Define what qualified means, and build the system that enforces it, in one day.
A full day, live and private. Marketing and sales agree what counts as a qualified lead, and the team builds the first working piece of the system that applies it, on a real account, before anyone leaves the room.
"Grace did a great job providing an overall framework for how to set goals for AI projects, define workflows, and establish governance. She combined this with live demos on building agents for surfacing prospects that match your ICP."
Craig PremoDirector of Marketing, RosmanSearch"Grace is probably one of the best presenters I’ve learned from. I attend a lot of sessions just as background, and Grace’s sessions always draw me in."
Logan Mathew JungAnalytics Manager, Meta Business Services, Meta"Grace is an inspiring leader with a real gift for making complex ideas feel clear and approachable. Whether you are just starting out or looking to take things to the next level, her guidance and insights are incredibly valuable."
Toey ChitmanasakRegional CRM and Data Manager, Michelin"Hands-on and interactive. Good at showing that it is possible to build a stable context structure and then interesting content from small everyday snippets like transcripts."
Maibritt Harritz GrosenChief Operations Officer, Clearwater"Really useful course, whether you are new to AI agents or already have some experience. It meets beginners where they are and still gives intermediate users real depth. You walk out with working agents you can refine and start using right away."
Geraldina Scarascia OlsonFractional Marketing LeaderNot two four-hour sessions. One day, and none of it is a deck.
A written qualification definition
What counts as fit and what counts as intent, agreed by marketing and sales in the room, from your own material: positioning, ICP, proof, and the objections that actually come up.
One part of the system running
Not a prototype and not a slide. The definition applied to a real segment of leads already in your CRM, gated by a named human, and producing a number before the day ends.
The rest of the path architected
Scoring, routing, the sales handoff and the feedback loop, each designed with its input, its agent, its gate, its output, and its one number. The build order agreed in writing.
A scorecard and a 90 day plan
One number per part of the path, the three hats named on each, the weekly loop written into your own calendar, and the decisions about what you stop doing.
Nothing is practiced on a sample dataset. What your team builds in the room is running on Monday, on the accounts they were already working.
When sales says the leads are bad, the channel is almost never the problem.
If more than one person in the room owns a piece of the funnel, marketing or sales, four sentences will sound familiar. Team size is not the qualifier. Whether anyone has written down what qualified means is.
We get leads, and sales says they are not qualified.
The lead count goes up and the pipeline number does not follow it, because almost everyone gets waved through the same loose threshold.
Deals go quiet in the middle and we find out at quarter end.
Nobody watches the long middle. By the time the pipeline gets attention it is too late to act on it.
Marketing and sales have never agreed on what qualified actually means.
Each side scores the same lead differently, and neither definition has ever been written down in a form the other side can see.
Everyone uses AI, and none of it starts from the same definition.
Individual scoring and drafting from a blank context, every time. Two people score the same lead two different ways.
None of that is a discipline problem. A team cannot out-work a definition that lets almost everyone through.
Who this is for, and who it is not.
- The team already produces GTM output with AI, at volume, and nobody is fully sure it is on-brand.
- There is real demand already. The job is converting more of what exists, not creating demand from zero.
- Someone owns the pipeline number, a CMO, VP Marketing, or Head of Growth, with sales on the other end of the handoff.
- You want it handed over finished. We train people to build. If nobody on your side will touch it, there is no product here, and I would rather say so now than take the fee.
- Nobody owns the pipeline number. If marketing and sales cannot name who is accountable for pipeline created, there is nobody to agree a definition with, and the day will not stick.
- Nobody is named to own it afterward. Every team that has drifted after a workshop shares that one thing. We ask for the owner before you book rather than hoping one appears.
Turning a team down costs one fee. Taking the wrong one costs both sides a day and a story that travels, which is why the disqualifiers are stated here rather than discovered on the call.
And one thing that is not a disqualifier: not every gap is reachable by a pipeline system. In any funnel a real share of the gap sits behind pricing, product, and market timing decisions that no system touches. We name those, with owners, and hold the system accountable for everything else. Saying it in the first session is what makes the scorecard credible at day 90.
Grace Man, founder
Nine layers, one operating model.
Nine separate AI experiments fail. One system built on a single definition of qualified compounds, because every layer reads the same approved truth and reports one pipeline number. Open any of them to see the workflow.
The context layer
The floor everything else stands on. Positioning, ICP, segments, proof, objections, and the claims the company is allowed to make, structured so a machine can read them rather than written as prose for a human.
The two at the bottom are built first and it is not negotiable. A team that starts at content is rewriting its output six weeks later. A team that starts at the floor is not, because everything above it writes from what the floor holds.
These are not five things to learn. They are one shape, applied five times.
Hover any step to see what it looks like on the day.
