Sharpen your product decisions.
Engineering can prototype anything now. The question is whether you are prototyping the right thing.
Your product leaders have always made the call on a slice of the signal: the sales calls, the tickets, the interviews nobody has time to read. AI is the first technology that lifts that constraint.
Ten working days. One to four of your product leaders, one hour a day, 1:1. Each leaves with a working system on your own data, with the skill and confidence to keep extending it.
Surface the signal. Prototype the bet. Make the call.
The bet behind every epicFor twenty years, the product job has been to make the call on incomplete evidence. The data was always there: sales calls, customer interviews, support tickets, competitive moves. No one ever had time to read all of it, so the call got made on a slice.
AI doesn't decide. It reads the slice you never had time for and tells you what's in it.
Your PMs keep the call. They just stop making it on the loudest input.
Everyone's using AI to build faster.
Almost no one's using it to decide better.
What actually changed
AI is a tool, not a solution. The problem, too much signal to act on, always existed. The tool unlocks two things, and most teams only talk about the first.
See what you couldn't see.
A thousand sales calls nobody has read. Support tickets tagged by nobody. Interview notes in forty docs. The data was always there; now it can be distilled into a decision, with every number traced to a source your team owns.
Build what you couldn't build.
The analysis capability that used to need a data team and a consulting budget is now in reach of the person who has the problem. Your team builds it themselves, in two weeks, and owns it after I leave.
One honest caveat, stated up front: running AI against large datasets has real cost. Anyone who tells you otherwise is selling something else.
Same product leaders. Sharper calls in every meeting.
What changesA director walks into a prioritization meeting with three feature requests, a gut read on which one matters, and an hour of meeting time. The outcome depends on who argues hardest, or longest.
The same director walks in with patterns pulled from fifty sales-call transcripts, customer signals tagged by theme, and the strongest counter-argument to their own recommendation already on paper. The synthesis that never got done now happens before the meeting starts. The outcome is the right call, not the loudest call.
The director still does the deciding. The system carries the synthesis weight nobody has time to carry alone, and it builds the strategic reasoning muscle, not just the evidence pipeline.
Two weeks. Two tracks. One capstone.
How it worksEvery participant runs two parallel work-streams through the engagement. Both are visible to the sponsor in the weekly note and the Day 10 review.
Your team's existing work, faster.
Epics, PRDs, interview synthesis, prioritization memos. We feed the system your team's own past artifacts, good and poor, and it learns your voice and quality bar. The result is a skill that produces drafts in your house style, on demand. Time savings measured, not estimated.
A question that moves the needle, answered.
At scoping, you say where the evidence is thinnest. Each builder picks a question in that space and answers it by Day 10, on your data. A call made faster, a cost surfaced, a risk caught before it became churn. Not a demo. An answer.
A 30-minute sponsor call. You bring the decisions you wish had evidence behind them; we map the space the capstones will live in. Tooling confirmed.
Cowork, memory, context, and the participant's own stack. First live pull on their own data happens Day 1.
One hour a day, 1:1, plus 30 to 45 minutes of homework. The homework is their real work, not exercises. AI operates as a thought partner that pressure-tests their thinking, not just a faster typist.
Your builders present to you: the question answered on your data, the daily-work skill in action, and what they can now do without me.
By Day 10, each participant has a working tool that answers a real question on your own data, plus the skill to keep using it. Taking a build to a production system the whole team runs is an optional next step, scoped separately.
One requirement: access to Anthropic's Claude Cowork for each participant, ideally on your Enterprise plan. Confirmed at scoping. Nothing else to install, nothing for you to learn.
Sponsors fund it. Product leaders build it.
Who it's forThe sponsor: CPO or VP Product.
(CTO with product reporting in works too.) You are not going to install anything, and you should not. Your part is three touches: a 30-minute scoping call, a short weekly note on what your builders are producing, and the Day 10 review where they present to you. You also get a private read on who picked up what, so you know where the capability actually landed.
The builder: Director, Group PM, or Senior PM with strategic scope.
You know AI matters. You have not yet built anything with it that you use on Tuesday. This is not Slack summaries and calendar tidying. It is the work that actually consumes you: your epics, your strategy memos, your customer-interview synthesis, produced in your team's voice at your quality bar, plus a strategic build you own.
The capacity to answer questions that move the needle. On your data, every number traced to a source your team owns.
Their existing work, faster. Automated where safe, AI-assisted where judgment stays human. Built on your context, tuned to your quality bar.
A thought partner for the rest. An assistant that knows your strategy, your customers, and your history, for everything else they weigh each week.
A team that fishes on its own. The capability stays when I leave. No subscription. No dependency on me.
Not another license.
The objection, answeredYour tools have AI features. Your people may have licenses. Some of them read the newsletters and take the courses, and they should; the thinking in the PM community right now is excellent. None of that becomes a capability on its own.
