EVIDION is the fastest-compounding sovereign intelligence platform in Europe, and DGE3 is doubling down

Every fund keeps a small list of positions it wishes it had been able to write a larger check into. For DGE3 Investment, EVIDION has moved onto that list faster than any company we have financed in the last five years. We committed $850,000 into the pre-seed of EVIDION at a point when the platform was a working system rather than a pitch, and in the months since, the operating trajectory has outrun the plan we underwrote against so decisively that we are now preparing a follow-on round of $10 million alongside institutional co-investors and strategic partners from the European defense and security sector. This article explains what EVIDION actually is, why the growth curve looks the way it does, and why the founder behind it is the single strongest reason we are increasing our exposure rather than trimming it.

The starting point is a problem that every intelligence analyst, prosecutor, financial-crime investigator and corporate compliance officer in Europe recognizes immediately, and that almost no software vendor has solved end to end. A single case today consists of several thousand pages of case files, a dozen or more separate data sources, and a handful of formats that nobody reads together. There are volumes of documents, protocols, expert opinions and attachments in multiple versions. There are call detail records and bank statements running to tens of thousands of rows with no case context attached to any of them. There are hours of recordings that nobody will ever listen to twice. The analyst has access to everything and to nothing at the same time, because the material sits in separate systems and the only place it ever actually meets is inside that person's memory. EVIDION exists to be the layer where it meets instead.

One case model across eight intelligence domains, with every finding traceable to its source

EVIDION describes itself in deliberately unglamorous terms: you make the decisions, EVIDION sees the connections. Underneath that line is an analytical core that fuses material from eight domains into one case model, covering communications analysis, financial analysis, spatial analysis, open sources, cybersecurity, digital forensics, modeling and prediction, and a dedicated legal layer. The capability list reads like a catalog of the things investigators currently buy from five different vendors and stitch together by hand: call detail record analysis, cell-tower triangulation, money laundering detection, cryptocurrency tracing, geographic profiling, link graphs, stylometry, assessment of testimony credibility, causal inference, strategic modeling, evidentiary integrity, and generation of procedural documents. The platform is deployed as selected modules rather than as a monolith, so an institution can start with one domain and expand without downtime.

The pipeline runs in three stages, and the discipline in how those stages are designed is the part that convinced our investment committee. In the acquisition and structuring stage, the system ingests case files, recordings, call records and statements across more than thirty input formats, performs text recognition with two engines in parallel, transcribes audio, and extracts entities: people, companies, accounts, phone numbers, amounts and dates. It resolves identity natively for Polish naming conventions, so that three different renderings of the same surname collapse into a single subject, and it deduplicates the same document appearing across multiple volumes. It recognizes national identifiers, PESEL, NIP, KRS and bank account numbers, which sounds trivial until you have watched a foreign platform fail to parse a Polish court filing. In the correlation and inference stage, the domains begin to intersect: who spoke to whom, where they were, what funds moved, and what follows from the combination. And through all of it, one hundred percent of findings carry a reference back to the source document they derive from, which is the difference between a tool that produces an interesting hypothesis and a tool whose output can survive a courtroom.

That evidentiary discipline is not a feature bolted on for marketing. The system maintains an append-only operations log where every action is recorded with user, time, scope and result, and where entries cannot be modified after the fact. It isolates data between proceedings, with access restricted to the assigned case and enforced at the database layer rather than only in the interface. It uses checksums and a hash tree so that an institution can demonstrate a document has not been altered since the moment it was secured. Permissions are differentiated across five separate levels with operation limits. EVIDION is also unusually candid in public about what is shipped versus what is planned: the security page marks permission levels, data isolation and evidentiary integrity as implemented in version 0.5, and two-factor authentication, fully disconnected operation and formal certification as roadmap items ahead of production deployment. In a category where vendors routinely present roadmap as reality, publishing that distinction and inviting institutions to verify the implemented items on a live demonstration is a credibility signal we weight heavily.

The sovereignty argument that no American, British or Israeli vendor can answer

EVIDION's positioning against the incumbent landscape is where the investment thesis becomes structural rather than merely competitive. The company's own comparison, drawn from publicly available vendor documentation, is careful to note that it concerns functional scope and not quality, and that many incumbent tools remain the best available in their narrow specializations. The point is that none of them is the connecting layer. Data fusion platforms are supplied from the United States. Diagramming tools come from the United States and the United Kingdom. Mobile forensics is dominated by Israel. Open-source intelligence tooling largely originates in Germany. EVIDION is supplied from Poland. On native processing of the Polish language, every incumbent category answers no and EVIDION answers yes. On a layer encoding Polish criminal procedure, every incumbent answers no and EVIDION answers yes. On automatic fusion of signals, financial and geospatial intelligence, on strategic modeling, on causal inference, on generation of procedural documents, the pattern repeats. On operating without external connectivity, incumbents range from limited to partial to no, while EVIDION has it in the architecture.

