Semantic Agent, grounded in what you trust
Bring your sources, plug in your apps, get cited answers from the source they came from.
What the Semantic Agent does
A reasoning agent over the sources you trust and the apps you already use
Cited answers, every time
Every claim links back to the source it came from — open it, verify it, move on.
Web search when your sources don't have it
The agent decides when to look beyond your corpus and brings the citation back with it.
Connect the apps you already use
Google Drive, Notion, SharePoint. Live access, no stale copies.
Plug in your own integrations
Any service that speaks MCP — the agent learns whatever endpoints you expose.
Visuals come with the answer
Diagrams, schematics, screenshots pulled straight out of your sources.
Long-form artifacts that stick around
Reports, code, charts the agent generates persist across the conversation.
Ask in any language. The agent understands and replies in yours.
Your sources, on your terms
The agent grounds on what you decide to feed it — and you decide who else gets to ask
Ingest any source — documents, manuals, slides, spreadsheets, drawings. Audio, video, and more on the way.
Just you
Private workspace, never leaves your account.
Share with a person
Pick a person among your groups: they accept the share themselves, with no email and no link.
Share with a group
An administrator of the group accepts the share for all its members; you withdraw it at any time.
Groups run by your company
Your company creates its groups and chooses their members and administrators: each group reads only what is entrusted to it.
In the wild
A few of the places HeyJo answers questions
A field tech on a solar install
On site, midway through wiring an inverter, an installer asks the agent a question against the manufacturer's manual. The cited answer comes back with the diagram from the spec sheet — no flipping through a PDF on a phone.
“Where exactly does the AC disconnect bracket mount on this inverter model?” — answered, cited, with the diagram.
For teams and enterprises
Deploy with confidence
Security & Compliance
- No training on your data — Your sources never train any model — not ours, not our AI providers'.
- Security testing — Automated data-isolation tests on every change; independent penetration test planned.
- Data location — Your documents are stored in France. Some AI processing runs outside the EU; the list of subprocessors is available on request.
- Controlled file access — Encrypted in transit; files are opened through short-lived signed links, and every access is logged.
Access management & integration
- Single sign-on — SAML single sign-on with your identity provider, planned for Q1 2027.
- Fine-grained permissions — Per-user and per-group access on the sources, the agent, and the tools it can call.
- Centralized controls — Configure model access, source visibility, and MCP allowlists from one place.
Control & advisory
- Budget control — Flat, predictable pricing with caps and forecasts that align with procurement.
- Dedicated guidance — Our team supports rollout, from pilot to organization-wide deployment.
- Premium support — Tailored SLAs and priority response for teams with specialized needs.
Frequently asked questions
Start grounding your AI in what you trust
Pick a plan to get going, or get in touch if you're rolling HeyJo out across a team or an organization.