ChatVLM is a complete AI workspace: chat, documents, research and agents, with a private pool of models behind it. It runs where your data already lives, so nothing sensitive has to leave the building to get answered.

Truly private AI.

Runs on a rack in your building, in a private cloud, or in ours.

Everything your people need, in one place

A chat that can do the work, not just talk about it. Answers come with the documents, files and sources they produced, and the same workspace is shared by the whole team.

Turn this report into a PowerPoint deck PowerPoint

Due diligence questionnaire

Northwind Growth Fund III · Answered from the fund documents

1. Firm overview

Describe the ownership and history of the firm.

Northwind Capital is an independent manager founded in 2011 and wholly owned by its four partners.

2. Investment process

How is a new investment approved?

By a unanimous vote of the three-person investment committee, after two rounds of diligence.

3. Service providers

Who audits and administers the fund?

Halvorsen & Co audits the fund each year, and Brightwater Fund Services administers it.

Target size: $400m
Management fee: 1.75%
Carried interest: 20%

Fill out this DDQ from our fund documents Word

	A	B	C	D	E
1	Bakery budget
2	Category	January	February	March	Q1 total
3	Flour and grains	2,140	1,980	2,260	6,380
4	Butter and dairy	1,620	1,540	1,710	4,870
5	Staff wages	8,400	8,400	8,900	25,700
6	Rent	2,800	2,800	2,800	8,400
7	Utilities	690	640	610	1,940
8	Packaging	420	380	450	1,250
9	Marketing	300	250	400	950
10	Total	16,370	15,990	17,130	49,490
11
12

Make me a monthly budget spreadsheet for a small bakery Excel

Proposal

Operations review for Halden Logistics

Scope

We map order-to-cash across the three warehouses, interview the twelve team leads and benchmark cycle times against the sector.

Approach

Two weeks on site, three of analysis, and a working session where your managers test every recommendation.

Deliverables

A findings report, a prioritised plan and a ninety-day roadmap your team can run without us.

Timeline

Six weeks from kickoff, with a progress review at the end of every second week.

Team

A partner leads the work, with two consultants on site and an analyst for the benchmarks.

Fees

A fixed fee of $48,000, invoiced in two instalments, with expenses at cost.

Next steps

Sign the enclosed letter and we start on the first Monday after.

Write a one-page consulting proposal as a PDF PDF

A watercolor lighthouse on a headland at sunset — Create a watercolor illustration of a lighthouse at sunset Image

One workspace, the whole toolbox. Nothing to install or hunt down.

Word docs, slide decks, spreadsheets, PDFs, image generation and editing, web search, deep research, a code interpreter, charts and analysis, translation, memory, questions about your files, Council Mode, personas and share links.

Council Mode. For decisions that matter, several models draft, and a judge checks and writes the final answer.

Projects and teams. Group chats and files into projects, share them with the people who need them, and build on the same context.

Personas. Assistants built for a job, such as a legal analyst or a research assistant, or your own with exactly the tools it needs.

Memory and files. Long documents, spreadsheets and past conversations stay searchable, and the assistant remembers what you told it.

Answers with receipts. Every reply shows which model answered, what was searched and what was cited, so anyone can check the work.

Standard APIs for agents. Coding agents and internal tools connect to the same private models by changing a base URL.

What did we promise Halden on response times?

A four-hour response on priority incidents, around the clock, and a service credit for every hour past it.

Halden MSA.pdf page 14“Supplier shall respond to Priority 1 incidents within four (4) hours, twenty-four hours a day…”

Legal analyst Answered by Tensor X Pro Sources · 3 retrieved · 1 cited

See how it works, in one conversation

Nothing retained. Nothing trained on.

Privacy here is not a switch in an admin panel, it is what the system does by default. Zero data retention and no training are in the contract on every plan, and a private deployment runs open-source models on machines you control, so there is no third party in the path at all.

Your buildingOutside

In our cloud

Content goes to the model to answer you and is dropped when the request completes, under contractual zero data retention on every plan, the free one included.

Your workspaceModel

In a private cloud

Sovereign Presets run leading open-weight models on ChatVLM-operated private infrastructure with zero third-party subprocessors.

Your workspaceOpen model, run for you alone

On a rack in your building

A Sovereign Box serves your people and nobody else’s. An air-gapped Box does not even need the internet to be up.

Your workspaceSovereign BoxAir gapThe internet

Nothing trains a model. Your conversations never become training data, for us or for anyone else. There is no opt-out to find in the settings, because there is nothing to opt out of.

Sovereign Box: the whole stack in one rack

Two appliances, one sized for an office and one for a datacenter. Both ship as clusters of at least three nodes, with unlimited seats, offline license enforcement, and quarterly model updates on encrypted physical media.

Theta: The office-ready private AI server, on standard wall power. NVIDIA or AMD datacenter and workstation GPUs
Omega: The datacenter deployment, for the largest open-weight models. Datacenter-class NVIDIA or AMD GPUs

Enterprise pricing

Three nodes to a cluster. Full capacity with two nodes down.

Privacy is not a paid upgrade.

Every plan, free included, comes with zero data retention in the contract and no training on your data.

Free

The core of the product, with daily limits.

Pro

$20 a month

No daily message limit and the larger models.

Ultra

$50 a month

The largest models and no daily tool limits.

Max

$400 a month

Coming soon. Everything in Ultra, with eight times the monthly allowance.

Teams

$30 a month, from

Shared workspaces, a common prompt library and one bill.

Enterprise

Custom

SSO, SCIM, private models and the Sovereign Box.

Start free See pricing

Put AI where the work happens

Talk to us about your deployment, or start on the free plan today.

Talk to us Start free