For PE, Growth Investors & Boards
A defensibility lens for your AI portfolio.
Most AI diligence still leans on TAM, growth rate, and a vibes-based read of the founder. That's how you end up funding a wrapper. Supply Chain of Intelligence™ scores every AI product across 10 layers, compute, data, models, workflows, surfaces, memory, and tells you, in one page, whether value is accruing to the company or leaking to the platform above it.
By Anand Arivukkarasu, Ex-Meta (Instagram) Product Leader & AI Product Architect. Personal thinking, shared freely.
The diligence gap
JTBD finds demand. It doesn't prove defensibility.
A product can be desirable and still be erased, by the foundation model below it, the distribution layer above it, or the workflow giant beside it. Without a layer map, you cannot tell which is which.
L1 + L5 + L8, owns data, workflow, and memory. Compounds over time.
L5/L6, earns its keep, but exposed if the platform layer above absorbs the surface. Counter-move exists.
L7-only on someone else's L2. One platform release away from being absorbed, counter-move is to deepen into L1, L5, or L8.
Three reads
Three ways the framework reads on the investor side.
Pre-investment read
AI Defensibility Diligence
Before you wire the check, score the target across all 10 layers. The framework surfaces whether you're funding a moat (L1 data + L5 workflow + L8 memory) or a wrapper that a platform release note will erase. Run it on your own deals.
Portfolio read
AI Roadmap Audit
A one-page scorecard pattern per portfolio company. Audit / 100, layer-by-layer verdict, and a 90-day deepening plan. The framework surfaces which assets are compounding and which are about to be commoditized, use it on your own portfolio.
Board read
AI Strategy at the Board Level
The 10 layers, the structural laws, and where each portfolio bet sits on the map, the language a board needs to translate AI hype into moats, margins, and capex. The framework is public; bring it into your own board packs.
The PE playbook
Valuation protection & expansion, in stack language.
Private equity does not buy architecture. It buys structural insulation against margin compression, multiple expansion, churn reduction, NRR, gross-margin protection. The Supply Chain of Intelligence™ translates each of those into specific moves on the stack. Any L7-heavy portfolio company (orchestration, workflow, dashboard, "agent" wrapper) faces the same threat: Law I, intelligence commoditizes downward, and the native platform below compresses the surface above. The defense is to climb down the stack into layers the platform cannot absorb.
The core threat
L7 / L6 compression.
Revenue built on orchestration and surface-level automation sits one platform release away from being absorbed. Foundation models below (L2Models) and native ecosystems above (L3Gates, L7Surface) are building intelligent agents that talk directly to APIs. If a buyer can route around the interface, the interface is a feature, not a moat, and the counter-move is to add a deeper layer (L1, L5, or L8) before multiples reprice.
The three structural moves
Move the product from a workflow interface to an un-bypassable reasoning node. Three defensible pillars, each tied to a PE-grade outcome.
Move 1 · Weaponize Access
Turn the MCP / API layer into a tollbooth.
Action
Position the company's MCP server or access layer as the only compliance, guardrail, and governance pipe for enterprise AI agents acting in your domain. Internal AI buyers do not connect to the upstream platform directly, they route through your layer to enforce budget pacing, brand safety, policy, and cross-channel rules.
PE Value
Churn collapses toward zero. Turning off the product means breaking the enterprise's internal AI architecture. Critical infrastructure, not a dashboard.
Move 2 · Monetize Decisions
Package historical telemetry as routing playbooks.
Action
Stop selling only software seats. Package years of proprietary transaction, attribution, and outcome data into fine-tuned decision frameworks at the L5 layer. When an external agent queries the domain, charge a premium to inject your closed-loop playbooks into its context window.
PE Value
High-margin, usage-based data revenue that scales independently of seat licenses. NRR expands without sales headcount.
Move 3 · Compound Memory
Build institutional memory the platform cannot replicate.
Action
Every automated decision, every override, every seasonal pivot the system executes feeds a compounding graph of why. Context drift, the largest enterprise AI failure mode, gets solved on your side of the wall.
PE Value
A data moat measured in years, not features. A cheaper UI or a faster orchestrator cannot copy compounded memory. Multiple expansion at exit.
How to frame it in the room
Present it as a valuation-protection audit, not an engineering overhaul.
- 1. Map the vulnerability. Show how much of current EBITDA sits on execution layers (L6 / L7) exposed to native-platform compression.
- 2. Name the structural fix. The three moves above are a product blueprint to defend NRR and gross margin, not a re-platforming project.
- 3. Tie it to the exit. Each move maps to a PE lever: churn, NRR, gross margin, data revenue, multiple expansion. The language travels from product to IC memo without translation.
What the framework surfaces
What you can pull out of the public framework.
Layer-by-layer scorecard
All 10 layers · 50 sublayers · score / 100 with verdict band. Run it on the audit page.
Defensibility verdict
Fortress / Workflow / Wrapper, with the structural reason cited.
Platform compression risk
Which layers above and below are about to absorb a product.
90-day deepening plan
The two or three layer moves that meaningfully change the verdict.
Comparable mapping
How a target sits next to the 24 worked case studies in the public corpus.
Board-ready one-pager
Run it yourself, drop it into your own IC memo or board pack.
Voices on the framework
Product leaders, founders, and investors using the 10-layer map.
Reactions from workshops, 1:1 reviews, and LinkedIn exchanges with people who have applied the framework to their own roadmap, thesis, or category. Venture partners have called it “a macroeconomic, industry-defining model” once the Applications view and vertical maps click into place.
Most 'AI frameworks' are taxonomies. This one is absolutely designed to be a macroeconomic, industry-defining model. Once you see the Applications view, the vertical maps, and the Cube together, it stops feeling like a stack diagram and starts feeling like a way to price an entire industry.
Partner, Multi-Stage Venture Fund
Investment Partner (attribution withheld)
JTBD tells you the length of the customer need. Supply Chain of Intelligence tells you the depth of the answer, how many layers you have to own to deliver it durably. 'Trust the output' is one job; you can answer it shallow with a verifier widget, or deep with an L3 gatekeeping layer baked in. The framework finally gave me a vocabulary for that trade-off.
I have sat through a hundred 'AI strategy' decks. This is the first one that told me which layer a product was actually on, and which layer it had to move to before the model layer ate it. The diagnostic is brutal in a useful way.
We were calling ourselves an 'AI platform' and the framework made us see we were a thin L7 surface on top of someone else's L2. We rewrote the roadmap inside a week to compound on L1b proprietary data instead. The language travels, engineering and GTM both speak it.
Working on AI and ads inside a platform company, you feel the layer compression in real time, what was an app last quarter is a feature this quarter. The 10-layer map is the first framework that names that dynamic instead of describing it after the fact.
Use it
Bring the framework to your next deal or board meeting.
Use it on your portfolio, cite it in your IC memos, link to it in your decks. Questions, pushback, or corrections are welcome on LinkedIn.
Personal thinking by Anand Arivukkarasu, shared freely. Views are my own. Disclaimer →