A fixed-scope, fixed-price, fixed-deadline study of one technical question, ending in a written go/no-go verdict with the evidence to reproduce it.
Written in one paragraph and agreed before work starts, so the verdict answers what you actually need to decide.
Papers, patents, datasheets, standards and your own data, listed so an evaluator can see what the verdict rests on.
First-principles analysis, simulation, code review or re-analysis of data, and the checks each method was put through before being trusted.
What the evidence shows, with the numbers, and where the evidence is thin.
Go, go with conditions, or no-go. Conditions are spelled out as measurable statements, not adjectives.
Each technical risk ranked by severity and likelihood, with what would settle it.
The measurements or tests to run next, in order, with what each one would prove and a rough cost.
The scripts, inputs and a run guide, so you or anyone you choose can reproduce every figure in the report.
The structure and tone below are those of a real report. The content is invented for illustration; client reports stay confidential, and redacted excerpts are shared on request with the client's permission.
Can the proposed passive cooling coating reach the claimed 8 °C sub-ambient surface temperature under the client's deployment conditions (humid coastal site, 40 % average cloud cover)? The decision this informs: whether to fund a 50 m² pilot installation in Q1.
Two supplier datasheets, the client's emissivity measurements (3 samples, FTIR, 2.5–25 µm), twelve peer-reviewed papers on daytime radiative cooling, hourly weather data for the site (3 years).
Radiative balance model built from first principles and reproduced against two published results before use; sensitivity sweep over humidity, cloud cover and wind; spectral data checked for the atmospheric window.
Go, with conditionsThe 8 °C figure holds only on clear, dry nights. Under the site's measured conditions the model gives 2.5–4 °C sub-ambient in daytime. The pilot is worth funding if the business case survives at 3 °C.
| Risk | Severity | Likelihood | What would settle it |
|---|---|---|---|
| Humidity closes the atmospheric window | High | Likely | One month of on-site surface temperature logging |
| Emissivity drops with soiling | Medium | Possible | Re-measure samples after 8 weeks outdoors |
| Supplier spectral data overstates the 8–13 µm band | Medium | Possible | Independent FTIR on supplier-grade sample |
Log surface and ambient temperature on one coated and one uncoated panel for 30 days before committing to the 50 m² pilot. Estimated cost of the test is a small fraction of the pilot.
Twenty minutes to define the question and the decision; I send the scope and the price the same day.
Day 0
Evidence gathered into the project library; I confirm with you that nothing relevant is missing.
Days 1–2
Agents run the literature synthesis, calculations and simulations; I verify each step and send back what does not hold.
Days 3–7
The draft report and the reproducibility package go through a final check against the question.
Day 8
You read the draft and we talk through anything you dispute.
Day 9
The final report, the package and a 45-minute walkthrough with your team.
Day 10
When the question is clear and the clock is short, a three-day Red-Flag Read (€900) tells you whether a sprint is worth it.
All pricesDoctorate from CEA-Leti and Université Grenoble Alpes, invited speaker at the MOS-AK workshop, peer reviewer for Solid-State Electronics, thirteen completed client missions across physics, engineering, finance and AI. I have no prototype to sell you afterwards, so the verdict has no thumb on the scale.
Does not clear target Redesign directions flagged for the next phase.
No-go Verdict delivered with the backtest code.
Then I say so on the scoping call and propose either a narrower question that can be settled in a sprint, or a two-sprint plan with a gate between them. You pay nothing until the scope is agreed.
Only what the question requires, and only after the NDA. Many sprints run on datasheets, published results and a first-principles model. When your data is the evidence, it is handled under the data-handling terms we agree on the call.
The reproducibility package exists for this reason. Run the scripts, change the assumptions you dispute, and we discuss the result on the walkthrough call. One revision round is included; if the evidence changes the verdict, the report changes with it.
Yes. The report is written so that it can be placed in front of an evaluator or an investment committee as it is. Many clients use the risk register and the next-experiments section directly in their plans.
Send a short description of the technology and the decision it informs. I reply personally, usually within one working day, and a first call is twenty minutes with no commitment on either side.