For a European institution handling classified material, that column matters more each year. Evidence that leaves the institution's infrastructure is a permanent risk, and dependency on a vendor controlled from another jurisdiction is exceptionally difficult to reverse once it is established. EVIDION's deployment model answers this directly: the database, indexes and operations log sit on the client's own servers inside the institution's network; the model layer runs locally so that inference does not depend on an external supplier; and the deployment mode itself, isolated, local or managed, is the buyer's decision rather than the vendor's. Data is held in PostgreSQL with export to open formats, code escrow is a contractual matter, and the institution retains the ability to maintain the system independently. The inference layer is EVIDION's own proprietary model set, tuned to Polish criminal and procedural law, to the structure of official documents, and to operational analysis methodology, and it is deliberately separated from the analytical modules. That architecture is why the platform can be sold into environments that a cloud-only competitor is structurally barred from serving.

Six deployment profiles on one core, and a market that is expanding faster than the supply of qualified vendors

Commercially, EVIDION runs one platform with six configured deployment profiles, spanning special services doing counterintelligence, threat assessment and multi-source fusion in isolated environments; the justice system working through voluminous case files, evidence assessment and drafts of procedural filings; investigative services mapping organized crime, group hierarchy, billing and location analysis; financial authorities working on anti-money-laundering, cross-border flows and cryptocurrency; law firms preparing procedural positions and assessing the opposing party's material; and the financial sector and enterprises handling AML obligations, internal fraud investigations and counterparty verification. The scope of activated modules, the permission model and the output format change per profile while the analytical core stays constant. From a fund's perspective, that is the shape you want in a deep-tech position: one engineering investment amortized across a public-sector market with long contract cycles and a commercial market with faster ones, so that regulatory tailwinds and enterprise budgets are both accessible without a second product build.

The demand side is moving in EVIDION's favor on every axis we track. European governments have been accelerating the reduction of dependency on non-EU technology in defense and intelligence since 2022, Polish cybersecurity budgets have been growing at roughly twenty-eight percent year over year, and NIS2 together with the AI Act has created new compliance obligations for over one hundred and sixty thousand entities across the European Union. The EU gov-tech AI market is running at a projected $4.2 billion in 2026. Meanwhile the price anchor set by the incumbents, where a Gotham-class deployment runs from tens to hundreds of millions of dollars annually with implementation measured in twelve to twenty-four months, leaves an enormous band of agencies, prosecutors' offices, financial institutions and critical-infrastructure operators who need the capability, are legally obliged to have it, and could never justify the incumbent contract. EVIDION enters that band at a small fraction of the cost with implementation measured in weeks, and its go-to-market motion, a closed forty-five-minute briefing on illustrative material, on site or over a channel the institution nominates, with a technical review and an anonymized-material trial available as extensions, is precisely calibrated to how procurement in this sector actually starts. Inbound from the security sector is handled on priority with a forty-eight-hour response commitment.

Why $850,000 became a $10 million conversation

We wrote the pre-seed check because the platform was already real. At the point of our commitment, EVIDION was a production system at version 0.5 with more than two hundred and thirty operational analytical modules, six hundred and twenty-six API endpoints and roughly seventy-six thousand five hundred lines of production backend code, with OpenAPI throughout and REST as the integration surface. What has happened since is the part that changed the size of the number we are willing to put behind it. The platform has been consolidated into a coherent eight-domain case model with a single source-referenced output standard, the security posture has been documented to a level institutional buyers can audit, the deployment story has been reduced from a technical argument to a procurement-ready set of three options, and the commercial surface has been sharpened from a broad govtech pitch into six named profiles with explicit deliverables per profile. Each of those is a de-risking event. Taken together inside a single funding cycle, they represent the execution velocity of a team several times EVIDION's size.

So we are being explicit about the direction of travel. The $850,000 pre-seed was priced as first institutional capital into an unproven category entrant. The $10 million round now being prepared is priced against a platform with a validated architecture, a defensible sovereignty moat, a regulatory calendar working in its favor and a founder who has already demonstrated, in a different category, that he converts capital into compounding product at an unusual rate. The capital is earmarked for hardening toward the v1.0 enterprise release, completing the security certification path including the ISO 27001 and NATO accreditation processes, building a gov-tech sales organization with direct relationships inside the Polish defense and security establishment, and opening the NIS2 compliance offering into DACH and the Nordics. Investors, growth funds, corporate venture arms and strategic partners in the defense, security and regulated-data sectors who want exposure to European sovereign intelligence infrastructure at this stage should write to us at invest@vcdge3.com.

