Assessment at scale
You're rejecting over 90% of ventures based on a pitch deck and a thirty-minute call. Somewhere in that 90% are the deals that would have returned the fund. The problem was never finding more ventures. It was assessing the ones you already have — all of them, at depth.
The mathematics of manual assessment
A typical early-stage fund receives 500–1,500 applications per year. First-pass screening — deck scan, pattern match, gut feel — filters out 75–90%. The survivors get a second pass: an analyst reviews the deck, researches the market, checks the team, writes a memo. That takes 8–12 hours minimum. At €150–200 per hour loaded, each second-pass costs €1,200–2,400.
For the top 25 candidates that make it to partner meetings, a proper deep-dive — comparable to what EviDimensional produces — takes 33–43 hours of analyst work.
And the output of the 25 that do get one? An unstructured memo that varies with the analyst. No consistent scoring. No evidence grading. No cross-pipeline comparability. Company A's memo was written by a sharp analyst on Monday morning. Company B's was written by an exhausted one on Friday afternoon. Both memos look the same. The quality is not.
A thousand inbound. Twenty-five assessed at depth. Nine hundred and seventy-five filtered on pitch quality. That is not a selection process. It is a lottery with better odds.
What gets lost in the 90%
The 90% you reject after a deck scan includes the founder who can't pitch but has five validated pricing conversations. The technical founder who built something extraordinary but doesn't have the warm intro. The founder building in a language you don't speak, in a market you don't recognise, solving a problem you've never encountered.
A polished pitch deck proves someone can make a polished pitch deck. It doesn't prove they can build a company — and vice versa. Bessemer's anti-portfolio is what happens when this filter runs across decades: Apple, Google, Facebook, Airbnb, PayPal, Tesla, eBay, FedEx (passed on seven times). None of them failed during in-depth due diligence. They were screened out earlier, on pitch quality.
The question is not whether hidden gems exist in your rejected pile. They do. The question is whether you have a way to find them that doesn't require 40 hours per company.
Construction
The reason classical commercial DD does not scale is that it is built around partner judgment. A senior partner reads the materials, applies pattern recognition accumulated over decades, and forms a view. Junior analysts produce the desk synthesis the partner reads. The partner is the bottleneck. Adding more analysts produces more memos for one partner to read; it does not produce more decisions. The model caps at the partner's reading capacity.
EviDimensional engine inverts this. The methodology is encoded — 37 dimensions of value (22 elemental, 15 composite), five evidence levels distinguishing founder assertion from third-party validation, six readiness gates, seven valuation methods aggregated to a defensible range, 500 tarpit patterns screened against the company. The partner judgment is in the encoding, not in each individual run. Once encoded, the same depth applies on company number 1 and company number 1,000.
Same engine. More cycles. No degradation.
Manual analyst engagement
Each deep dive consumes 33–43 hours of analyst time at the partner's reading bottleneck. Quality drifts with analyst, day, fatigue, mood. Memos are not comparable across companies because each was written from scratch.
Cost per deep-dive: €5,000–8,500 · Time: weeks · Comparability: noneThe encoded engine
Each assessment runs the same 37 dimensions, the same evidence framework, the same gate logic. Output is structured, comparable, and produced at minutes-not-weeks compression. Quality does not degrade with volume because the methodology does not.
Cost per assessment: €2,950 / €4,950 · Time: minutes · Comparability: structuralToday
The two products that ship today already scale by repetition. A fund or programme commissions one assessment per company in the pipeline; the engine runs the same methodology on each one; the outputs come back structured and comparable. Manual coordination at the front, automated production at the back.
Pipeline coverage
Run a Public-Source DD on every long-list candidate. The same 37-dimension framework that classical commercial DD applies, against publicly available information only. The 975 you couldn't afford to assess at depth manually become assessable.
Sort the entire pipeline by evidence quality, by validation level, by specific dimension scores. The founder who pitched badly but has E4 evidence on customer validation becomes visible. The founder who pitched brilliantly but has E1 evidence on everything becomes visible too.
