It starts with the business decision.
Not a checklist of activities. rocket asks what the analysis should change, what evidence supports it, and what the next commitment would cost.
With your consent, I use Google Analytics to count visits and see which pages are useful. No advertising, no cross-site profile. Nothing is set until you choose, and you can change your mind any time from “Manage cookies” at the bottom of every page. Details
Remember your cookie choice (in local storage, 6 months), keep content unlocked after you confirm your email (rocketpm_unlocked, 1 year) and keep your Quick assessment answers in your browser. No tracking, never sent to anyone.
Google Analytics 4 measures visits and clicks on the assessment buttons, without advertising features. Cookies _ga and _ga_*, kept up to 13 months. Off until you turn it on.
A Product Marketing skill for Claude. Challenge the opportunity before building. Design how it wins and makes money. Bring it to market. Learn what actually worked, and whether the next feature is worth the investment.
Early beta. Human decisions stay human. No invented customer evidence.
rocket triage "Sales wants an invoicing module, half our lost deals mention it"
Restates the idea as a problem
Checks what traces exist
First guess: build, buy or partner
Recommendation: study.
It asks whoever brought the idea to confirm the problem as restated.
Ideas are everywhere. The CEO saw a competitor demo. Sales heard the same request three times. A customer wants a feature. Product has a technical idea that could work.
Then the usual machinery kicks in: roadmap, development, launch, announcement. What often gets lost is the question that should have come first: what would have to be true for this to be worth building?
rocket connects the decisions that too often live in different documents, functions and meetings: the market evidence, the customer value, the investment case, the offer, the route to market, and what the launch actually earned.
On all 12 disciplines. Assist is the default. Decision is there when you ask for it, and at every gate of a project. Escalation is offered, never imposed.
rocket price "should the reporting module be an add-on or sit in the top plan?"
The default when you ask for help outside a project. Useful work at once, with the context it has.
Here is a working recommendation now. Given the price policy at stake, I also recommend a formal review before committing.
rocket price "should the reporting module be an add-on or sit in the top plan?"
When you ask for it, and at every gate of a project. "Challenge my business case, the board meets on Tuesday."
A decision memo: the recommendation, its risks, its confidence and its conditions. Ready only when the essential evidence is there.
The same request, at each level. What rocket does is described, not quoted: the example is invented, and it holds no figure because the skill never supplies one.
The whole process, 8 stages in 5 phases, or one of the 12 disciplines on the decision in front of you.
Take a feature, a new product, an offer change, a segment entry or a repositioning from its first hypothesis to a grounded business review. Each stage asks one question and ends on a decision taken by a named person. The skill recommends; it never decides.
Phase I · Find the opportunity
rocket triageDoes this idea deserve a study?
Ends on: A verdict: study, merge, park or drop
rocket problemIs there a problem worth sizing?
Ends on: A checkpoint
Phase II · Decide where to invest
rocket caseIs it worth pursuing, against doing nothing and the best other use of the same means?
Ends on: A gate: pursue, reshape, park or stop
rocket pathNothing new, build, buy, partner, or a mix?
Ends on: A gate: the path
Phase III · Design the business
rocket offerWhat do we sell, to whom, against what, at what price, through which route?
Ends on: A gate: the offer
Phase IV · Entry to growth
rocket readinessCan we bill, and is the field ready?
Ends on: A gate: go or no-go
rocket entryDoes the market see, understand and reach the offer?
Ends on: A first reading
Phase V · Learn and reinvest
rocket reviewWhat revenue or retention did it produce, and what do we change?
Ends on: A gate: scale, iterate, pivot, maintain or sunset
Not everything is a project. Need to rethink a pricing tier, understand why you lose, or test a reasoning before the meeting? Ask. rocket applies the discipline that fits, works on the topic alone, and does not push you through a whole project.
rocket price "should the reporting module be an add-on or sit in the top plan?"rocket compete "why do we lose to the spreadsheet?"rocket challenge "we should make the entry plan free to fight the free alternative"The examples are invented. What you get stands on its own: no project to open first.
rocket delivers before it asks, imposes no template, and pushes you to gather what a good decision needs.
