The founder's score
Friday morning. Marcus had filled in the AIM Index spreadsheet honestly the night before. Ten dimensions. Each scored 1 to 5. The average had landed at 2.1.
He had expected a 3. He had written, in his own private estimate three weeks earlier, that he was probably a 3.2. Sarah Kessler (capital markets advisor) had emailed him the framework on Monday after walking an LP through a peer firm's diagnostic. Her email had two lines.
This is the tool the institutional capital is using to measure the second moat. Score your firm honestly before we talk Friday.
He had carried the spreadsheet around the office for four days. The first pass landed at 2.8. The second at 2.5. By Thursday night, sitting at his desk after everyone had left, he had scored it the way a stranger reading the firm cold would score it.
Two-point-one.
The Zoom with Sarah was at 9 AM. He had asked Priya (investment associate) to join. Sarah's idea, two days earlier: Get Priya in the room. She has seen 4.0 firms.
Scoring the firm an LP would see
Sarah called from her Houston office. The window behind her showed Memorial Park, gray with January rain. Marcus and Priya were in the small conference room at Chen Capital with the spreadsheet projected on the wall.
"Walk me through the score," Sarah said. "Start with the foundation."
"Data Infrastructure: 1.5. The quarterly report pulls from four systems. Priya bridges them by hand."
"Why not 1?"
Marcus paused. "Because each system works in isolation. The data exists. We just cannot get it to flow."
"That is Level 1. Level 2 is when 'NOI' means the same thing in every system without a person translating. Does it?"
"No."
"Then the score is 1."
He changed it.
"Workflow Integration: 1. The IC memo takes three weeks because five people contribute sections without shared context."
Sarah nodded. "Decision Intelligence?"
"1.5. The investment thesis is in my head."
"And the buy box?"
"Also in my head."
"And the IC criteria?"
"Mostly in my head. Some lives in IC memos from prior deals."
"And the LP preferences?"
"In my head, my co-founder Greg's head, and a CRM that nobody updates."
"That is a 1, Marcus. Decision Intelligence at 2 means a documented version exists that an analyst can screen against without asking what you meant this week. You do not have one."
He changed the score.
The dimensions kept moving downward. Institutional Knowledge: 1. Team AI Capability: 2. AI Governance: 1. Validation Architecture: 1.5. LP Reporting Maturity: 2. Scalability Architecture: 1. Intelligence Compounding: 1.5.
(The full rubric for all ten dimensions, with the level definitions Sarah was holding Marcus against, is in Appendix A. The exercise Marcus is running is one any reader can repeat in an afternoon.)
The new average landed at 1.3.
Sarah had one more question. "Who else needs to be in this room?"
Marcus texted Jordan Wells (Head of Investor Relations) and Anika Reeves (General Counsel). They arrived within ten minutes. Claudia (CFO/COO) followed them in without being asked. She had been listening from the doorway.
Jordan scored LP Reporting Maturity. "I want to say 2.5. The reports go out on time. The content is solid. But I rebuild every LP brief from scratch because we have no system that remembers what each investor asked for last quarter. I carry a spreadsheet of LP preferences that I update from memory. If I got hit by a bus, nobody could reconstruct what I know about forty-seven investor relationships. The score is a 1.5. Maybe a 1."
Anika scored compliance and governance. "AI Governance: 1. We have no written policy. Validation Architecture: 1. Most outputs leave the firm without a structured verification step. I told Marcus six months ago that our compliance documentation was inadequate for institutional diligence. The ODD team confirmed it."
Claudia scored Scalability Architecture. "The answer is 1. We cannot add fifty percent more AUM without adding proportional headcount. The platform does not absorb complexity. The team absorbs it." She paused. "Workflow Integration from my seat? We have no weekly scorecard, no issue-resolution rhythm, no quarterly planning cycle that connects strategy to execution. When something breaks, it escalates to you or it does not get fixed. That is how you get thirty-five unresolved items on a single Monday. I know because I count them. Score: 1."
The new average, with their honest scores folded in, dropped to 1.2.
Marcus sat back from the spreadsheet and did not say anything for a full thirty seconds.
"Sarah, this is humiliating."
"It is accurate. Humiliating is your reaction. Accurate is the score. The two are different. The 2.1 you came in with was the founder's score. The 1.2 we landed on is the LP's score. They are scoring you against this framework whether you know it or not."
