AI, underwritten. The problem with most AI spending is not the technology — it is the missing underwriting standard. Pilots demo well, die in diligence, and never reconcile to a number finance can check. Return Engineering is the discipline that fixes that: every build starts as an operating case and has to keep passing it.
Every business has a short list of goals that actually move enterprise value: a margin problem, a capacity ceiling, a revenue line that should compound. Return Engineering starts there — with the goal leadership would defend in a board meeting — not with a model shortlist or a vendor demo.
The sponsor, the operating owner and the builder agree the result in advance: what changes, who owns it, what the baseline is today, and which measure settles whether it worked. Alignment on the measure is what keeps a project from being declared a success by whoever shipped it.
Build, buy, process change, or no build — each option is priced against the baseline with its full cost: delivery, review effort, client time, running cost and downside. A forecast is labeled a forecast. If the case does not earn a build, the recommendation is no build, in writing.
Secure AI around your existing systems — sources, permissions, exceptions and human review defined before the first release, with held-out acceptance tasks and stop conditions agreed up front. The smallest system that moves the measure ships first; breadth follows evidence.
Operators and maintainers demonstrate the work on fresh tasks, not attendance certificates. Feedback becomes a candidate change that is evaluated against the baseline and released by a named human decision — or rolled back. The measure keeps being reported after launch.
Name the lever — revenue, gross margin, operating cost, working capital, capacity, risk avoided. Show baseline → change → dollar effect, with the evidence labeled: measured in production, client-reported, or modeled estimate. State the clock: time-to-value and payback.
Finish the multiplication. ΔEBITDA × your multiple = enterprise value created. And build the kind of capability a buyer re-rates: recurring revenue, gross margin, growth, proprietary data.
Worked example — labeled MODELED
Value created: $2.8M – $92.8M
At one end, a single workflow at a mid-market operator: release $400K of annualized cost in a services business at 7× and you have not saved $400K. You have created $2.8M of enterprise value — if the capacity converts to cash. At the other end, a company-wide program at a larger enterprise: $11.6M of ΔEBITDA at 8× is $92.8M of enterprise value — and $11.6M is roughly a 10% EBITDA lift at a company with $100M+ of EBITDA. One workflow → the whole business. Capacity released is not cash saved; the conversion is the engineering.
The right answer is sometimes no build: when there is no named owner who can change the workflow, no baseline anyone trusts, no measure leadership will accept, or a process change or an existing tool solves the problem at a fraction of the cost. A firm that will tell you that — before anyone writes code — is the firm to trust with a build.
Goal-first AI transformation, end to end: underwrite the case, build the smallest production system that moves the measure, embed it and hand it over.
Adam is brought in as a part-time Chief AI Officer to look at the entirety of a business and tackle workflows across the whole business — embedded with the leadership team, not scoped to one department or one tool.
AI underwriting for sponsors — diligence, contract and procurement analysis, deal knowledge systems — and portfolio operating systems for their companies.
A working session that underwrites one goal live and ends with the measure, the case, and a build / buy / no-build call.
Current work touches healthcare services, marketplaces, hospitality, retail and restaurants, commercial and industrial services, shipping, trucking and logistics, repair and the trades, and financial services — and PE portfolio companies across all of them.
Private equity sponsors and middle-market company leaders in the US and Europe. The work connects operating goals, application engineering and workforce capability. Fit depends on the workflow and available evidence, not the industry label alone.
Founder Adam Roseman performs the hands-on AI engineering himself; the person who scopes the work builds it, trains the team and maintains it. For the past three years he has led full-scale AI transformation and restructuring for organizations from startups to billion-dollar enterprises, including regulated environments. He previously co-founded and led as CEO an AI and machine-learning fintech that turned billions of unstructured transaction records into financial intelligence (No. 86 on the Inc. 5000), earlier founded a media data business (Ad Age Global 40 Under 40), and began as an M&A investment banker with private-equity investor experience. Engineering capacity, specialist coverage and backup responsibilities are agreed for each engagement.
Start with your operating goal, accountable owner and first workflow. A mandate session is a fit conversation about constraints and next decision—not free diligence or a commitment to a build. If there is a fit, the next step is a scoped assessment or operating mandate for review.
Otomat underwrites the case, builds secure AI around your existing systems, and stays through adoption—strategy, engineering, training and maintenance led by one senior AI engineer. The discipline connects investment judgment, operating judgment and engineering judgment into one delivery process with explicit acceptance and handover.
A tool rollout deploys a product. AI restructuring changes the workflow, ownership, controls, adoption and measurement around a business goal. Installing software alone does not establish operating change. The work may recommend using an existing tool, a process change or a custom build depending on the evidence.
No. There is no promise of a particular financial return or rollout date. Fees, capacity, exclusions and any performance component are negotiated for the specific mandate. Agree the baseline, measurement window, attribution and client dependencies before committing to a build.
Handover requires that the client primary and backup each perform a fresh task, diagnose an exception, release an approved change and restore a known version. Workforce capability, controlled improvement and operating measurement are acceptance tests. Training attendance alone does not establish independence.
Bring the project — or bring the problem. Start with a Return Review: one goal, underwritten live, with the measure, the case, and a build / buy / no-build call.
Book a Return Review