AI agents that follow your team’s approval rules.
An AI agent completes a specific task from start to finish, such as onboarding a new supplier or matching invoices to payments. We fix the process first, then build the agent, launch it and run it. Your team decides which steps need approval.
Why we fix the process first.
If nobody agrees who approves a request or which spreadsheet is correct, those decisions need to be settled before an agent can follow the process.
We map how the work is done today, remove unnecessary steps and agree which tasks the agent will handle.
Who it's for.
- Skilled people on your team spend hours each week copying data between systems, checking documents or chasing approvals.
- You tried an off-the-shelf AI tool, and it couldn't handle the exceptions in your process.
- You want to automate, but you can't accept an AI making decisions nobody checks.
What you get.
A redesigned process
We map the work step by step with the people who do it. Then we agree which steps the agent takes on and which stay with a person.
An agent that works in your existing systems
The agent works inside the tools your team already uses, such as email, your finance system or your CRM. For example: it receives a supplier request, checks for duplicates, verifies the documents, asks your team to approve, then creates the supplier record.
Rules you control
You decide what the agent can do on its own and what needs a person's sign-off. The activity log records the steps taken, the results and any approvals, so your team can review each run.
Results measured against today
We agree on one measure, record the starting point and set a target. For example: hours spent onboarding suppliers each week. Reports compare the results with that starting point and account for running costs.
How it works.
The schedule below is illustrative and shows a 12-week project. Allow 8–12 weeks for a first agent; we agree your schedule before work starts. Ongoing support is covered by a separate monthly retainer.
- Weeks 1–3
Map and redesign the process
We sit with the people who do the work, map every step and record today's baseline. We remove unnecessary steps and agree where a person approves.
- Weeks 4–8
Build and test
We build the agent and test it on real past cases, including the awkward ones, before it touches live work.
- Weeks 9–12
Launch alongside your team
The agent handles live work while your team checks its output. We expand its scope only after reviewing the results with you.
- Ongoing
Run and improve
On a monthly retainer, we monitor the agent, fix issues and update it when your process changes. You get a regular report on the agreed measure.
What we need from you.
One process owner who can decide how the work should run.
Time with the people who do the work today, because they know the exceptions.
Past examples of the work, so we test on real cases, not demos.
Access to the systems the agent will use, agreed with your IT team.
We keep it running.
Our monthly retainer covers maintenance as your processes, documents and software change.
- Agents. We monitor them, fix issues and update them when your process changes.
- Context. We add new documents, retire outdated ones and resolve conflicts with the owners.
- Systems. We maintain the connections your agents and Context use, and update them when the connected software changes.
We agree what is covered, set the fee to match and report on the measures you chose.
You stay in control.
Before we build, we agree what data the agent can access, where it's processed and who can use it. The agent can only use the systems and actions you approve. You can pause it at any time.
Bring us one process.
Tell us about a task your team repeats every week. On an initial call, we'll discuss whether an agent could help, what we would need to check and whether a simpler tool could do the job.
Get in touch