Welcome
Chats
Channels
Talk to your AI from your phone — outward actions still come back to you to approve.
Plugins
Everything your AI can do, and the skills that come with it.
🔒 Credentials are stored locally in data/aios.db and never uploaded.
Heartbeat
What runs on its own — scheduled skills and the proactive check-in.
🏢 Agent Organization
Create a new organization or join an existing one to collaborate with other AI agents and people.
Create new org
Start your own organization. You will be the owner and can invite peers by URL.
Join an org
Paste the invitation URL or code you received from the admin.
🏢 Company
Your members, teams and network.
People and AI recruits in your Company. Recruit a role to add an AI member.
Group members into teams you can pull into a Discussion.
All reachable agents across your organization and paired peers.
Everyone in the org uses the same token. Rotating it disconnects all current peers until they receive a new invite.
—Peers use your current network token automatically. Manual peer entry is optional — most members will join via an invite.
Memory
What your AI has learned about you, captured from chats and kept on this machine.
Files
Everything your AI made while working. Stored only on this machine.
Photometrics
Read an IES or LDT file: output, beam angle, distribution and glare. Nothing is uploaded.
Mind maps
Structures your AI drew, and the ones you saved. Open one to rearrange it.
Company
Create or join a company to collaborate with your colleagues' AIs.
Create a company
Start your own company. You become the master (owner) — you set it up, invite people with a code, and manage members and teams.
Join a company
Got an invite code from a company's owner? Join it as a member and collaborate with your colleagues' AIs.
Group members into teams you can pull into a Discussion.
Company profile
Your company's shared identity and the knowledge every AI in it is grounded in.
A short brief carried in every AI's context on every message — who you are, your terms, how you work. Capped at 8,000 characters, because it is paid for on every turn.
Your processes, policies and field meanings, stored whole and searched only when a question needs them — so they can be as long as you like.
What every AI in this workspace knows about your company — culture, values, processes, people, terminology. It's woven into the AI's context for chat, skills, task work and consultations.
Your company's own documents — how a process works, what your terms mean, which system holds what, who approves what. Unlike the box above (carried in every AI's context at all times, and capped), these are searched only when a question needs them, so they can be as long as you like. Upload the file you already have rather than retyping it.
Ready-made reference documents that ship with QuorieAI — the law, not your policies. Your AI can only answer statutory-deadline and entitlement questions once one is installed; it holds no rates of its own on purpose, so a figure can be corrected here without a software update. Install once from any workspace — it publishes to everyone in the company, including your AI teammates. Installing again refreshes it rather than adding a second copy. Your own employee handbook is a separate thing: upload it above.
Type a real question. This runs the same search your AI runs — if nothing comes back here, your AI won't find it either.
Your company's colours, typeface and letterhead, used by every document your AI produces — Word files, reports, decks. Set it once and it applies to everything built from here on. Nothing already made is changed: a document somebody has quoted must not silently become a different document.
🔒 A company admin, or someone holding the brand author permission, can change this. It applies to documents built on this workspace.A logo goes out of the building. Every document built with it travels — Teams files, shared rooms, emailed reports — so upload the mark you already publish, and keep anything internal out of the footer line above. PNG, JPEG or WebP; no SVG.
Documents this workspace produced in the last 30 days, and which revision of the brand each was built with. This is here to answer the question you have while looking at the card above: did my change actually reach anything?
Permissions & approvals
What people may install, who signs off on what, and how far the AI may go on its own. Applies to everyone in the company.
Plugins already installed keep working whatever you choose here.
Below the line, people approve their own actions. From the line up, their manager does — and can block it. A person's own limit, set on their job role, overrides this.
Applies to everyone in the company and takes effect immediately.
Email, Teams messages and posts wait for Approve in the Inbox.
PDF, Word, PowerPoint, QR codes, charts… — installed safely in its sandbox.
With “Approved sites”, you get a card naming each site before its code first contacts it.
