Operations & capacity · 3:42 video
How do you redesign a workflow around AI?
Redesign a workflow around AI by mapping the real journey, identifying where work waits and ranking those delays by business value and effort to change. Specify what AI prepares, what a person checks and who owns the next action. Start with one manageable pilot, agree the measures and a stop rule, then decide whether to expand.
English captions available in the player. Watch on YouTube ↗ · Read the transcript ↓
Map the job before choosing an AI tool
Shaz describes mapping a typical job with the team from the first call through to invoice and renewal. The aim is to find where information or revenue gets stuck. Write down the actual handovers, including the informal chasing that is easy to miss in a process document. A useful map shows the trigger, the information needed, the owner and the next step.
Rank opportunities by value and effort
The review asks two questions: what could the change be worth to the business, and how hard would it be to make? The examples include quote follow-up, invoice checks, email sorting and renewals. Value estimates help set priorities, but remain estimates until tested. Choose one problem with enough volume to learn from and a clear owner who can assess the output.
Write the split between AI and people
The proposal uses AI for mechanical work and brings a person in where judgment is needed. Make that split explicit: which inputs can AI read, which output can it prepare, what must be reviewed, and what happens if information is missing? The video mentions quote chasing after 3, 7 and 14 days. That cadence is an example to adapt to your customers and communication rules.
Give the pilot a decision point
Agree the baseline, review date and conditions for continuing, changing or stopping before the pilot starts. The video includes a stop rule and phased work, rather than an open-ended commitment to automation. In a workshop, turn these decisions into a team playbook and pilot scorecard. Your business retains ownership of implementation and ongoing decisions.
Before you decide
Questions from business owners and operations teams
Where should we start if we have lots of AI ideas?
Map the end-to-end journey, then compare a small number of bottlenecks by value and effort. Pick one with a clear owner, usable information and a result you can measure.
Does this mean reducing headcount?
The stated intention in the video is to help the existing team handle mechanical work. Capacity is an outcome to measure; no headcount reduction or productivity result is established by this review.
What happens in the 30-minute AI review?
Bring one bottleneck. You and Shaz will explore where AI could help and whether a practical workshop is a suitable next step. It is an initial conversation, not a completed workflow redesign.
Video transcript
English caption transcript, reviewed for content and grouped by topic. Spoken wording is retained. Examples, estimates and earlier service references reflect the recording; the explanation above describes the current workshop offer.
Read the full transcript 3:42
0:00 A proposed quote follow-up workflow
So, I redesign, uh, workflows around AI. This was a HVAC company where I proposed a completely reimagined workflow. You can see quotes were going dead. And in the redesigned workflow, I proposed that AI would chase automatically on 3, 7, and 14 days. This is so that, obviously, they could recover, uh, quotes. Now, I didn't propose a new team or a new system. It was simply a redesigned workflow so that those quotes would not die and they were worth thousands of pounds each. And imagine they're getting hundreds of these every quarter. So, the
0:42 Map the journey and prioritize value against effort
question we answered was, firstly, how would AI actually help their business? This is before they invest, um, anything further into AI. So, I mapped out their typical job from call right the way through to invoice and renewal with their team. And you could see here the different places where, uh, revenue was getting stuck. So, the way that we thought about this was, um, map out complete journeys, but more importantly, figure out which specific part of those journeys we should fix first. Um, and the way that we did that was asking two questions. What's it worth to the business in recovered revenue? And how hard would it be to actually change? And this gave them an idea of how much they could recover before they went ahead and redesign, um, their workflows. So,
1:42 Invoice checks, email and renewals
for example, invoice checks. Before, lots and lots of, um, manual checks required. The proposition, AI reads that and can do the check for you. Coordinators then, obviously, have the um opportunity to to, I guess, outsource a lot of the mechanical, monotonous stuff. In email, the same thing. It lands, we could get AI to sort the emails for them. This is because a lot of queries come out of hours. Jobs and renewals, same thing. Um lots of these renewals were sitting waiting for responses, and we could get AI to automatically respond to those with a human stepping in when required. That way renewals would not slip. The
2:33 A phased proposal with a stop rule
proposition was a build which would take approximately 4 weeks, and you could see here starting with the quickest wins first. Um and, you know, of course, there's a stop rule as well. So, if it does not work, or it hasn't worked after we embed, we can completely uh overhaul the project.
2:55 Help the team with mechanical work
So, we didn't want to necessarily add more head count or reduce head count. Um the whole point of this redesign was to just get AI to assist them with a lot of the mechanical work that they were doing across those different parts of a typical journey. Um what we proved to them was that they didn't have to buy more tools, but these changes could essentially work with their current systems. All of that revenue was they were already sitting on this, and this is probably something that's similar with your business as well. You maybe have a similar use case or pain point. If you do, then just get in touch, and we can map something out.
