All of your data in one place, with one agreed definition of each number. Every source is organized the same way, so a new engineer or a new agent finds their way around without a tour.
Vecten CoreFor the agentic workplace
Your agent is the interface
Setting it up, extending it, using it and keeping it running all happen by talking to an AI agent. The know-how ships with it, and how much it helps the agent is measured in public.
Skill evaluations (opens in a new tab)Secure by the warehouse's own rules
Every question as the person asking
Your warehouse decides who sees what, not the AI. Every answer carries a receipt, and admins see who asked what and which questions failed.
Vecten MCPProven open-source tools. Put together for agents.
Vecten Open puts widely used open-source tools into one consistent setup an agent can drive. dlt loads each source, Apache Airflow runs it on a schedule and dbt models it, on Postgres or Snowflake. People reach the warehouse over the Model Context Protocol, the open standard Claude and other assistants use to reach data.
Other products ship parts of it: data platforms on one side, MCP servers and semantic layers on the other. None ships both together, open source, on Postgres and Snowflake, with each person querying as themselves. The comparison is public (opens in a new tab).
- Pieces
- Vecten Core, the platform, and Vecten MCP, the server, behind one command line:
vecten. - Stack
- dlt, Apache Airflow and dbt, with an Apache Ossie semantic model.
- Warehouses
- Postgres, which Core can set up for you, or your own Snowflake. Serving BigQuery, Databricks and Redshift through the MCP is planned.
- Clouds
- Major cloud providers like AWS, deployed with Terraform and GitHub Actions.
- Assistants
- Claude, ChatGPT and any client that speaks the Model Context Protocol (MCP).
- Install
uv tool install vecten, or point Claude Code at an empty directory and say what you want.

Run it yourself, or let us run it for you.
Vecten Open is free and complete. It runs in your own cloud account, nothing passes through Vecten, and you don't need a partner or a paid plan. If you stop working with us, you keep everything: the code, the data, the definitions and the record of every question. When you'd rather not run it yourself, or need it built, connected and compliant, that's the work Vecten does.