What an agentic process redesign actually looks like
The phrase “agentic process redesign” is the kind of language that makes a sensible operator’s eyes glaze over. Fair enough — it sounds like a consultancy slide deck.
So let me ground it. Here’s what an engagement actually looks like, end to end, with a worked example.
The shape of the work
Six weeks. Four stages. One process picked on purpose, redesigned end to end, and live by the end of week six.
Week 1–2 — Diagnose. We map your repeatable processes (it’s usually about 15–25 of them in a £20m business), then score each one for agent-fit. Which are high-volume, painful, well-defined, with data that already exists? Those are the candidates. We pick one to start.
Week 2–3 — Design. We take that one process and rebuild it on paper. What does the new workflow look like? Which steps are agent-led? Where does a human stay in the loop? What tools do the agents need access to? Where does the data come from and go to?
Week 3–6 — Deploy. We stand the agents up in MAF. Wire the integrations. Configure the templates. Train the people who’ll work alongside them. Switch it on, with telemetry watching from day one.
Ongoing — Drive. Iterate against real data. Add the next process. Build the organisational memory that makes your agents sharper over time.
A worked example: post-meeting follow-up
Let’s take a process most service businesses have: the loop after a customer meeting.
Before the redesign: 1. Account manager has a discovery call. 2. They take notes (badly, while talking). 3. They mean to write a proposal afterwards. It happens 2–4 days later, often when something else has slipped. 4. They draft an email with next steps. Forget to add the new contact to the CRM. 5. The follow-up either happens or doesn’t.
This is fine when you’ve got two prospects. It’s a disaster when you’ve got fifty.
After the redesign: 1. The meeting is captured automatically — a bot joins the call, or the account manager records it locally. 2. Within 30 minutes, an agent has produced a clean summary, extracted the discussed pain points, and drafted a proposal in the right format for that prospect type. 3. The account manager reviews the draft on the train home. Edits the parts that matter. Sends it. 4. The CRM is updated automatically. Next-action tasks are created. A follow-up reminder is scheduled. 5. Three days later, if there’s been no response, the agent drafts a follow-up nudge. The account manager approves before send.
Same process. Same people. Much shorter cycle, much higher hit-rate, much fewer dropped balls.
What’s actually different
The thing to notice: the agent isn’t replacing the account manager. The account manager still owns the relationship. They still write the strategic parts. They still make the calls. What the agent removes is the administrative drag — the bits no human enjoys and no human does consistently.
That’s the pattern. Agentic process redesign isn’t about substituting AI for people. It’s about deleting the drag that’s currently between your people and the work they actually want to do.
What we don’t do
We don’t build chatbots. We don’t sell “AI strategy”. We don’t deliver decks. We don’t recommend a 12-month transformation programme when a six-week project would do.
We build working software that runs your business differently, on the MAF platform we’ve already built and tested. By the end of the engagement, something concrete has changed.
How to start
If this resonates, take the five-minute diagnostic — it’ll surface the processes in your business most likely to fit this pattern. Or read the full method for the engagement shape, pricing band, and what happens next.
Or just send us a message. We’d rather have a real conversation than send you another marketing email.