The Wrong Asset

We were financing the wrong thing. The engagement was never the asset. The asset was the office that decides whose engagement is worth backing.

When we argued that AI is turning professional services into a financeable asset, two very different rooms pushed back. The second room was sharper than the first.

The bankers asked a legal question: can you actually claim the cash flow when the firm fails? We wrote about that in a companion piece on assignability. The honest answer narrowed the thesis from securitisation to underwritable working capital. A smaller phrase for a bigger near-term prize.

Then the private equity people read it. They did not ask whether the thesis was elegant. They asked four questions, and they are the only four that matter to them. Can I buy this asset. Can I lever it. Can I improve its cash conversion. Can I exit at a higher multiple.

Run the essay through those four and it stops being a piece about AI. It becomes one claim about re-rating: can a services firm move from an earnings multiple to an asset-backed one.

That is a more interesting claim than the one we made. It is also far more attackable.

What They Conceded

The PE lens found real money in the idea. It sits in three places, and none of them require a single security to be issued.

1. Multiple expansion. Consulting trades at a modest earnings multiple, often somewhere from 7 to 11 times, because the revenue is read as non-recurring, people-dependent, and hard to finance. High-quality recurring data businesses trade at twice that. Change how a lender and a buyer perceive that revenue, with a clean underwriting spine, and the multiple can move. PE lives for that arbitrage.

2. Leverage. This is the one that matters. A buyout improves when debt capacity rises. Let underwritten pipeline quality carry a consulting business at, say, 3.5 to 4 times net leverage instead of 2, and the returns change dramatically. So in PE terms the thesis was never about building a tradable instrument. It was about debt capacity. A more boring sentence, and a much more compelling one.

3. The roll-up. This is the strongest investable version of the whole idea. Buy twenty niche consultancies, in cyber, in regulatory advisory, in ERP migration. Impose one underwriting spine across all of them: standardised scope, acceptance, evidence, collection. The platform becomes more financeable than any firm inside it. That is a classic platform play, and it needs none of the asset-class language. The spine alone does the work.

The bull case survived. It just stopped depending on the word we led with.

What They Would Not Buy

Then they went after the risks. Three of them land hard.

Concentration. Most consulting firms carry ugly revenue concentration, top five clients at forty to sixty percent. One cancellation craters the year. An underwriter sees that and stops reading.

Key-man risk, again. PE hates a business whose value walks out the door at night and might walk into a competitor by morning. A pipeline-quality model is worth very little if the partner who owns the relationships resigns.

Goodhart's Law. The sharpest one, the one we had not priced. The moment financing depends on structured milestones, firms optimise for the milestone instead of the outcome. Fragment the deliverables to clear acceptance gates faster. Soften the criteria. Dress up the evidence. When a measure becomes a target it stops being a good measure. An underwriting spine is exactly a measure, and we were proposing to turn it into a target worth real money. That is not a detail. It is a live failure mode.

The Pre-Mortem

A good investment committee does not argue. It runs a pre-mortem. Assume it is 2031, the thesis failed, now say why. Two of the answers kept us up.

Adverse selection. The best firms never needed this. They self-finance, they generate cash, they borrow cheaply against the corporate balance sheet. So who shows up for pipeline financing? The cash-constrained, the weaker operators, the firms with broken working capital. The financed pool skews bad before the first model runs. That is how credit markets quietly poison themselves, and we had not addressed it.

Observable is not predictable. AI can measure formal delivery quality beautifully. It cannot see the client's politics, the new CFO, the budget freeze, the strategic reprioritisation that ends the engagement for reasons that were never in the data. The variables that kill deals are off-model. Build a credit market on a confident score of the observable and you rebuild, in miniature, the thing that failed in 2008: a model that priced what it could see and ignored what it could not.

The Turn

Here is where the PE lens did something the banking lens did not. It did not narrow the thesis. It told us we had been looking at the wrong asset.

We kept asking how to finance the engagement. The engagement may never be the asset. Look at who captured the value in credit markets. Not, mostly, the lenders. The infrastructure. FICO. Moody's. S&P. MSCI. The firms that took none of the credit risk and instead sold the language everyone else used to price it. The rating layer outearned the lending layer for a century, with fatter margins, a data moat, and network effects no lender ever had.

The opportunity hiding inside the essay was never to securitise consulting. It was to build the rating engine for knowledge-work delivery risk. The Moody's of whether an engagement will land.

That is a software business, with recurring revenue, a widening data advantage, and a defensibility a consulting roll-up will never have. We find this slightly painful to write, because it is obviously right, and it reframes everything we argued. The thesis was not "professional services becomes an asset class." The version that survives two rooms of professionals trying to break it is quieter and stronger: AI is creating a new underwriting infrastructure for knowledge-work cash flows, and the money is in the infrastructure, not in the cash flows.

The cargo was never the business. The business was always the office that decides whose cargo is worth insuring. We led with the bond. We should have led with the bureau.