- Input
Approved truth
The knowledge layer, plus your own systems. Never a blank context.
On the dayOne priority account, its history, and the proof your team has already approved.
- Agent
The volume work
Inside boundaries you set. It drafts and scores, it does not decide.
On the dayFive roles mapped, a narrative and a first touch drafted for each, in minutes.
- Gate
A named human
Reviews before it acts. Operator, Reviewer, Approver, named people.
On the dayRed pen on the pack. What is right, what is wrong, what is missing.
- Output
Work, and one number
Something that ships, and the single measure it reports.
On the dayThreads out per seat, and stakeholders engaged per account on the scorecard.
Learn the shape once and every future layer is an afternoon, not a project. That is why this is one day rather than eight weeks.
Four blocks, in build order, across one day.
The floor first, then demand, then pipeline, then the loop that keeps it honest. Nothing is covered because it is interesting, and nothing above the floor is built before the floor exists.
Built first, in the room, from the team’s own material and CRM data. Positioning, ICP, segments, proof and objections, plus the fit-and-intent definition marketing and sales agree on together. Then the evidence layer underneath: calls, campaigns and outcomes, indexed so the definition has data behind it rather than opinion. Nothing above this holds without it, which is why it is not homework.
Research that never gets done because nobody has the afternoon, run against the floor so it arrives framed. Then one approved source through the full channel set, timed in the room, which becomes the baseline. Then answer engine and generative engine optimisation, because buyers increasingly ask a model rather than a search box, and a model answers from what it can retrieve and attribute.
Leads scored against the definition rather than a form-fill count, and outbound as a system rather than a spare afternoon. A CRM record that writes itself as the work happens and flags what does not meet the bar. And the layer most teams skip: marketing and sales reading the same objection answers, so a lead does not get a different story at the handoff than it got on the landing page.
Monitor and analyse, which is what most weeks are missing: plan, execute, execute, and nothing from last month changes next month. One number per layer, and every sales rejection fed back into a sharper definition next quarter. Three hats named, with names rather than roles. Then the build order, written into the team’s own calendar before anyone leaves.
The running order inside each block flexes with the room. The blocks do not.
Three things, and you already have most of them.
We do not sell a stack and we do not ask you to move to one. Every team arrives on something different, and the architecture is the same either way. We build on what you already run.
Your CRM and databases
Pipeline state, account history, the lead list, and whatever else already holds the record. The capabilities read from your systems and write back to them. Nothing gets migrated, and nothing gets rebuilt in a tool of ours.
An agentic workflow builder
Whatever you use to put a model behind a schedule or a trigger, pull from your systems, and hold the result at a gate. Several do this well and they change every quarter, so we teach the shape rather than the buttons. Learn it once and the tool becomes a detail.
A named human on the team
The part nobody can buy. Every capability gets an Operator who runs it, a Reviewer who checks it, and an Approver who releases it, and those are people with names rather than roles in a policy document. Nothing auto-sends, in the room or afterward.
The gap was never access to the technology. It is the order of assembly, and who is accountable when the output is wrong.
Two people have to agree. Here is what each one gets.
This section is the deliverable. Forward this page to the other one and see whether they land on the same read without you explaining it.
CEO or VP Marketing
Is this a real fix, or another AI expense?
It is signed on your own authority. No vendor onboarding, no security questionnaire, no procurement cycle. The definition gets tested against real leads in the room, so you know before the invoice whether it moves the number.
VP Sales or CRO
Will this actually change what lands on my reps’ desks?
Sales writes the definition alongside marketing, not after. Every lead a rep sees has already passed a bar sales helped set, and every rejection feeds back into what gets sent next.
Marketing operations
Does this mean rebuilding our CRM or MAP?
No. The definition and the scoring rules plug into the systems you already run. Nothing gets migrated, and no new platform needs approving.
Grace Man
Fifteen years building and running growth and marketing, at Microsoft, Pfizer, Canon, and Hyundai, influencing more than $2B in revenue: ICP research and buyer interviews, full-funnel and conversion work, and the qualification definitions that decide what a lead actually is. Over 3,000 professionals trained in agentic AI for growth and marketing.
Every session is taught by me. I have run this exact diagnosis twice: a lead threshold that let almost everyone through, promoted as qualified, and a pipeline number that never matched the lead count. Both times the fix was the definition, not the channel. That is what this workshop builds, and it is the reason the first thing built is the definition rather than a dashboard.
Twelve things, and none of them stay behind when we leave.
