The 70/30 boundary applied to NorthStar, Tech Vector, and Defense BD.
The hardest move in productization is also the simplest one: drawing a line around what the standard package contains, and refusing to put anything else in it. What came up on the call was exactly this realization: “the insight from your product, you have to get it in the hands of decision makers. And the problem is you’re putting a product that’s complex because it does a lot. It does a lot, but it’s super powerful too, and the problem is for folks who will make decisions about this, they don’t have time to do this, right? They just want the answers.”
That observation is the entire productization thesis. Decision-makers do not want to operate a complex tool. They want the answer the tool produces. The productized form, then, is not the tool with reduced features. It is the tool plus a defined deliverable plus a defined scope plus a defined cadence. The buyer gets a quarterly universe pack on a named technology, delivered as a structured PDF, with a point-of-contact arrangement for the eight follow-up questions they will actually have. The tool stays inside NorthAI. The answer goes to the buyer.
Book 1 chapter 11 names a rule that applies cleanly here: 70% of each productized SKU is the standardized core that ships identically to every buyer. 30% is configurable — the named technology area, the time horizon, the specific accounts and agencies in scope, the format adaptations a particular buyer needs. Anything outside the 30% configurable surface is custom work that gets priced separately under a different CLIN. That boundary is what protects the gross-margin math. What came up on the call: “60 to 80% gross profit, which we never had on services.” That number is achievable only if the 70/30 boundary holds in delivery.
Tech Vector is the cleanest Level 2 candidate. The standardized core (70%) is the universe-construction workflow against the 200M-document corpus and the structured output template. The configurable surface (30%) is the technology area, the depth of analysis, the time horizon, and the named-account focus. Excluded: anything that requires live ingestion of a buyer’s internal data — that is a separate engagement.
NorthStar is harder to productize without losing the conversational character buyers like. The standardized core would be a structured Q&A session against a defined technology domain, with a transcribed and indexed output. The configurable surface is the domain selection and the depth of follow-up. The risk is that conversational tools resist scope-fencing — the next question is always a new question. NorthStar ships as a Level 2 SKU second, after Tech Vector has set the precedent for what the cadence looks like.
Defense BD is the longest path to Level 2 because the value proposition is closer to operational integration than to deliverable. Buyers want a live dashboard against their planning cycle, not a quarterly artifact. The Level 2 form here is probably a quarterly “budget posture” brief that highlights movement on the line items the buyer cares about — a slice of the dashboard’s output, delivered as a structured memo. That ships third, once the first two SKUs are in market and the cadence is proven.
Boundaries only work if you defend them. Make the cost visible. It protects everyone.
Scoping the engagement — figuring out what the buyer actually needs from the SKU — stays human. Senior-buyer conversation about the implications of the universe pack stays human. Edge-case interpretation when the data surfaces something genuinely ambiguous stays human. The clean Builder role and the clean Operator role both need to stay protected from being eaten by services-callback urgency — that is the structural job HARBOR plays as Steward during the engagement window.
What becomes machine: the universe construction itself (Tech Vector already does this at speed), the structured output template generation, the cross-reference checks against the 200M-document corpus, the indexed Q&A archive for follow-up. The amount of human time per dollar of revenue drops by something like 60% when this boundary holds, which is what creates the gross-margin uplift.