ServiceNow and Oxford Economics' 2025 Enterprise AI Maturity Index surveyed 4,500 executives. The average global score dropped from 44 to 35 year-over-year, even as Gartner forecast worldwide generative AI spending would rise 76 percent in 2025. Fewer than 1 percent scored above 50.
Sources: ServiceNow / Oxford Economics, Enterprise AI Maturity Index (2025); Gartner, GenAI spending forecast (2025)
The industry is spending more and measuring worse. In the diagnostics I have run, firms rate their maturity one to two full levels above where a structured scoring places them. They assess based on their best use case rather than their average. The analyst using a chat assistant for deal memos becomes the firm's evidence of AI adoption. Manually assembled quarterly reports, scattered LP preference data, an IC process running on email, all invisible to the self-scoring exercise.
Three statements I hear in CEO interviews, each offered as evidence of maturity, each actually Level 1:
"We use AI for deal screening." Level 1 on Team AI Capability if one analyst is using a public chat assistant on his own. Maturity is not the presence of the activity; it is the activity as a repeatable, governed workflow.
"Our data is in good systems." Level 1 on Data Infrastructure if those systems do not share a common data architecture. Three good systems and one bad infrastructure is still Level 1.
"We have a clear investment thesis." Level 1 on Decision Intelligence if the thesis is clear to the CEO who built it and undocumented for everyone else.
The CEO who scores their own firm at a 3 is the CEO who has not had a Sarah read the score back to them.
What a 4.0 firm looks like
Sarah looked into the camera. "Priya, you have been quiet. Walk us through what each of these dimensions looked like at your old firm."
Priya glanced at Marcus. He nodded.
"Data Infrastructure was a 4. Maybe a 4.5. Every system spoke to every other system through a central data warehouse. NOI was defined once at the data layer and pulled into every report. There was a data dictionary. I did not know what a data dictionary was for the first six months."
"Workflow Integration?"
"4. Every workflow had a documented owner, input, output, and handoff. The IC memo template loaded the deal data already mapped from the underwriting model. The memo took five days from kickoff to vote."
"Decision Intelligence?"
"3.5. The buy box was a written document with version history. The IC meeting did not vote on what the criteria were. It voted on whether the deal met them."
She kept going. Institutional Knowledge: 4. LP Reporting Maturity: 4.5. Validation Architecture: 4.
When she finished, the screen showed the comparison side by side. Chen Capital averaged 1.2. Her old firm averaged 4.1.
Marcus looked at the gap.
"Why did you not tell me this fourteen months ago?"
"Because I did not know what I was looking at," Priya said. "I had never worked anywhere that operated below a 4. I assumed the way Chen Capital ran was just the way smaller firms ran. I am seeing it now."
Sarah let the silence sit for a beat.
"Marcus, that gap is the second moat. The firm Priya came from is not five times better than yours at sourcing. They are five times better at infrastructure. The LP across the table from you in March is going to ask the same questions Priya just answered."
The build sequence
Sarah pulled up a build sequence on her screen.
"The dimensions do not weight equally. Data Infrastructure and Workflow Integration form the foundation. A firm cannot advance past Level 2 on any other dimension if these two are at Level 1. The Tire Principle as a diagnostic. It does not matter if your Team AI Capability is at Level 4 if your Data Infrastructure is at Level 1. The tire is only as fast as its flattest side."
"AI automation is a function of upstream data-shape stability. Asset-management reporting is the easiest place to start because the data is stable. Property-management software produces operating statements in a consistent format. Accounting reconciles the same general-ledger structure every quarter. The output, the LP-grade quarterly report, has a known-good baseline the firm has produced manually for years. The platform can be wrong, and the firm can tell."
"Acquisitions automation is the hardest. A deal enters as a one-sentence email, a fully papered data room, a phone call, or a dinner ask. The same downstream screening process absorbs wildly different upstream data shapes. Until the ingestion side is stabilized, the AI layer has nothing stable to operate on. Start where data is stable, because that is where the platform can be tested."
Compliance infrastructure sits underneath every dimension. The data has to be auditable. Workflows have to preserve evidentiary trails. The institutional LP will ask for the compliance manual, the CCO's background, Form ADV Part 2, and the cybersecurity plan before the investment team meets the GP. Compliance is a gating function pre-diligence. A firm arriving to a first meeting with compliance treated as an afterthought loses allocation to a peer whose compliance was built into operational architecture from the beginning.
What codification surfaces
The build is primarily a codification deliverable. The automation is what the codification makes possible. The codification exercise is where most firms discover they do not actually know what they agree on.