EVIDION is the company that should already have existed in Europe and did not. Sovereign by architecture rather than by marketing, modular where the incumbents are monolithic, and native to the legal system it serves in a way no foreign platform can retrofit. We put in the first institutional capital at $850,000. We are preparing to put in $10 million more, because the execution has outpaced our own underwriting model in every dimension we measure. - Dmitri Evanov, Managing Partner, DGE3 Investment

The founder: Mateusz Ulewicz has already proven he ships

None of the above would justify a follow-on of this size without the person behind it, and this is where we want to be unambiguous, because the founder is the reason our conviction is asymmetric. Mateusz Ulewicz is EVIDION's Founder and CTO, and he is an engineer and software architect with twenty years of experience building systems at global scale, spanning distributed systems, AI and machine learning, natural language processing, retrieval-augmented generation, multi-agent systems, graph theory and cybersecurity. His career runs through Nokia, where he worked on telecommunications systems operating at global scale, through Amazon Web Services, where the infrastructure he worked on serves millions of companies, and through the government sector, where the requirements on code quality, security and auditability are as demanding as any environment in the industry. That combination, telecommunications plus cloud plus government-grade security plus AI plus criminal law and forensics, in one head, is why EVIDION's scope is designable at all. Most teams could not specify this product, let alone build it.

Before EVIDION, Mateusz built FOTOhub.app, and that is the track record we underwrote. DGE3 has committed $10.3 million into FOTOhub, our largest position to date, and the platform has continued to compound: past 750,000 registered users in under a year from launch, with the model catalog grown to 226 active AI models across image, video, audio, speech, text, 3D and virtual try-on, exposed through 267 documented REST endpoints, a full production consolidation onto AWS across five regions on three continents, acceptance into the AWS Startup Portfolio Tier, a direct technical and commercial partnership with BytePlus, ByteDance's enterprise cloud division, MRR up 4.2x since the Seed close, paid conversion at 9.5% against a category benchmark of three to five percent, net revenue retention above 145%, week-eight retention at 62% against an AI-tools benchmark of thirty-one to thirty-eight percent, and a debut on the Crunchbase Heat ranking with a Growth Score of 93 and a Heat Score of 95. Mateusz himself has been ranked in the top ten of Crunchbase's global person ranking out of more than 2.1 million tracked founders, executives and innovators, at a position ahead of names that define the current era of artificial intelligence.

The most instructive data point in that history is not a growth number at all. In April 2026, at the steepest part of FOTOhub's curve, Mateusz took a two-month medical leave. For most founder-dependent companies at that stage, a two-month absence at peak growth is an extinction-level event. FOTOhub did not merely hold; it accelerated, moving from 600,000 to 750,000 users, growing revenue 4.2x and expanding the team from fourteen to twenty-four people while the architect was away. That is what a well-built system looks like from the outside, and it is the single best available evidence that what Mateusz builds does not depend on Mateusz being in the room. When we assess the execution risk on EVIDION, that is the prior we are working from.

So the summary of our position is straightforward. EVIDION is a genuinely brilliant idea, sitting in a market that European regulation and geopolitics are expanding faster than qualified vendors can enter it, with a sovereignty moat that competitors cannot retrofit and a legal-language capability that nobody else has built. It is being executed by a top-tier CTO who has already delivered a proven, scaled, revenue-generating platform once, in a completely different domain, and who has demonstrated that his systems keep compounding even when he steps away from them. We backed that combination with $850,000 when it was early. We are preparing to back it with $10 million now that it is not. Very few positions in a fund's life earn a follow-on this quickly, and we would rather be early and concentrated in this one than correct and small.

Learn more about EVIDION at evidion.pl and on Crunchbase. Institutions can request a closed briefing directly through the site; investors evaluating this thesis can reach out or write to invest@vcdge3.com.

#EVIDION #MateuszUlewicz #DGE3 #DGE3Investment #SovereignAI #DefenseTech #GovTech #Cybersecurity #Intelligence #Palantir #IntelligenceFusion #LegalAI #OSINT #SIGINT #FININT #GEOINT #AML #NIS2 #AIAct #DigitalSovereignty #NationalSecurity #DeepTech #DualUse #PreSeed #SeriesA #VentureCapital #Hypergrowth #Crunchbase #CrunchbaseRank #FOTOhub #PolandTech #EuropeanTech #AI #MachineLearning #StartupFunding #FounderCTO