About
Built for the documents
that break generic tools.
Dokko is the document AI platform from Mono Software. It exists because most AI tools were built for simple documents — and most of the documents that actually matter to a business are not simple.
The thesis
Document AI broke because retrieval broke.
The popular story about AI in business is that the models aren't good enough yet. Our experience disagrees. The models are good. The retrieval is broken.
Most AI platforms treat documents as a bag of fragments to embed and search — a strategy that works on short, simple content and breaks on dense, cross-referenced documents where one clause only makes sense in the context of the whole. Legal contracts. Compliance frameworks. Technical manuals. Regulatory archives. The documents that decide outcomes.
"Almost right" is indistinguishable from wrong when the document is a 300-page contract. Sometimes it's more dangerous.
Dokko was built to solve the retrieval problem first — by reading every document in full before indexing it, cross-referencing within and across the library, and citing the source documents on every answer. Once the answers are reliable, action becomes the natural next step: Skills inside the conversation, Agents across the stack.
Deployment
One engine.
Two deployment shapes.
Most document AI platforms picked a side. Either they were built for internal teams (an enterprise search box for the legal department) or for external customers (a chat widget for the support page). Dokko was built for both — on the same engine, the same subscription, the same configuration model. The legal team's internal assistant and the company's customer-facing assistant share document understanding, security posture, and dashboard. The teams who configure them are different. The audiences are different. The engine isn't.
What we believe
Three commitments.
/01
Augmentation, not replacement
We don't think AI should replace experts. We think AI should make experts faster and more accurate — by handling the volume so the human can focus on the edge. Every Dokko surface is built around human-in-the-loop. The system knows its limits and escalates when it should.
/02
Transparency over lock-in
Dokko publishes its rates. Caps consumption. Previews every import. Your data exports. Your knowledge bases are yours. We make leaving straightforward because making it hard is how vendors signal they're not confident in their product. We are confident in ours.
/03
The team that uses it owns it
The legal team configures its assistant. Compliance owns theirs. Operations runs its own. IT supports the platform; the teams configure the work. We don't sell tools that need a six-month deployment project to start delivering value.
Who we are
From Mono Software.
Mono is a Croatian software company. We've spent years building production AI systems for clients in legal, compliance, manufacturing, and healthcare — the sectors where document complexity decides outcomes. Dokko is the platform that grew out of that work. The contextual retrieval architecture, the Skill and Agent layers, the per-tenant cost tracking, the calibrated confidence detection — they exist because we needed them in production work, not because they sounded good in a deck.
We build for the EU AI Act era because we live in it. The auditability posture in Dokko is the same one we'd want a regulator to ask us about.
Where we're going
What's next.
The roadmap is shaped by what production customers ask for.
Near-term
Broader coverage of the integration library, deeper Agent composition surfaces, a growing library of pre-built Skills.
Mid-term
The configurable guardrails layer that makes Dokko's behaviour as auditable as its answers.
Long-term
The document AI engine that serves every team and every customer surface in an organisation — on one subscription, with one configuration model.
Get started
Read deep. Act fast.
On your documents.
The 14-day trial runs on your real library, not a sandbox. Bring your contracts, manuals, or frameworks — and see how Dokko reads them.