- A governed knowledge layer, built in the room from your own material
- A working layer running, on a real account, before the day ends
- Four more architected, each with its input, agent, gate, output, and number
- A segment of leads re-scored, against the new definition, so you see how many were mis-qualified before today
- A silence threshold per stage, set by you, not a vendor default
- A timed content baseline, source to live, measured on a stopwatch in the room
- Your backlog sized, never contacted and never connected, counted
- Your leads tiered, pass, nurture, disqualify, on fit and intent rather than form fills
- A scorecard, one number per capability, with an owner on each
- Three hats named, Operator, Reviewer, Approver, on every capability
- The weekly loop, written into your own calendar before anyone leaves
- A 90 day build order, agreed in writing, weighted to this quarter
All of it sits in your own workspace and stays there. Every prompt, context file, and workflow is yours on the day, not licensed from us, and there is no platform of ours to stay subscribed to in order to keep using what your team built.
One number, and everything above is inside it.
No seat tiers, no per-person pricing, no upsell on the day. One fee for the workshop.
For context, a typical marketing automation and lead-routing stack runs several thousand dollars a month before anyone has agreed what the leads it scores should actually count as. This is the day that decides that first.
Where this sits, and what comes after it.
Shown in full rather than revealed later. Some teams read this and buy the monthly instead, which is a better outcome for both of us than finding out in month four.
Team Training
$4,500 · up to 5 seatsOne full day. The qualification definition written, the nine layers architected, the first ones running, the 90 day build order agreed in writing.
Boardroom
$4,950 a month, rolling. Pause anytimeOne part of the pipeline system built, launched, iterated and refined every month, with the team that owns the number. Unlimited requests queued, one active build at a time.
Running a practice of your own rather than a team? The Agentic AI Practitioner Program is the same system scoped to one practice, at $1,100 a month.
The two quotes above, in full.
“Grace did a great job providing an overall framework for how to set goals for AI projects, define workflows, and establish governance. She combined this with live demos on building agents for surfacing prospects that match your ICP, establishing an outbound sequence of LinkedIn or email communications, and developing content for various platforms.”Craig Premo, Director of Marketing, RosmanSearch
“Grace is an inspiring leader with a real gift for making complex ideas feel clear and approachable. Whether you are just starting out or looking to take things to the next level, her guidance and insights are incredibly valuable. I would highly recommend her to anyone looking to deepen their understanding of AI strategy or leadership.”Toey Chitmanasak, Regional CRM and Data Manager, Michelin
What you are actually thinking
Because the nine layers are not nine things to learn. They are one shape applied nine times: input, agent, gate, output, one number. Once a team has built one end to end and can name the shape, the next is an afternoon rather than a project. A longer course teaches the same shape eight more times and charges for the repetition.
A license is a tool. This is the operating model that decides what to build for your pipeline, which approved truth it reads first, who reviews it before it acts, and which number it reports. It runs on whatever models and tools you already pay for. Tools change every quarter. The model does not.
The honest question back is: what is still running from it? A generic session teaches the tool on someone else’s examples. This builds one capability live on a real account of yours, hands it to a named owner, and attaches one number to it before anyone leaves the room.
No. Everything is directed in plain language, and where code is involved the AI writes it and we show your team how to direct it. The right five people are the ones who own the pipeline number, usually a VP of Marketing or Sales plus the people who run demand, marketing operations, and the SDR or BDR team.
Probably nothing new. You need three things: the systems that already hold your pipeline and account history, something that can put a model behind a schedule or a trigger and hold the result at a gate, and a named human to approve. Most teams already have the first and some version of the second. We deliberately do not sell or mandate a stack, because the ones that are best today change every quarter and the architecture does not. If you are missing the middle piece we will help you pick one on the scoping call, sized to what you will actually keep using.
Yes, and we would rather say so than pretend. The difference is whose truth it starts from and who decides. Every capability reads your approved knowledge layer, and a named human approves before anything moves. Nothing auto-sends, in the room or afterward.
Then the capability is in the system and the library, not in the person. That is the whole reason the workshop trains a triad rather than a single owner, and why everything built sits in your own workspace.
Additional seats are quoted on the call, and the honest answer is that five is not an arbitrary cap. The second half of each session is hands on, on one real account. Past about six people, half the room watches instead of builds, and watching is what makes a workshop fade.
Most teams build the remaining capabilities themselves off the 90 day plan, which is the outcome this is designed for. If you want the build sessions, the architecture review, and the roadmap to continue monthly, that is Boardroom, and it is scoped on a call rather than sold from a page.
Email grace@aistrategyleague.com and I will send one. The card path exists because it is faster for the teams that want to start this month.
Scoring is the enforcement, not the decision. What counts as qualified is a human call, made by marketing and sales together, in the room, before anything gets built. The system applies that call at every lead, and every rejection reason feeds back to sharpen it.
Define what qualified means, and build the system that enforces it.
One full day, up to five seats, on your own accounts and your own CRM. Book it and I will email you within one business day to agree the dates and the baseline number we start from.