The partner thesis
Scale the work, not the headcount.
Infra One’s thesis is a human-and-agent production firm, starting with fund administration. Own the delivery and its service economics, rather than sell a copilot. Agents and deterministic services execute routine work; humans retain judgment, review and finality.
An operating hypothesis. Not measured company performance.
Working hypothesis: an operating advantage to prove.
500 funds · same scope · annual EUR- Cost advantage
- 48.8%
- Operating margin
- 62.2%
- Δ annual operating profit
- +€9m
Lower all-in cost vs. modeled incumbent
Infra One · after 10.0% price discount
More than incumbent · after discount
This case assigns 80% of routine production to systems, assumes 12% of routed cases need a person, and retains 48 human review hours per fund-year. It models 132 total human FTE, versus 317 at the incumbent, including engineering and corporate staff.
All inputs are illustrative, not calibrated to company disclosures. Cases share volume and pricing. Active incumbent: 20% system share, 1.0× residual-work speed. Later edits in the full model do not change these reference cases.
Compare the operating assumptions11 rows
Key operating assumptions plus every difference between these cases. System share applies to routine work only; exceptions and review are additional. All 50 exact inputs travel with “Open this case in full model” and the export.
| Assumption | Execution downside | Working hypothesis · selected | High reuse |
|---|---|---|---|
| Infra One system share% | 50% | 80% | 92% |
| System cases needing a person% | 40% | 12% | 5% |
| Infra One review hours / fundhours | 90 | 48 | 36 |
| Infra One residual-work speed× | 1.0× | 1.2× | 1.5× |
| Infra One engineering floorpeople | 18 | 12 | 12 |
| Infra One corporate staffing% | 15% | 15% | 15% |
| Infra One software / fundEUR | €800 | €800 | €800 |
| Cost / system executionEUR | €15 | €2 | €2 |
| Infra One funds / extra engineerfunds | 250 | 250 | 500 |
| Annual exception-rate improvement% | 0% | 20% | 25% |
| Infra One annual system-share gainpp | 0 pp | 3 pp | 2 pp |
Growth and learning differences are preserved for the full model; they do not affect this annual steady-state brief.
01 / The economics
Where the economic value goes
less €13,610,200 Infra One cost
Work volume stays unchanged
less incumbent operating profit
One bridge, not three additive benefits. Retained incremental profit is after the customer concession. Amounts rounded for display.
Automation alone is not the whole thesis. The automation-only comparator gives 40.4% lower cost and a 60.3% operating margin at 0.0% discount. Engineering and training remain costed.
48.8% maximum total discount would match the incumbent’s percentage operating margin. Not an extra discount on top of today’s 10.0%, and not the same absolute profit.
- First cost parity
- 40 funds
- First Infra One operating break-even
- 65 funds
- First incumbent operating break-even
- 68 funds
First integer crossing in a 1–20,000 fund scan; hiring steps can cause later reversals. Cost parity means Infra One cost ≤ incumbent cost; operating break-even means profit ≥ 0. Operating scale, not cash runway or a demand forecast.
02 / The investment logic
Why this could be a venture business
- 01
Own delivery economics.
Capture the value of completed service work, rather than the price of a software seat. The firm owns the work and accountability, not just the tool.
- 02
Make operating knowledge reusable.
Rules, tests and reviewer feedback could compound across repeatable work. The test is lower total human effort, not manual work moved off-book.
- 03
Earn the right to expand.
Adjacent regulated workflows are an option only after proving the wedge. Scope, authority, customer distribution and delivery quality gate each expansion.
Demand and retention still need proof. Defensibility must come from reusable operations and accountable delivery, not privileged access to a model.
03 / The evidence boundary
What is evidence, what is hypothesis
Company revenue and workforce disclosures provide context. They do not demonstrate the cause of any productivity difference or validate Infra One.
Founder materials describe Harness engineering primitives. They do not evidence live regulated fund-production efficiency.
Fund-production quality, system and exception performance, review burden, customer demand, retention and commercial economics.
Alter Domus
Company disclosureAlter Domus Global S.à r.l. consolidated IFRS financial statements (top consolidating entity of the Alter Domus group)
Eligible EUR ratio · not a productivity claim
Revenue: FY2025 (year ended 31 December 2025) · Workforce: FY2025 · average number of people employed (employees 5,719 + directors 233)
Same consolidated group scope and same FY2025 period; denominator is the company-disclosed average number of people employed (5,952 incl. 233 directors). Using employees only (5,719) gives EUR 163,648.
TMF Group
Company disclosureTMF Group consolidated (TMF Group Holding, redomiciled to Jersey 2 Jan 2025); adjusted figures as presented in the FY2025 annual report
Eligible EUR ratio · not a productivity claim
Revenue: FY2025 · Workforce: FY2025 · average number of FTEs
Same group scope and same FY2025 period; denominator is company-disclosed average FTE (11,906), the cleanest denominator in the cohort.
Even eligible period-matched revenue/workforce averages have different service mixes and workforce bases. No blended peer average; no causal claim. A missing profit metric does not, by itself, invalidate available revenue.
Read evidence delivery and provider statuses
Bundled snapshot · 12 Sep 2026: 30 company records bundled in the application. No connection is required to read or export the snapshot; external source pages and optional refresh require internet access.
Research: Complete dated research and archived standardized financial evidence, included in the app download.
Financial data: Bundled snapshot, not live data. FY2025 provider figures were fetched on 12 September 2026. Five public firms have operating profit; three remain intentionally partial. Live refresh requires the hosted data service.
04 / The underwriting gates
What must be true
- 01
80% system share cannot conceal manual work.
Prove the 12% exception assumption on representative routine cases. Count human intervention, contractors, handoffs and remediation, not just successful system runs.
- 02
48 review hours must preserve quality.
Demonstrate comparable accuracy, deadlines and effective controls at this annual per-fund burden. Review floors and peak staffing are planning assumptions, not regulatory validation.
- 03
500 funds at €72,000 must be earned.
Validate demand and retention at the annual net fee after 10.0% discount. Prove acquisition and onboarding economics; a volume input is not a customer pipeline.
1 selected-model assumption to validate
- Faster review is assumed, not proven. Validate review quality and deadline capacity before underwriting the time saving.
Outside this model: financing, taxes, depreciation/amortization, capital investment, M&A, transition and onboarding cash ramp, working-capital timing, churn and acquisition economics. Error, liability and service-failure losses are not monetized. Comparable scope and quality are assumed, not proven. No forecast, investment, legal or regulatory assurance.
The next discussion