The CRAFT Room
A free, 60-minute working session on bringing AI into post-sales.
What's getting in the way of your AI build? Most post-sales teams have AI by now, and 6 months later very little has changed in their customer conversations. Pilots stall, the team doesn't trust the outputs, and the question of which work goes to AI and which stays with people never gets properly answered.
That's why I run The CRAFT Room, a free working session I hold on a regular basis. It's by application, for CS leaders who are building with AI right now, and you can apply below for an upcoming session at any time. I read every application, pick the problems that come up most, and work through them live with CRAFT, the method I use every day with my clients. Places in each session are limited, and I come back to every applicant personally.
CRAFT, in 5 steps
Why it works. Most AI rollouts stall because nobody decided where AI belongs, so the tools get pointed at whatever looks easiest and the team never comes to trust the result. CRAFT works because it makes that decision first and runs it like any other business project, with an owner, a baseline and an agreed picture of success, so each step earns the trust the next one needs.
CRAFT is the method I use to decide how much AI a post-sales team should use, where and when, and then to build it well enough that the team actually chooses to use it. Some work should go to AI, some should stay with people, and some should stop altogether. You can't apply AI to everything, so every step starts with why, and each decision is made on purpose rather than in reaction to the latest tool. It runs like a project, with an owner, a baseline and dates, and it sits next to ACE, the way I run customer success.
Capture the work
We start with conversations with the CRO, the CS leader and the people doing the work, because the picture has to be human before it's operational. Then we list every recurring task across onboarding, adoption, value realisation, renewal and expansion: who does it, how often, how long it takes, what it draws on, and above all why it exists and who still uses what it produces. A lot of post-sales work survives on habit, so those tasks stop here, before anyone teaches AI to do them faster.
Why it matters. AI has to be matched to the right work, and you can only do that once the work is written down. This picture of the week is also the baseline every later result is proven against.
Rank human and machine
Every task that's left goes into one of 3 lanes. Human, where trust is built or lost, like a difficult renewal conversation. Assisted, where AI prepares and a person decides, like a pre-call brief. Automated, where a person only spot-checks, like CRM updates and meeting notes. Three questions decide it: does the moment carry emotional weight for the customer, would they notice or care whether a person or a system did it, and would we lose information or trust if it were automated? A yes to any of them keeps a task out of the automated lane.
Why it matters. This is where AI goes in on purpose rather than everywhere at once, and where the team sees that the conversations they care about most stay theirs. We start where a lot of time comes back and the risk to the relationship is small.
Audit the inputs
AI works with the data it's given and can't see what was never captured. So for each use case we check four things: is the data correct, current, complete and consistent across the CRM, the CS platform and finance? We also look at what it's organised around, because customer milestones give AI real signals, while journey stages mostly describe our own process.
Why it matters. AI won't fix bad data; it only helps us find it faster. Where the inputs fall short, we fix them first, or the task stays with a person until they're fixed.
Fire-test the output
For each use case we write down what good looks like, in the words of the people who'll use it, and the output goes back until it meets that picture. It often takes a third attempt. The CSMs decide: we ask whether they'd trust it with 12 minutes before a call, and if they wouldn't, it isn't finished.
Why it matters. People trust what they've helped shape. That rejection loop is taste, the human skill at the centre of the method, and it's where adoption is won.
Track success
Before anything is built, we agree which level of the AI success pyramid we're aiming for: adoption in the first weeks, efficiency from month one, capacity around month six and revenue at month twelve. Then we measure adoption weekly, hours and capacity monthly, and retention and expansion quarterly, and we decide where the hours we give back will go.
Why it matters. A programme measured against everyone's hopes disappoints all of them. Agreeing the levels up front turns a vague "AI-ROI" into something a team and a board can steer by, and stops the freed hours disappearing into email, internal meetings or the headcount line.
What we'll do together
I'll show you how I'm running CRAFT with my clients right now, stage by stage, including what we sent back and why. Then we'll put the tasks you bring through it live, and you'll leave with your own task placed in a lane, the reason it belongs there, and the level of the AI success pyramid it should move.
Alongside the sessions, I regularly run workshops that explain CRAFT in depth, for leadership teams who want to put it into practice with their own people. Apply below or get in touch to ask about one.
Who gets a place
CS and post-sales leaders who are building with AI now, or about to, and will bring a real task.
Who it isn't for
Anyone not building yet, or looking for a tool demo. I write about CRAFT on Substack as it develops.
Apply for a place
7 questions, about 5 minutes. I read every application myself.
Thank you, it's in.
I'll read your application myself and email you to let you know whether there's a place for you, and the date if there is.
Let's talk
Questions before you apply, or interested in a CRAFT workshop for your leadership team? Tell me what you are working on and I will tell you honestly whether I can help.