What is an AI workflow redesign workshop?

A working session where the people who run a workflow map it as it really happens, price where it loses time, and build and demo AI solutions for the worst parts, using their own tools.

Who should be in the room?

The team that does the work, the process owner, and some people who've never built anything with AI. Up to 15.

What do people learn?

Four skills they take to the next workflow: reading a workflow through the eyes of AI, screening use cases with FOCUS, stating how a decision really gets made, and writing up what they built so someone else can run it.

Do we pick the workflow first?

Yes, on a scoping call. The morning goes into the map rather than the agenda.

What can a team build in an afternoon?

A saved prompt, a project or template others reuse, a checklist, or a small agent. Ugly is fine. It has to run.

Do we need training first?

No. People bring their own work and their own tools.

Programs · Prove it

Applied AI Workshop

How do you get a team to redesign its own workflow with AI? You put the people who run the workflow in a room with their own AI tools, and nobody leaves until something they built is running. Six hours, one team, one real workflow, prototypes demoed live.

Your people are faster at their tasks. End to end, the work takes what it took last year. The day goes after that gap, and it teaches four skills they keep using after this workflow has been rebuilt twice.

AI keeps moving. A team that never learned to redesign its own work waits for someone else at every model release.

Find a delivery partner

6 hours · one team, one workflow · up to 15 people

What you've built

Your team uses AI daily and tasks are quicker. People have prompts they'd defend. Somebody built something useful and told two colleagues.

The thing most likely holding you back

Task speed up, delivery time flat. The work sits in the same queues and gets retyped at the same handoffs, and nobody has taken the workflow apart to see where the day goes.

The four skills your team keeps

Fix one workflow and you call someone again next quarter. Each exercise here teaches a skill your people use on the next one without us.

Reading a workflow through the eyes of AI.

Taking work apart into tasks, and spotting the ones that are high volume, rule bound and document heavy.
Screening for the highest-value use case.

FOCUS: organizational Fit, Operational pull, Capability in the room, Underlying data, Success metrics. Without it, teams pick what's nearest and get what Melissa calls random acts of AI.

Melissa · 62 sec

The FOCUS Framework for Choosing AI Use Cases

Short from YouTube, plays on click
Stating how a decision really gets made.

Teams decide by whoever's in the room with the best case. AI can't work from that. Writing the real criteria down is what makes a workflow automatable, and almost everyone skips it.
Handing the pattern over.

Writing what you built so a colleague runs it without asking you.

By the end of the day, each person can take a workflow they own, find where it loses time and fidelity, put a number on that loss they can defend to finance, and build and hand over a working AI solution for it.

The six hours

Diagnosis in the morning, building in the afternoon. Your people are in your own AI tools the whole time. Less than a fifth of the day is anyone presenting.

Be messy, the first output is a draft. Bring the real thing. Demo or die.

Morning

The paradox. Everyone answers "why aren't we faster yet?" themselves, then asks their AI. What the AI missed is usually the thing only your team knows.
The messy map. The version that happens when three people are out and the system is slow. Not the process document.
Friction and fidelity loss. Friction is wait time. Fidelity loss is rework. Test: did somebody have to do it twice?
The math of waste. Ten minutes, five people, ten times a week, and it's 400 hours a year. That number gets you a sponsor.
The FOCUS screen. Score every candidate on all five letters.
Three targets, three names. No name on a target means it's a wish, and the room says so before lunch.

Afternoon

The marketplace. Three minutes of posting, no debate. It surfaces what's already in the room, including the people quietly building agents on the side.
Pitch, vote, team up. Sixty seconds a pitch. Each team needs someone who does the work daily and someone who's never built with AI.
Sixty minutes on the keyboard. Two people on call: someone from IT, and someone who knows what's allowed. Answers in minutes instead of next week.
The demo. Five minutes a team, live, on real input. The pain, the build, the new number.
Codify it. What it does, how to run it, when not to, owner by name, where it lives, the thirty-day measure.
Who spreads it. The people who helped their neighbor instead of finishing their own build. That's the AI Lead role, and it needs five protected hours a week.
The scorecard, out loud. Friction points found, hours priced, solutions demoed, patterns written, AI Leads named.

What your team leaves with

One workflow mapped as it really runs
An annual hours number on the five worst clusters
Three targets, each with an owner
Working prototypes, demoed on real input
A written pattern per build
Names of the people who want the AI Lead role
The board itself, open and editable

What another quarter costs

Nothing breaks if you wait, which is what makes this easy to postpone. The arithmetic:

The workflow keeps leaking whatever the room would have priced that morning. Usually hundreds of hours a year, in one process.
People keep building anyway. Without the screen, a team picks the ambitious thing, gets sixty or eighty percent there, and stops. The political capital and the budget are gone, and there's nothing running. The next request is harder to fund than the first.
The models move again in a few months. Teams with these skills take the next pass themselves. Teams without them wait for a budget cycle.

Why teams bring in a partner

Plenty of teams could run a day like this. Here's what usually happens when they do.

The room maps the process document instead of the Tuesday version, because that one's safe to say out loud.
The senior person answers everything, and the day becomes their opinion.
Nobody prices a delay, because guessing badly in front of your boss is uncomfortable.
The afternoon ends with four prototypes and no owners, and nobody wants to be the one to say so.

Then there's the curriculum. Every exercise teaches a named skill, in an order that builds, with an outcome you can check before people leave. That's the difference between a good offsite and a capability your team still has next quarter.

How rare workflow redesign still is

Of organizations using gen AI, 21% have fundamentally redesigned at least some workflows. Of 25 attributes McKinsey tested, workflow redesign had the biggest effect on seeing EBIT impact. Source: McKinsey, The State of AI, March 2025 (1,491 respondents, 101 nations)

Only 42% of employees say they know how to identify where AI could improve their own work. Source: Gartner, October 2025 (2,986 employees)

Start here

Redesign one workflow with the people who run it

Right now

Your team is faster at its tasks, the end-to-end time hasn't moved, and nobody has priced the delay.

With this move

One workflow priced, rebuilt and running, and the people who did it can do it again when the models change.

Logistics

$12,000

for up to 15 people, or $1,000 per person with a minimum of 6. You pay whichever is less.

Find a delivery partner

Format: Six hours, one team and one real workflow
Delivery: Virtual or in person. In-person delivery adds travel and expenses, billed at cost.
Group: Up to 15 people. Two teams that share a workflow can come together.
Prerequisites: None
Tools: Bring your own. People work in the AI tools you already license, with someone from IT on call.
Before the day: A scoping call picks the workflow, its trigger, its endpoint and its process owner, and names the two people on call.
Terms: The proposal sets final scope and terms.

Next step: the AI Leads Accelerator

The people who spent the build helping their neighbor are your first AI leads.

Questions about the Applied AI Workshop

What is an AI workflow redesign workshop?: A working session where the people who run a workflow map it as it really happens, price where it loses time, and build and demo AI solutions for the worst parts, using their own tools.
Who should be in the room?: The team that does the work, the process owner, and some people who've never built anything with AI. Up to 15.
What do people learn?: Four skills they take to the next workflow: reading a workflow through the eyes of AI, screening use cases with FOCUS, stating how a decision really gets made, and writing up what they built so someone else can run it.
Do we pick the workflow first?: Yes, on a scoping call. The morning goes into the map rather than the agenda.
What can a team build in an afternoon?: A saved prompt, a project or template others reuse, a checklist, or a small agent. Ugly is fine. It has to run.
Do we need training first?: No. People bring their own work and their own tools.