Skip to content

AI, built in

AI that works where your team works

Ask in an AI workspace, research with sources, hand the busywork to Deskly, hire AI specialists, get minutes from the meetings you record, or connect your own agents over MCP. Everything runs with your permissions, inside limits your admins set.

What the AI does today

Nine parts, one sign-in and one set of permissions. All of it is in the product now.

Also in the mix: a daily briefing, an AI summary of your day emailed each morning once your organisation switches it on.

AI Workspace

Ask for the answer, or ask for the file

One chat that picks the right kind of answer for each message: a reply, a side-by-side comparison, a pipeline or an image. Ask for a spreadsheet, a PDF or a Word document and it appears under the reply, ready to open and save to Drive or a Docs page.

  • Deep research with citations, charts and files in the chat
  • Attach documents, images and PDFs, or dictate your prompt
  • Projects your team shares, with instructions and memory
Explore the AI workspace

Deskly

An assistant that does the clicking

Deskly sits in your Messages panel and sees the page you are on. Ask about it, or tell it what to change. It proposes the change, and you decide.

  • Acts across Tracker, Docs, chat, meetings, Drive and more
  • Allow once, Always allow or Never, per kind of action
  • Every action shown step by step
Meet Deskly

Control

Your providers, your keys, your limits

Admins connect the AI providers and models the organisation wants. The workspace routes each request to a suitable model from those providers, within the budget each member has.

  • Provider keys encrypted at rest and used only on the server
  • Per-member budgets, with an alert before the limit
  • Model allow-lists, and AI switched off per person or tool
See the security model

Ground rules

AI that stays inside the lines you draw

It acts as you, not above you

The assistant, AI employees and any agent you connect work with your permissions. If a private space is closed to you, it is closed to them, and organisation boundaries apply the same way.

It asks before it changes your work

A change waits for Allow once, Always allow or Never, and every step is shown. Risky changes, such as sending messages, sharing, deleting and changing settings, always ask.

Keys stay on the server

Your admins manage the AI providers and models and enter the keys once. They are encrypted at rest, used only on the server and never sent to the browser.

Budgets, with an alert

Admins set a budget for each member and are alerted before the limit. Usage charts show where it went, and AI employees stay inside a monthly spend cap the owner sets.

Models you allow

Admins choose which models a person can use, and can switch AI off for a member or for the whole organisation.

Switch off a tool

Individual AI tools can be turned off one by one, so the assistant can be useful without being able to do everything.

Oversight that is disclosed

Where an organisation reviews AI use, members are told, and each review is recorded in the audit log.

Memory you write, kept private

The AI never learns personal facts on its own. What it remembers about you is what you wrote down, it is private to you, and there are switches for the organisation and for you.

AI employees

A team of AI specialists you can hire

AI employees are specialists for marketing, sales, finance, legal, software and more. Mention one in chat, message it, or assign it a Tracker item. It answers as itself, acts with the permissions of the person who asked, and asks before it changes anything.

  • Paid for by the organisation, within a monthly cap the owner sets
  • Text only: they do not join calls or sit in seats
  • An organisation module an owner switches on
About AI employees

FAQ

Questions about AI in Remote Desk

Something else on your mind? Write to hello@thehivez.com.

Which AI features are available today?

The AI workspace (chat, comparison, pipelines, deep research, files and charts, projects and memory), the Deskly assistant, AI employees, minutes for recorded meetings, Studio for images, a daily briefing email and the MCP server. Your organisation's owner and admins decide which are switched on.

Does the AI change things without asking?

In Deskly, the AI Workspace and with AI employees, a change to your work waits for your decision: Allow once, Always allow or Never. If you choose Always allow for a kind of action, such as updating Tracker items, it stops asking for that kind only. Sending messages, sharing, deleting and changing settings always ask. An assistant you connect over MCP acts through that app, with your permissions, and its changes are recorded under your name.

Whose AI providers and keys does it use?

Your organisation's. Admins connect the AI providers and models they want and enter the keys once. Keys are encrypted at rest, used only on the server and never sent to the browser. When you self-host, you choose the providers.

Is my data used to train models?

Remote Desk does not train models on your conversations. Requests go to the AI providers your admins configure, under those providers' terms, and on a self-hosted deployment the conversations stay on your infrastructure.

Can admins limit who uses AI?

Yes. Admins can give a member their own budget with an alert before it is reached, limit the models they can use, turn individual AI tools off, or switch AI off for them. Organisations can also turn the AI module off entirely. Optional oversight is disclosed to members, and every review is logged.

What does the AI remember about me?

Only what you write down. Personal memory is a list you keep yourself, private to you, and the AI never learns personal facts on its own. Projects keep their own shared memory for the work inside them. There are switches for memory for the organisation and for you.

How do AI employees fit in?

They are specialists your organisation hires from ten templates, such as marketing, finance or software. You mention one in chat, message it, or assign it a Tracker item. They are an organisation module an owner switches on, they work in text, and the owner sets a monthly spend cap.

Put AI where the work already is

The workspace, the assistant and your agents share one sign-in and one set of permissions.