Private AI chat

Private AI chat for business teams, with EU-hosted models

A secure AI workspace for your whole team: 118+ models from 18 providers, EU-hosted options clearly labelled, personal data masked before it leaves, and answers grounded in your own documents. Your data is never used to train models.

Start free Book a demo

A shared FlowSentric chat with three participants online, typing indicators and the model set to Claude Opus 4.6 (EU)

118+models in one picker

18model providers

26EU-hosted models, labelled (EU)

4organization roles

14–365days of audit history, by plan

0prompts or documents used for training

Why teams need a private AI workspace

When a team uses AI without a shared workspace, everyone picks their own tool, pastes client data into it and keeps the results in personal accounts. Nobody can say afterwards which model saw what. A private AI workspace fixes the parts that matter to a business: one place with roles, a clear choice of where models run, personal data masked before it leaves, and a record of what happened.

FlowSentric lists every model with its hosting region. Models routed through our EU gateway carry “(EU)” in their name, and the picker can be filtered to EU-hosted or local models. If a chat model fails before writing anything, FlowSentric switches to another working model in the same hosting region, never across one. Application and database are hosted in the EU. We do not promise that data never leaves the EU: you choose the model, and we label where it runs.

People join your organization with a role: admin, editor, member or viewer. Restricted accounts see only the agents, models, tools, knowledge and features you allow, and can have Privacy Guard locked on. In a shared chat, colleagues work in one thread, and a participant's message runs on the owner's model, tools and privacy setting. Company sign-in through OpenID Connect or SAML, user provisioning through SCIM and two-factor authentication are available.

Upload the handbooks, contracts and price lists your team keeps asking about. Answers name the document and the page, and a citation opens the exact passage it stands for. Each person only gets answers from documents they are allowed to read, and you can restrict a question to one folder, one document or only documents someone has marked as verified.

What it does for you

Choose where models run

26 models labelled (EU) through our EU gateway, locally hosted models, and the direct providers, side by side in one picker grouped by hosting region.

Roles and restricted seats

Four organization roles, plus restricted accounts that see only what you allow and can keep Privacy Guard switched on whatever they prefer.

An audit trail you can export

Security-relevant actions are recorded with actor and time, kept 14 to 365 days depending on plan and exportable as CSV; a full account export comes as JSON.

How it works

01Start free and create your organization.

02Invite colleagues by email, give each a role and set up restricted accounts where needed.

03Upload the documents your team keeps asking about and decide which models are allowed.

04Chat, compare models side by side, share a chat with a colleague and export the result.

What is included

Chat

118+ models from 18 providers, grouped by hosting region
Compare 2 to 5 models on one prompt
Temporary chats that are not saved
Your own standing instructions and answer style

Knowledge

Answers cite document and page
Ask everything, one folder or one document
Only documents the person may read
A verified-only option for checked documents

Governance

Admin, editor, member and viewer roles
OpenID Connect and SAML sign-in, SCIM provisioning
Privacy Guard, lockable for restricted accounts
Never sold, never used to train models

Built for

Real-world scenarios

Consultancies working on client documents

Firms in regulated industries

Onboarding new colleagues with the handbook

Research with a second-opinion model

Good to know

The free plan (€0, no card) includes the chat, Privacy Guard, 4 agents, 2 workflows, 10 knowledge documents and 200,000 tokens a month. The EU-hosted “(EU)” models come with the Business and Custom packages.

Frequently asked questions

Is FlowSentric a GDPR-compliant AI chat?

It is built for GDPR use: application and database are hosted in the EU, a data processing agreement is ready to sign, the subprocessor list is public, personal data can be masked before any model and retention and deletion are under your control. Whether a particular use is compliant also depends on your legal basis and the model you choose.

Is our data used to train AI models?

No. Your prompts, documents and outputs are never sold and never used to train models, ours or anyone else's. Long-term memory is opt-in and can be deleted completely.

Which models are EU-hosted?

26 models are routed through our EU gateway and carry “(EU)” in their name, for example “Claude Opus 4.6 (EU)”. Locally hosted models run on our own servers. All others are labelled with the provider's region.

Can we control which models and features each person may use?

Yes. A restricted account gets exactly the agents, models, tools, knowledge and features you list, and Privacy Guard can be locked on for it. Everything else stays out of sight.

Can we share a chat with a colleague?

Yes. The owner invites colleagues into a shared chat; an invitation only becomes membership when the person joins. Everyone sees who is online, and a participant's message runs on the owner's model, tools and Privacy Guard setting.

What does the free plan include?

4 agents, 2 workflows, 200,000 tokens a month, 10 knowledge documents and Privacy Guard, at €0 and without a card. Everything you build carries over to a package; see pricing.