A firm says "seven-cap retail in Austin is attractive." Every seat nods in the conference room. Then the live deal runs through the full economic stack: management fees, asset-management fees, debt service, distributions, required after-fee cash-on-cash return. The fifty-basis-point cap-rate pickup that looked like alpha at the cap-rate layer evaporates by the time equity reaches the LP's cash position. The team codified at the wrong altitude.
"Fifteen percent IRR" reads as gross or net-to-LP depending on who is reading. In the funds I have seen measured, gross and net-to-LP separate by some three hundred basis points. Pass-or-fail flips depending on which reading sits in which head. The analyst screens pass on gross. The IC committee evaluates fail on net-to-LP. The disconnect was built into the criterion before the first deal crossed the desk.
The codification exercise exposes disagreements the team has been hiding. The firm has already been paying for this divergence without knowing it.
What the score does to the conversation
Before the diagnostic, talk about ops is abstract. "We need to improve our operations." True and useless.
After the diagnostic, the conversation becomes concrete. "Our Data Infrastructure is at Level 1, which is why the quarterly report takes seventeen days (what we used to round to two weeks) and burns ten days of asset-manager capacity every quarter." "Should we invest two hundred thousand dollars and three months to move Data Infrastructure from 1 to 2?" produces a real decision with real resource implications, defensible to a board, measurable in outcomes.
The CEO whose number stings defaults to one of two postures. Inertia: the firm is at 1.2 and always will be. Or a burst of activity (announce the initiative, hire a CTO, sign three vendor contracts) that creates the appearance of change without doing the work. The firms whose scores advance are the ones whose CEO could carry the original number for a week before moving.
The number we work from
The Zoom ended at 10:35. Sarah had a call with an LP at eleven. Priya stood up to leave. Marcus stayed at the table.
"Priya. Sit down for one more minute."
She sat.
"What you said about not knowing what you were looking at. I have been doing the same thing in reverse. I have been looking at this firm for twelve years and seeing the version of it I want to see, not the version that an LP would see. The 1.2 is the version they would score. The 3.2 I had in my head was the version I had built without measuring."
Priya nodded.
"What I want is for you to score the firm the way you would have scored your old firm if you had known the framework when you were there. Honest. Without softening. The number you give me is the number we work from."
"Marcus, I just gave you that number with Sarah."
"I know. I needed to hear myself say back to you that I accepted it."
She nodded once more and left.
You cannot see your own firm from the outside, and that is not a flaw in you. The proximity that let you build it is the same proximity that hides it from you now. You carry the firm in your head, and the version in your head is the one that survives the day's pressures: the founder's story, the favorable benchmark, the workflow as it should run rather than as it does. The version capital scores is a different firm. The version a data room reveals is a different firm.
I have watched operators I deeply respect take this number and read it as an indictment. It is not one. The diagnostic is not there to humble you. It is there to hand you the outside view you cannot generate on your own, the way a sharp board member or a trusted advisor does on their best day. The 3.2 in Marcus's head was the firm he had built without measuring. The 1.2 on the page was the firm a sophisticated outsider would meet. The number stings because you built the thing, and then it stops stinging, because a firm you can finally see is a firm you can build.
The work is not to feel worse about today's number. It is to make today's number obsolete by next quarter, and that one obsolete by the quarter after. The CEO who steps outside their firm long enough to take the independent view is the one who gets to begin the real work of building the platform.
Marcus opened a new document.
AIM Index: Current State Assessment
Below it, the ten dimensions and their honest scores. A single line below that:
Dimension 1 and Dimension 2 first. Everything else follows.
The build was eighteen months whose progress would not feel like progress for most of it. No quarter would produce a 1.2-to-3 leap. The dimensions would move half a level at a time. The firms whose AIM scores compounded were the firms whose CEOs accepted the work as discipline applied steadily over years.
Marcus had built this firm on inspiration's cadence. The sourcing trip that produced the deal of the quarter. The IC session that ran late. The fundraise that closed because he out-worked the calendar. The platform required a different cadence. Daily. Scheduled. Protected. Boring on any single day. Decisive over a decade.
Stop optimizing deals. Start building the system the deals flow through.
Part II was over. The vocabulary was built. The diagnostic was scored. The distance was measured in infrastructure. Specific infrastructure. Built in a specific sequence.
Part I had answered: what is wrong with my firm? Part II had answered: what should my firm look like?
The question Part III answers is operational, concrete, and urgent: how do I build it?