The Research
The thesis was stress-tested by a five-workstream program — ~60 competitors, 4 geos, and two independent reviews that converged on the same answer.
Structure: one shared base memo + five workstreams. The base memo held every workstream to the same thesis (Commercial Memory Layer; commercial event as core object; demo must not look like voice-to-pretty-diary).
- Competitor landscape analysis
- Partner-facing strategy deck
- Construction SOP + field-evidence ruleset
- Live demo product spec
- Future-state mockup + meeting/data-ask plan
The first two were load-bearing. The broad competitor pass + SOP study confirmed the wedge held — no attractive opening in AI daily reports, takeoff, BIM, AR/360 capture, or golden-thread handover (all crowded or specialist-led). They grounded the domain logic onto Procore’s Daily Log, the SCL Protocol’s record-keeping, and JCT/NEC variation and compensation-event concepts.
Method — what makes the conclusions defensible. Research method. Every competitor scored on the same two instruments:
- 21-area coverage grid — where a tool plays.
- Four-axis opportunity lens — whether and how we could attack.
- Segmented reviews from real Capterra data where it existed (248 analysed for Raken; samples for Procore, Fieldwire, Kreo, Pype, Sablono), bias-tagged: vendor-solicited 5-stars down-weighted, field vs office voice separated.
- No absence-claims without proof — assert “lacks X” only from having searched and found nothing.
Twelve full dossiers:
| Dossier | Dossier |
|---|---|
| Procore | OpenSpace |
| Raken | Trunk Tools |
| Fieldwire | Civils.ai |
| PlanRadar | Kreo |
| Document Crunch | Pype |
| Sablono | BuildPass |
Seven landscape scans (last two added to test the thesis head-on):
| Scan |
|---|
| Reality capture |
| Estimating & takeoff |
| Field daily reporting |
| Platforms & system of record |
| Materials & handover |
| Claims & delay |
| Commercial QS |
~60 competitors scored on the same scales, feeding a market problem map and an opportunity lens.
De-biasing. Early thinking was UK-shaped. A four-country beachhead scan (US, Canada, Australia, UK) by independent agents against real government/industry/vendor sources checked whether the “empty market” story was an artifact of one country.
Two independent strategic reviews, run as separate passes, each reaching its own go/no-go verdict — to test whether independent analysis of the same evidence converges or diverges.
It converged — same beachhead, same vehicle, same corrected thesis. The program’s verdict on itself: “Stop researching the field — it is mapped. The two target columns are empty and the supply chain is identified.” Next: the landscape and the decision.