€2,950 per company · Same depth at company 1 and company 1,000
Portfolio depth
Run a Confidential-Source DD on every portfolio company on the same cadence. Same framework, same evidence grading, same readiness gates — applied with confidential materials. A sharp Friday-afternoon assessment looks the same as a sharp Monday-morning one because the methodology is the same.
Cross-portfolio comparability emerges naturally. Which companies are accelerating on the dimensions that matter for their stage. Which are not. Which evidence levels are climbing. Which are flat for two quarters.
€4,950 per company · Structured for cross-portfolio comparison
Twenty-five Confidential-Source DDs per year at €4,950 each = €124K. A junior-analyst-team budget that previously bought 25 unstructured memos now buys 25 structured assessments at the same depth, in minutes-not-weeks, with cross-portfolio comparability built in. The other 975 in your pipeline stop being a lottery — commission Public-Source DDs across them and the long list becomes navigable for the first time.
Next
Repetition gets you coverage. Automation gets you trajectory. The next layer of products closes the loop — same engine, same methodology, but applied at recurring cadence and surfacing the cross-cutting patterns that emerge when the assessment population accumulates.
An assessment is a photograph of a company at a moment in time. Portfolio Monitor plays the photographs forward into a movie reel. Lateral plays the cast on screen at once. Both run at public-source and confidential-source depth.
Public Portfolio Monitor — long-list watchlists for funds tracking companies they have not yet engaged. Quarterly re-assessments using public information only. The same Public-Source DD framework applied at recurring cadence.
Private Portfolio Monitor — portfolio dashboards for funds and programmes tracking companies they hold. Quarterly re-assessments using updated confidential materials. Evidence velocity becomes a leading indicator of company health. A founder whose validation levels are rising is executing. One whose levels are flat for two quarters is stalling. The data shows it before the board meeting does.
Public Lateral and Private Lateral — cross-cutting benchmarks that emerge naturally as the assessment population accumulates. Sector percentiles. Stage distributions. Cohort comparison. Ecosystem trends.
See the full Roadmap → — portfolio management lives in the Next column on the grid. Committed direction, in build, not yet shipping.
What this enables
When every venture in a pipeline or cohort is assessed on the same framework, a category of insight becomes available that no individual assessment can produce.
An accelerator with thirty ventures sees that fifteen are stuck at the same dimension. That is not fifteen individual problems. It is one programme-level pattern — a curriculum gap, a selection bias, or a systematic misalignment between what the programme expects and what founders arrive ready to do. Invisible with thirty individual mentor reports. Immediate with structured data.
Ventures rejected on pitch quality but with strong evidence on customer validation, monetisation, or moat become visible the moment the filter sorts on substance rather than presentation. The 90% rejection pile becomes searchable instead of opaque.
A founder whose validation levels are climbing is executing on hypothesis. One whose levels are flat for two quarters is stalling. The pattern is in the data before it is in the pitch deck. You catch drift in week eight, not month four.
Patterns across funds, accelerators, regions. Which idea categories are overcrowded. Which founder profiles are systematically underserved. Where regional clusters diverge from global benchmarks. None of this is possible when assessments are produced individually by different people with different frameworks.
The principle
The two were always the same problem. A system that can assess one venture at the depth a partner-grade analyst would produce can assess a thousand — because the methodology is the same, the framework is the same, the evidence grading is the same, and the output quality does not degrade with volume.
Manual analyst engagements scale by adding analysts; the Big Four can't productize the desk-synthesis layer because their margin model depends on partner-leverage rather than encoding. The competitive landscape argument is exactly this: the methodology has to predate the LLM, the LLM has to be capable of holding the methodology, and both pieces had to mature in that order. The gap is structural and the structure produced it.
What scales is what was already encoded. The engine is not getting smarter as it runs more assessments. It is getting more useful, because comparability across the assessment population is itself the value.
Sample assessments on a fictional DeepTech company are available, no signup required. Read both at the same depth on the same company — one from public information, one from confidential materials. Multiply by your pipeline.