An AI agent inside your EHS system — not a chatbot on top of it
Ask in plain language, by text or voice. The agent reads your inventory, executes tasks with the platform's own tools, shows the result on screen — and keeps every change as a draft until someone approves it.
Things people ask it
What the agent does today
Occupational risks
Query, filter, group and count risks; show the probability × severity matrix and indicators per unit; create, edit, approve or delete risks — always as drafts with an audit trail; run a completeness “x-ray” of what is missing for approval; attach evidence from the conversation; restore its own operations.
Action plans
Create, edit and link actions to risks; delete them when asked.
Custom forms — full authoring
Read a form's pages, fields, questionnaires and rules; build a form from scratch; move, edit or remove elements; add conditional rules and AI instructions.
Screen bridge
The agent knows where you are and what is visible. It opens the risk inventory, risk detail, action board and form builder, applies filters, pages through lists and opens popups — screens update live as it writes.
HTML reports & dashboards
Writes and edits formal, printable HTML reports and dashboards from your data, with the client logo, and keeps them in the conversation.
Images & attachments
Lists, loads and shows images in the chat — attachments, evidence, logos.
Spreadsheets
Reads Excel structure (sheets, formulas, dropdowns), cells and ranges; searches text; turns rows into records.
Scheduled extractions
Creates, lists and deletes scheduled extractions and reports delivered by e-mail — AI mode or pure extraction.
Support
Opens a support ticket to the RiskEx team by e-mail, with attachments.
Memory & supervised learning
Per-conversation memory, resumption of older requests, question cards and unit selectors, plus a supervised mode that records consolidated knowledge.
Human in control, by design
Draft until approved
Anything the agent writes is a draft. Approval — by a person with the right profile — is what changes the inventory.
Your permissions, not its own
The agent acts with the permissions of the user talking to it: unit scope, module access, view/edit/delete.
Audit trail and undo
Every operation is logged and restorable. You can list what the agent changed and roll it back.
Honest about limits
“AI can make mistakes. Always validate the data.” is written under the prompt box — on purpose.
Hosted on Microsoft Azure, alongside the platform. Model providers and usage limits are configured per customer. The agent is rolling out in beta with selected customers and improves continuously.
See RiskEx working on your data
A 30-minute live demonstration with real workflows — no slides. Bring one of your own risk scenarios and watch the platform (and the agent) handle it.
Book a demoor write to contact@riskex.ai · +55 (11) 2817-0735