Not a checklist of activities. rocket asks what the analysis should change, what evidence supports it, and what the next commitment would cost.
Revenue is not enough. Every case is sized against doing nothing and against the best other use of the same means. Build, buy or partner is asked early, then settled.
Subscriptions, transactions, marketplaces, one-time purchases and hybrid models do not share the same economic drivers. rocket does not force every company into a recurring-revenue template.
Inside a project, rocket writes the framing: the brief, the structure, the criteria. Your teams write the content. Outside a project, it drafts one complete asset at a time.
Every claim is a fact with its source, a hypothesis, or an opinion with its author. No invented figure, no invented quote. Costs and thresholds come from your team, never from the skill.
A profitable feature does not automatically deserve more development. A weak launch does not automatically justify killing the opportunity. The review ends on one of five outcomes: scale, iterate, pivot, maintain or sunset.
Two tools, one process, and a file that travels between them. A project can start on either side.
Launchpad
A web board that follows a project through the same stages and the same gates. No AI inside: it works on its own, and it shows where the project stands and who decides next.
Launchpad to rocket
Export the project from Launchpad as one file. rocket takes it back in, scores the evidence for the next decision and works the stage or the discipline with you.
rocket to Launchpad
One command writes the file that Launchpad imports. The decisions, the gates and the plan are back on the board, for the people who do not work in Claude.
The skill itself never connects to anything: you move the file. It brings up Launchpad only when it helps: a project followed over time, or shared.
What a product marketer knows how to do, in four pillars. Use each one on its own, or bring the right combination into a project stage.
marketcompetebuyerspositionmessagedistributepriceThe CEO wants to match a competitor’s new feature. What should we investigate first?
Build a credible business case for a paid add-on, including cannibalization and payback.
Compare build, buy and partnership approaches for this customer outcome.
Draft a complete positioning brief from these source notes.
We’ve shipped the feature. Adoption is weak. Is the issue value, experience, pricing or distribution?
Customers love the feature. Should we invest another quarter in enhancements or maintain it as-is?
1
Add rocket to Claude Code or to the Claude app.
2
A short question, your market context, an existing project, or source files.
3
rocket picks the stage or the discipline that fits, and the depth the decision needs.
4
You validate the evidence, approve the outputs, and take the investment and customer-facing decisions yourself.
who need to connect market work, offers, launch and growth to measurable business outcomes.
who want clearer investment theses, more defensible prioritization and better post-launch decisions.
who want to challenge product ideas before allocating capital.
who need the commercial narrative, the enablement and the customer-value assumptions to agree.
No. It is a skill for Claude: a method, commands, reference material and templates. Where Claude can write files, it keeps the project documents up to date. It does not replace your product management, analytics, billing or CRM tools.
No. It delivers first, with the context it has, and names the assumptions that would change its answer. It asks a question only when the work would mislead without it. In Decision it is more demanding, and it still gives a provisional assessment before asking for anything.
No. Start a project when the decision needs one. Otherwise run one discipline on one topic, or ask for a specific deliverable.
Yes, when you provide them or when the tools are connected in Claude. It has no access of its own to your company systems, and it keeps sourced facts apart from assumptions.
It structures the case and works from the figures your team provides. It never supplies a cost, an effort or a threshold itself: when one is missing, it says which, and who can give it. You stay responsible for the inputs, and Finance should review any decision that matters.
Yes, as a recommendation. At the review it can recommend to maintain an offer or to sunset it, and it treats a sunset like a launch: customers to migrate, commitments, who tells whom. The decision belongs to a named person, never to the skill.
Yes. The method covers other business models, but your evidence and your assumptions decide whether a recommendation fits.
No. rocket is an independent skill built for Claude by Mathieu Seguy.
rocket is in early beta. The terms of use will be published at launch.
The next feature does not need another opinion. It needs a better decision.
rocket is an independent skill built for Claude by Mathieu Seguy. It is not an Anthropic product.