AI Assistant · add-on

Ask your data anything.

The Portanus AI Assistant answers questions about your own program in plain language — grounded in your data, filtered by each person's permissions, and audited on every step. No new dashboards to learn; just ask.

See it on your data $99/mo · any plan

What it does

Answers, grounded in your workspace

Plain-language answers

"How many dogs are in advanced training?" "Which applications are waiting on interviews?" Ask in words, get answers from your live data.

Bound by permissions

The assistant only ever sees what the asking user is allowed to see. It can't reveal a record a person couldn't open themselves.

Every step audited

Each thing the assistant reads or does is written to an append-only audit log — you can always see exactly what it looked at.

Safe actions

Beyond answering, it can help with tasks (add a note, advance an application) — but consequential actions go through a propose → confirm → execute step. It never acts on its own.

You stay in control

Operators set daily request quotas, cost caps, and exactly which tools are enabled. A blocked request degrades to a friendly reply — never a runaway bill.

Your data stays yours

Your records are never pooled with other organizations or used to train external models. The assistant works on your workspace, for your workspace.

Built for organizations that handle sensitive information

Access is gated by license and per-user permission, every read and write is logged, and the whole assistant can be turned off with a single operator switch. It's an assistant your board and your handlers can trust.

Simple pricing

$99/month, on any plan

Add it to Essentials, Program, or Complete — or $990/year. Turn it on when you want it, off when you don't.

Can it change our data?

Only within guardrails. Reversible notes and logs can be written directly; anything consequential (advancing an application, sending a message) requires you to confirm the exact action first. It's off by default until an operator enables actions.

Will it show someone data they shouldn't see?

No. Every answer is filtered to the asking user's own permissions — the assistant can't surface a record that person couldn't already open in the app.

How do we keep costs predictable?

Operators set a daily request quota and a daily cost cap per organization. Once a limit is reached, the assistant politely declines further requests until the next day — no surprise bills.

Is our data used to train the AI?

No. Your records aren't pooled with other organizations or used to train external models. The assistant reasons over your workspace to answer your questions, and that's it.

See it answer questions about your program

Book a demo and we'll show the assistant working on a workspace like yours.

Request a demo