A fixed process with AI steps.
Every step is known before it runs. AI does the reading and sorting in the middle.
Best for: admin that repeats, at volume.
Explore Agentic Workflows →Company Brain & Company Agents
Ask your business a question and get the answer from your own records. Then hand the busywork to agents that check with you first.
Picture a building company. A client rings about a variation agreed three months ago. The answer is somewhere in an email thread, a Teams chat and a site note, and the one person who'd know is on leave.
So the job waits, or someone guesses.
A Company Brain finds it, and shows you where it came from.
The Brain reads your email, Teams chats, SharePoint and OneDrive, documents, tickets and calendars. Ask it something in plain English and it answers from your own records, with the sources attached.
What it reads
The Brain
What you get back
What did we quote the last time we did a job like this?
Two similar jobs last year. Here's what we quoted, and what changed on site.
Sources
Shown because you can open all three.
Built on open-source agent tools. The Brain isn't tied to one AI model, and the code we write for you stays yours.
Together they're your business's long-term memory. Each agent has a name and one job. Before it starts, it asks the Brain what's going on. When it's done, what it did goes back into the Brain, so the next agent, or the next person, starts with the full story.
Company Agents + Company Brain = long-term memory
The memory runs on the same open-source tools as the Brain.
Sorts the inbox, flags what's urgent and drafts replies for you to check.
Writes the weekly report from what actually happened in email, chats and tickets. Not from memory on a Friday afternoon.
Reads an enquiry, finds similar past jobs and drafts a quote for your review.
Chases approvals that have stalled and tells you who's holding things up.
Before a meeting, tells you who's coming, what was said last time and what's still open.
Spots work happening in email and chats with no ticket, so it's logged before it's forgotten.
A scenario
The Inbox Agent drafts the reply and notes the request.
The Quote Agent splits the new quote by site without being told.
A new starter asks why. The Brain shows them the original email.
They ask before they act. Sending an email, changing a record or committing money waits for your yes. Everything an agent does is logged, so you can always check what it did.
Before we offered agents to anyone, we built our own and we run them every day.
Our agents share one memory of our work: what was asked, what was done and what's still open. It's where we learned what a Company Brain needs.
Each agent job starts in its own microVM on our own private cloud, and the microVM is removed when the job's done. One job can't see into another.
Reads first-party AI news and writes each story. A second model checks every claim against the original source, and a story that fails is held as a draft.
One system you own, from CRM to BAS. You pay once for the build, with no monthly fees for core functions.
One place for a team's work, with a ledger that can't be edited. Now in pilot.
We run all of it ourselves.
Same invoice, two kinds of job. A workflow does the part that's the same every time. An agent handles the part that isn't.
Every step is known before it runs. AI does the reading and sorting in the middle.
Best for: admin that repeats, at volume.
Explore Agentic Workflows →You give it a job, not a script. It plans the steps itself and checks the Brain before it starts.
Best for: work that's different every time, like replies, reports and follow-ups.
Admin handled by agents. You approve the exceptions.
Small steps, each one useful on its own. Nothing gets access to change or send anything until you've seen it work.
What the agents can do at each step
Where your information lives, and which jobs are worth handing to an agent.
Your sources are connected read-only. Nothing can be changed or sent.
Your team asks it questions and checks the answers against what they know.
Each agent starts with one job and an approval step. You decide when it's earned more.
We watch what the agents do, step by step, and tune them as your business changes.
The Brain works inside the systems you already use and the permissions you've already set. Adding AI doesn't change who can see what.
Example · same question, two people
How did we go on the last kitchen job?
Under target, because of two variations we didn't charge for.
Source: Job costing sheetThat's in a file you don't have access to. The owner can share it with you.
Scripted examples of the format. They aren't connected to any business.
No. It connects to the email, chats, documents and tickets you already use, so your team doesn't move into anything new.
No. Answers follow the permissions you already have. If someone can't open a file today, the Brain won't show it to them.
Not for anything that matters. Sending an email, changing a record or committing money waits for your yes. Everything an agent does is logged.
You do. The code we write for you is yours, with no per-seat fees. If we use LaunchStack modules, those are rented from us and the module code stays ours.
No. The Brain and the agents are built on open-source tools, so there's no licence fee for them. Every step is recorded on our own account, in Australia. If your setup needs extra tracing and testing, we'll show that cost in the quote.
Every build is quoted to the job after a free 30-minute call, and the price is fixed before work starts.
We'll look at where your information lives, which jobs are worth handing to an agent, and where it makes sense to start.