Blocks the AI's code from network and operating-system access unless it is truly isolated.
Before a company document is handed to the AI as fact, a cheap model checks it answers what was asked. Off is faster and slightly cheaper, but a question none of your documents cover gets answered from whichever is closest.
The defaults suit almost everyone.
How many steps the AI may take on one request before it wraps up.
Tokens in one AI reply.
Megabytes per attachment.
Near the model's limit, older messages are summarised instead of re-sent. Off always sends the whole conversation.
Tokens the model can hold. 0 detects it from the model — recommended.
Plugins
What this workspace runs, plugin by plugin — who each skill is for, and whether its last run worked.
A skill can be individual (everyone runs their own), or belong to a role, a team or the whole company — one run, one answer, for that audience. Setting this records the decision; the shared run itself is the next piece of work, so today every member still runs their own copy.
Loading…
Automation & memory
What each person's AI does without being asked, what it learns, and how it keeps its memory tidy.
Each person's AI reviews their tasks and Inbox and posts a short “here's what I'd do” note. Report-only — it never sends or changes anything. People can still change their own in their Inbox.
AI teammates answer messages addressed to them in community and group-chat rooms.
Act on approvals, questions and tasks by themselves. “AI teammates only” leaves people's Inboxes for them.
Anything a scheduled run would send, post or reply waits until a person approves it in their Inbox.
The shared task board and its AI tools. Off hides it everywhere and stops the follow-up sweeps — nothing is deleted.
Each night every member's AI offers the working rules it has learned to the teams that member is on, and the team's lead decides what the team adopts. Nothing is added on its own; a lead who isn't a company admin has their approval co-signed by one.
Turns each person's chat history into memories their AI keeps. Off by default, for privacy.
When a job goes badly or expensively, the nightly review may read the web for a better method. Web pages only — it can't reach your plugins, and what it learns still needs approval and shows the page it read.
Tidies, de-duplicates and retires stale memories overnight.
“Suggest only” sends each change to the Inbox for a person to approve.
AI Model
Which model each job runs on, and how hard it thinks.
Each job runs on the best model or the standard one, and thinks harder or not at all. Chat should stay on the best model; routine background jobs rarely need it. Spend is over the range chosen on Usage & advanced.
Usage & advanced
Where your AI spend goes, network reachability, and advanced peer-request limits.
Where your AI spend actually goes — by model, by what the AI was doing, by tool, by person, and by chat. Costs are estimates from published list prices, not a bill. Visible to admins only.
Advanced
Reachability — Online mode
Turn on to make this company reachable by members on other machines over the internet, and to keep your invite code's address live.
Advanced — peer request limits
Guards against a flood of incoming requests from other machines on your network — each one runs the AI on your model & key, so these caps protect your usage budget. The defaults suit almost everyone; change them only if you know you need to.
Job roles
The jobs at your company. A role gives whoever holds it whole plugins, the systems they sign into, and how their AI should work.
People & roles
Click a teammate to set who they report to, assign roles, and see the org chart.
Mailboxes on your company domain. Creating one gives that person an address they can connect in Marketplace → Email.
HR
What your HR AI reads, and what it still needs before it can answer.
Each step below is something an HR skill depends on. A skill whose step is missing still runs — it just can't answer, so check here first if an answer looks empty.
Employment rules and filing deadlines for each country you employ people in. Your HR AI holds none of these itself — it quotes this document and says when it was last checked, so correcting a figure here fixes every future answer with no software update. Installing again refreshes it rather than adding a second copy.
Birthdays, join dates and work-permit expiries your AI watches, so it can remind you before each one. It reads these from Odoo first — add someone here only to fill in what Odoo doesn't have. Leave the year off a birthday you don't know by using 1900.
Title
About this skill
No summary yet. Press ▶ Run to generate it now, or open ⚙ Settings to choose what it includes and schedule it.
Runs on this machine's local time while the app is running. For a digest, set a “Post to Teams” channel above so it delivers itself.