Chatbots that cannot act
Most AI tools stop at answers; the real work still happens by hand.
Answers from your own knowledge.
Ask across your organisation’s documents and systems. Answers come from what each person is allowed to see, with the source lit up.
Ask across your organisation’s documents and systems. Answers come from what each person is allowed to see, with the source lit up.
Agents act through signed plugins, using the permissions of the person who asked, never a master key.
Anything that changes a system or carries risk stops at a clear checkpoint until the right person approves it.
Each action is written to a tamper-evident trail with the agent, the person it acted for and the outcome.
Enterprise product · myAi
myAi brings chat, knowledge and specialist agents together under the same permissions, approvals and audit as the people they work for.
In use at R92 today
Most AI tools stop at answers; the real work still happens by hand.
Giving an agent a shared key means it can do anything, for anyone, with no one checking.
When something goes wrong, there is no trail of which agent did what, for whom.
Ask questions across your organisation’s documents and systems, within each person’s access.
Agents act through signed plugins that run separately from the platform.
Calls to other R92 products use the requesting person’s own permissions, never a master key.
Actions that change systems or carry risk wait for a human approval.
Runs on your own servers today; the same package is designed for hosted and fully disconnected use.
A developer kit to build and sign agent plugins.
Sign-in and the per-person tokens agents act with.
Agents work in each product through its own permissions and approvals.