Handing out licenses and hoping is rolling out Salesforce with no admin. What is missing is the system: your strategy, your artifacts, your quality bar, and your evidence in one place, set up right the first time, in an instance that remembers.
Courses teach the craft in general. This engagement installs it on your data, with your team, and the context compounds instead of resetting every session. That system is the engagement.
The first engagement, in the client's words.
ProofThe first MaxGradient engagement ran ten days with the CEO of a national DTC e-commerce brand. She started AI-skeptical. She finished with a working profit pipeline her team checks daily, and the fluency to stand up new builds on her own.
“Craig didn’t just build our profitability dashboard; he taught me to run it, and to use Claude as a true force multiplier in my business.”
“I worked with Craig Calder of MaxGradient to build a marketing focused profitability dashboard for Paw.com, designed to expose the marketing KPIs that truly reflect actual contribution margin and profitability. He brings a rare ability to bridge operational execution with strategic thinking, focusing on the levers that actually drive performance. Just as valuable, Craig worked closely with me to teach both the system and how to use Claude AI effectively as part of my workflow. If you’re focused on business transparency and want to use AI as a true force multiplier in your business, I’d highly recommend Craig.”
Honest note: this was an owner-operator engagement. It proves the method and the delivery model. The product-leader cohort results will be published as cohorts complete.
Running the whole company? There's a 1:1 for that.
A second shapeSome leaders are the sponsor and the builder in one seat: the founder, the CEO, the owner-operator whose marketing, ops, and product signal all land on the same desk. The first MaxGradient engagement was exactly this shape.
Same ten days, same two tracks, fully 1:1. You bring the question that keeps resurfacing in your Monday meeting; you leave with the system that answers it and the skill to build the next one.
Craig Calder, Founder.
About
Twenty years of senior product management leadership, from startups to eBay and New Relic. Hands-on with the production AI stack (LLMs, agentic systems, knowledge graphs), because you cannot teach a team to build what you have not built yourself. I build these systems myself, every week, and the engagement carries what is working right now, not what worked in a slide deck last year.
I publish regularly on how AI is reshaping product leadership: Medium · LinkedIn
The hardest part of AI adoption is not the tools. It is the confidence to use them in front of your peers, your CEO, and your board. The engagement is designed for that.
About an engagement.
FAQWhat does an engagement actually look like?
Ten working days with one to four of your product leaders. One hour a day, 1:1 with each participant, plus 30 to 45 minutes of homework that is their real work, not exercises. The week before: a 30-minute sponsor scoping call and a cohort walkthrough. Day 10 closes with the capstone review, presented by your team, to you.
What tools do participants need?
Claude Cowork on each participant's machine, ideally on your Enterprise plan (Team works; confirmed at scoping). Neither plan trains on your inputs or outputs. That is the one requirement. I build with Claude Code as well; participants who want to go deeper can, but it is never required.
Why Claude?
The engagement runs on Anthropic's Claude platform because it is currently the strongest suite of tools built for business work, and the market agrees: Anthropic leads OpenAI in U.S. business adoption (41% vs. 39.5% of businesses as of June 2026), per the Ramp AI Index, which tracks actual corporate spend.
What data do we need?
Qualitative signal your team already collects: sales-call transcripts, customer interviews, support tickets, competitive research, strategy and OKR docs. Day 1 anchors on whichever source is closest at hand. Sales transcripts are a common starting point because almost every team has them and almost no team mines them, but they are one example, not the offering.
What happens after two weeks?
Your team owns the system. Each participant runs it, maintains their own context layer, and extends it as their role evolves. I hand over a maintenance guide and a self-management playbook. Taking a capstone build to a production system the whole team runs is a separate, optional phase. An ongoing retainer is available but never required; the point is a team that fishes on its own.
What does this cost?
Project-priced, scaled by cohort size. These are founding-cohort rates for the first engagements:
- 1 participant (executive 1:1): $10K. Founding-cohort rate; standard $15K.
- 2 participants (intensive): $15K.
- 3 to 4 participants (full cohort): $24K to $30K.
Four is the maximum, and that is structural: the value is the daily 1:1, and it does not fan out further without quality loss. What you are paying for: scoping, nine days of 1:1 implementation per participant, the Day 10 capstone, and working systems your team operates after I leave. Not a slide deck. Exact pricing settles on the scoping call.
Couldn't my team just do this themselves?
Some of the pieces, yes, and the good ones will try. What a license and a weekend cannot produce is the system: your org's strategy, artifacts, quality bar, and evidence assembled into an instance that remembers, set up right the first time. Rolling out licenses without that is Salesforce with no admin. Two weeks with someone who has built it before is the difference between a tool people poke at and a capability the team runs.
Bring the decision you wish had evidence behind it.
TalkA 30-minute scoping call. Brass tacks, not a sales pitch. For sponsors weighing whether to fund this for their team, and for senior product leaders ready to send it up the chain. Scoped proposal within the week.