Competitive landscape
Five categories of tool occupy adjacent territory. Each solves part of the problem investors face. None solves it the way classical commercial DD does — at depth, on both pre-engagement and in-engagement evidence, at minutes-not-weeks compression.
The structural argument
An investor evaluating a company looks through two lenses, often without naming them as separate. Most of the market covers one or the other. EviDimensional covers both, with the same framework, at the same depth.
Lens 1
Long list to short list. The company has not opened its data room. The investor's team is working from public sources only — website, registries, LinkedIn, news, published materials.
This is where most deals are decided. It is also where pitch quality currently substitutes for company quality — a proxy that research has shown to be false. Bessemer's published anti-portfolio — Apple, Google, Facebook, Airbnb, PayPal, Tesla, eBay, FedEx, all passed on — were screened out at this lens, not in due diligence.
Public-Source DDLens 2
Short list to investment committee. The company has opened its data room. Pitch deck, financial model, customer pipeline, technical documents, contracts.
Classical commercial DD synthesises this material into an investment-grade view: dimensions, evidence quality, valuation range, readiness against the target round. EYs and McKinseys deliver this in weeks at €25,000 or more. The desk-synthesis layer of that work is what compresses cleanly — not the partner judgment, the pattern application.
Confidential-Source DDThe market
Five existing categories of product touch part of what an investor needs at one or both lenses. None covers the structural gap that opens when the two lenses are taken together.
Category 1
EY · McKinsey · KPMG · BCG · Bain · boutique commercial DD firms
The Big Four and tier-one strategy houses produce serious commercial due diligence at depth. An analyst team spends two to four weeks synthesising the data room into an investment view. The output is the gold standard the rest of the market is implicitly compared against.
What they do well: in-engagement depth, partner-grade judgment, defensible methodology.
What they don't do: pre-engagement screening (the engagement model assumes you've already engaged), volume (one company at a time), speed (weeks not minutes), price (€25,000 minimum). The desk-synthesis layer that takes most of the analyst time is exactly the layer that compresses.
No pre-engagement · No compression · No volumeCategory 2
Dealroom · PitchBook · Crunchbase · Tracxn · Affinity
Aggregate company data, fundraising history, team composition, sector tagging, contact records. The plumbing of deal sourcing and pipeline tracking. They tell you that a company exists and what is publicly knowable about its surface. They do not produce assessment.
What they do well: discovery, search, pipeline management, signal detection across volume.
What they don't do: any form of structured assessment. They surface what's been observed; they do not evaluate what it means. A reader of a Dealroom record knows the company raised a Series A from named investors. They do not know whether that round was warranted by the evidence the company holds.
Data, not assessmentCategory 3
ValidatorAI · DimeADozen · GoldenEggCheck · IdeaProof · PitchBob · Founderpal
Founder-facing tools that take a description, an idea, or a pitch deck and produce some scoring or feedback. Fast, cheap, frequently free. A founder pastes their idea into a chat interface and gets a structured response back about market size, competitive landscape, and what they should think about next.
What they do well: founder-facing speed, low friction, idea-stage feedback.
What they don't do: distinguish what a founder knows from what they believe. The system scores what the founder told it; there is no evidence framework, no E0–E5 grading, no methodology corpus, no third-party validation. A polished pitch produces a confident assessment regardless of whether the underlying claims hold up. The output reads like an investment memo and validates exactly the wrong thing.
No methodology · No evidence grading · Founder-facing, not investor-facingCategory 4
Keye · V7 Go · CENTRL · Vantager · ToltIQ · DealPotential
Process data rooms, contracts, and financials with LLM-assisted extraction. Built for late-stage M&A and PE due diligence where the documents already exist and the question is what they reveal. Strong on contract review, financial-statement analysis, and structured-data extraction.
What they do well: confidential-document processing at scale, contract clause flagging, financial extraction.
What they don't do: work pre-engagement — their entire workflow assumes a data room exists. Useless on a long-list company you have not yet engaged. Also: methodology-light. Strong on processing inputs, weak on producing the methodologically-graded synthesis investors actually need to make a call.
Data room only · No pre-engagement · No structured assessmentCategory 5
ChatGPT · Claude · Gemini · Perplexity — used directly
The simplest competitor and the most underestimated. An investor pastes a deck or a company description into ChatGPT or Claude and asks for an assessment. Fast, free, and increasingly capable. Many investors do exactly this today.
What they do well: general intelligence applied to whatever you paste in. Reasonable prose. Surface-level synthesis.
What they don't do: the LLM has no methodology corpus, no evidence framework, no valuation engine, no readiness gate logic, no tarpit-pattern library. It produces fluent-sounding text that is methodologically generic. Two pastes of the same materials produce different outputs. Pasting confidential materials directly into a consumer chat interface also creates a privacy posture most investors would not want documented.
No methodology corpus · No structure · No evidence frameworkAt a glance
The matrix below maps six load-bearing capabilities against each category. Only one row checks every column.
| Pre-engagement | In-engagement | Evidence-graded | Valuation engine | Compression | Methodology corpus | |
|---|---|---|---|---|---|---|
| EviDimensional | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
| Manual analyst engagements | − | ✓ | ∂ | ✓ | − | ✓ |
| VC data platforms | − | − | − | − | − | − |
| AI startup validators | ∂ | − | − | − | ✓ | − |
| Document AI & data room tools | − | ✓ | − | − | ∂ | − |
| Direct frontier LLM use | ∂ | ∂ | − | − | ✓ | − |
✓ capability fully present · ∂ partial or limited · − not present
Pre-engagement: works on a company before you have confidential access. In-engagement: uses confidential materials when the data room is open. Evidence-graded: distinguishes founder assertion from third-party validation (E0–E5 or equivalent). Valuation engine: produces a defensible valuation range, not a single number. Compression: faster than an analyst team. Methodology corpus: encoded structured framework underlying the output.
Why the gap exists
Compressing the desk-synthesis layer of classical commercial DD requires two things working together: a structured assessment methodology with evidence grading already encoded, and a frontier LLM capable of holding the methodology and applying it consistently across many dimensions at once. Neither piece on its own is sufficient.
Most existing assessment products predate frontier LLMs. They have methodology but cannot scale — the partner-leverage model that produced the methodology also caps how many companies it can assess in a year. The Big Four can't productize the desk-synthesis layer because their margin structure depends on partner leverage, not on productization. Productizing destroys the model.
Most LLM products skip methodology. They are general-purpose intelligence applied to whatever the user pastes in — powerful, but methodologically generic. Direct LLM use can't add structured methodology because that would be the product itself, not a thin wrapper around a chat interface.
The specific gap covered by EviDimensional opens at the intersection: a methodology that predates LLMs, descending from ISO 9126 software-quality assessment work in 2003–2007, encoded across 1.5 million tokens, layered on top of the most powerful LLM engine available. Both pieces had to mature in that order.
The team wasn't first to recognise the gap. It was first to be in a position to fill it.
Sample assessments on a fictional DeepTech company are available, no signup required. Read the public-source DD. Read the confidential-source DD. Read both side by side. Then ask whether anything else you've seen produces the same shape of work.