Product/AI assistants

Bring your AI assistant to your architecture.

Connect Claude, ChatGPT or any assistant that supports the Model Context Protocol. Ask it anything about your estate, and let it draft diagrams and decision records for your team to approve.

Start free Book a demo
We're retiring the legacy rating engine next quarter. What still depends on it?
search_elements · get_element
Three things still call it: Quote & Bind, the Broker Portal and the nightly renewals batch. It appears on four diagrams across two teams, and the renewals batch is marked business critical.
Draft a decision record for the migration.
create_adr
Done. "Retire the legacy rating engine" is saved as a proposed decision, ready for your team to review.
An example conversation in Claude, using Voussar's AI tools

Ask

Answers from your real architecture, not guesses.

Your assistant reads the same catalog, diagrams, data flows and decisions your teams do, so its answers are grounded in what you actually run.

"What depends on the payments gateway?"Every system that calls it, and every diagram it's on.
"What are we running that's near end of support?"Every system on a technology that's due to lose support.
"Which integrations carry financial data?"Straight from your data flow register.
"Why did we choose Kafka for order events?"The decision record, with its status and history.

Draft

Let it do the drafting. You keep the say.

Hand your assistant a design document or a change note and ask it to model it. It shows you exactly what it plans to add before anything changes, then builds the diagram and lays it out.

  • Preview first. Every change is proposed as a plan you can read before it's applied.
  • Same approvals as everyone. Diagrams an assistant creates go through your normal review.
  • Clearly marked. Anything it adds to the catalog shows as unconfirmed until a person checks it.
  • It can't delete. Assistants can add and propose, but never remove.
A system context diagram of the kind an assistant can draft and lay out automatically.

Control

Your data, your rules.

Voussar runs no AI of its own and sends nothing to an AI provider. Your organisation's owners decide exactly who may connect an assistant, and every connection can be revoked in a click.

  • Sign in to connect. Assistants connect through a standard sign-in and consent screen, with no passwords shared.
  • Scoped access. Choose what each connection may do, from read-only upwards.
  • Keys that expire after 90 days by default, and never last more than a year.
  • A fair daily allowance on every plan, Free included.
The AI access settings, where organisation owners choose who may connect an AI assistant.

Up and running in three steps.

1. Allow it

An owner turns on AI access for the people who should have it.

2. Add Voussar to your assistant

Paste your organisation's Voussar address into Claude, ChatGPT or another assistant.

3. Sign in and ask

Approve the connection, and start asking questions about your architecture.

Architecture catalogThe source your assistant reads from. Reviews & governanceThe approvals its drafts go through. SecurityHow your data is protected.

Put your architecture on solid ground.

Start free and bring one real system across. You'll know within an afternoon whether your diagrams can stay true.

Start free, no card needed Book a demo

Free plan forever · 14-day trial of every paid plan · Hosted in the UK