Knowledge Manufacturing for L&D
Sravan Ankaraju
I build what L&D buys, and I measure it
Your content library is an expense. This is how you turn it into an asset.
Every course expires. Completion gets recorded, the artifact decays, and the next initiative starts near zero. That was survivable when production was expensive. Then generation became nearly free, and the problem got worse — more content, faster, on a foundation nobody had measured.
The line between an expense and an asset isn't cost or permanence. It's measurability.
I build these systems. My CMMC Level 2 substrate carries 110 competencies and 320 assessment objectives with ground truth behind every one — and 9 ontology triples, bad enough to render red on my own dashboard. Every session opens on a real screen with a real number, and at least one is bad. You can't learn to score a domain honestly from someone who only shows good days.
Five weeks, one domain you bring and own. Score it, classify the corpus, decompose it, write an assessment that measures judgment instead of recall, and finish with a one-page case for whoever controls your budget.
I won't claim a better-scored substrate improves learner outcomes. Nobody has shown that, including me. You'll leave able to instrument what you own — so in thirty days you have two measurements and a delta, not an opinion.
What you’ll learn
Stop buying content. Learn to score, structure, and instrument the knowledge your organization already owns — and prove it's an asset.
The ASSAY rubric, applied by hand in pre-work, then run live against a real substrate dashboard with its bad numbers showing.
Reading the shape, not the total: two domains can score 48 and need opposite fixes. You leave knowing which single axis to repair first.
Live access to the cubelet.ai catalog tools, so you can inspect a scored, production substrate from the inside rather than take my word.
Run the seven-archetype classifier by hand, then against the tool, then defend your archetype call in a 1:1 with me.
A source inventory across the six faces of a knowledge atom, plus the two-minute trace that tells you whether a claim is grounded or cited.
Licensing triage at scoping time — open, permissive, law, reference-only, unusable — so rights never surface as a surprise in month three.
The boundary test: a competency that is neither a topic nor a task, sized to a defensible judgment. Practiced three times before class.
Prerequisite ordering by dependency, not document order — and the assertion problem: why no source will hand you the relationships.
A live demo of the ontology layer rendered as a graph, so you see what you cannot yet build — and why every domain is weakest there.
The determination grain: a scenario with incomplete information, an attractive wrong answer, and a key that records reasoning not verdicts.
You sit a live scored simulation yourself and get a determination-grain result, rather than only reading about one.
A cohort attempt exchange: you are paired with another participant and score each other's scenario, which is what exposes a weak key.
Three tiers of measurement — production, capability, business — and the rule about which tier you are allowed to price on.
Baselining a real number you can re-measure in thirty days, plus decay and freshness ownership — the part content programs skip.
Cut-off grade, borrowed from mining: the line below which a domain is not worth manufacturing regardless of how well it scores.
The three inputs to the line: what the decision is worth, what production will cost, and your realistic recovery for that archetype.
A one-page internal case, drafted live and then defended against a hostile CFO role-play with five standard objections.
Learn directly from Sravan
Sravan Ankaraju
Divergence Academy CEO (4x Inc. 5000). Founder, GRID42. Author, Hackers & Heroes
Who this course is for
L&D leaders who can't defend their library
Years of content, and no answer to "how do you know this is any good?" You want an instrument.Architects and designers moving upstream
Faster production hit its ceiling. The work that matters now is the structure underneath it.Capability leads handed an AI mandate
A budget and a vague brief. This is the upstream work that decides whether the spend compounds.
Prerequisites
A bounded domain you own or can get access to
A compliance framework, certification curriculum, product knowledge, or role-specific onboarding. You need to reach the source material.
Roughly 4 hours a week, including the live session
90 minutes live, 45 of pre-work, 2 on the project. The pre-work isn't optional reading.
Enough seniority to name a budget holder
Week 5 ends with a one-page case addressed to a named person. You don't need budget authority, just to know whose it is.
What's included
Live sessions
Learn directly from Sravan Ankaraju in a real-time, interactive format.
Certificate of completion
Share your new skills with your employer or on LinkedIn
Private 1:1 with me, mid-course
A 25-minute working session between Week 2 and Week 3, with a fixed agenda: your archetype call, your licensing position, and one written instruction for the week ahead. Scheduled at the point where a wrong call is still cheap to fix.
Live access to production tooling
A cohort entitlement to the cubelet.ai catalog and simulation tools, unlocked one per week after you have done the same work by hand. You inspect a real scored substrate from the inside rather than take my word for it.
Real dashboards, including the bad numbers
Every live session opens on a current, unedited screen from a production system. At least one number on it is bad, and diagnosed. Nine domain substrates, shown as they actually are.
One domain, one finished artifact
You bring a bounded domain you already own in Week 1 and leave with a single substrate design package — scored, classified, decomposed, assessed and instrumented. Five increments, one document, revised as later weeks invalidate earlier calls.
Pre-work you run, not read
Each week opens with a timed procedure and a self-checkable done-condition: score a domain by hand, trace five claims, test a competency boundary, write a ground truth card. Roughly 30–45 minutes, and the tooling stays locked until it is done.
A paired attempt exchange
You are matched with another participant in Week 3 and you attempt each other's judgment scenarios in Week 4. A non-expert attempt is what exposes the assumptions your answer key was quietly relying on.
The ASSAY scorecard, yours to reuse
The five-axis rubric with band language, plus the cut-off calculation. Re-run it on any domain, in any organization, without a license or a login. The instrument is the takeaway.
Lifetime access
Recordings, worked examples and every procedure sheet, available whenever you need to run this on the next domain.
Maven Guarantee
Your purchase is backed by the Maven Guarantee.
Course syllabus
5 live sessions • 59 lessons • 5 projects
Week 1
Oct
20
The Asset Argument — Why content was never the asset
The knowledge management graveyard
Expense or asset: the measurement test
The two-minute trace and licensing triage
Week 2
Oct
27
Archetype and Sources — What your corpus can and cannot yield
The seven domain archetypes
Grounding is not citation
Book your 1:1 — this week
The boundary test
Free resource
The ASSAY Scorecard — find out if your content library is an asset
Schedule
Live sessions
1-2 hrs / week
One 90-minute session, Tuesdays at 12:00 PM CT. Every session opens on a live production dashboard and runs I do / we do / you do — I score a domain, we score one together, then you score your own in a breakout. Recorded, but the breakouts and the scoring arguments are the part that doesn't record well.
Tue, Oct 20
5:00 PM—6:30 PM (UTC)
Tue, Oct 27
5:00 PM—6:30 PM (UTC)
Tue, Nov 3
6:00 PM—7:30 PM (UTC)
Projects
1-2 hrs / week
One project per week, all five building a single substrate design package for a domain you bring and own. Each week adds a section and forces revision of the ones already written. Week 4 also involves attempting a paired participant's judgment scenario, which needs a 48-hour turnaround.
Async content
1 hr / week
A timed procedure before each session — score a domain by hand, trace five claims, test a competency boundary, write a ground truth card. Roughly 30–45 minutes, self-checkable, and the tooling stays locked until it's done. These are the manual pass that makes the live session land.
Frequently asked questions
Maven for Teams
Reimbursement
Get your company to pay
Everything L&D needs: email template, receipts, and certificate of completion.
Get reimbursedTeam discount
Learn with your teammates
Save 20%+ when 2 or more teammates enroll in the same cohort.
Save 20%+ with a teamPrivate cohort
Run a cohort for your org
A dedicated cohort with a custom schedule and curriculum, tailored to your team.
Book a private cohort$1,500
USD