Q: How is Mima aligned with the EU AI Act (Regulation 2024/1689)?

Mima is built as operational review infrastructure for the EU AI Act high-risk regime (Annex III). Architectural alignment maps to specific articles: Article 11 technical documentation (versioned rules registry, model versioning log), Article 12 record-keeping (every finding stored with source span, rule version, timestamp, named verifier — audit trail by design), Article 13 transparency (deterministic engines disclose which rule produced each finding), Article 14 human oversight (each decision routed through verify, warn, or block control signals before send), Article 15 accuracy and robustness (reproducible: same input always produces the same finding).

Q: What does 'deterministic, not generative' mean architecturally?

The language model reads input only. It does not generate findings. Findings are produced by versioned deterministic code running against a registered rule set. The same document, processed with the same rule version, produces the same finding every time — bit-for-bit. This is what makes Mima auditable, reproducible, and procurement-grade for regulated decisions.

Q: What is a 'named verifier' and why does it matter for EU AI Act Article 14?

A named verifier is the deterministic component that produced a specific finding, identifiable by version and signature. EU AI Act Article 14 requires meaningful human oversight of high-risk AI. By exposing which named verifier produced each finding, a human reviewer can interrogate the specific rule rather than an opaque model — making oversight operationally possible rather than nominal.

Q: Does Mima qualify as a high-risk AI system under EU AI Act Annex III?

Mima is deployed in contexts that EU AI Act Annex III classifies as high-risk: administration of justice and democratic processes, access to essential public services, and employment-adjacent decisions. Mima is engineered to the high-risk control set (Articles 11–15). No notified bodies are yet operational for Annex III conformity assessment as of 2026; Mima is not CE-marked. Mima provides the technical and procedural substrate customers need to pass conformity assessment when notified bodies become operational.

Q: What patents protect Mima's architecture?

Five patent applications filed with PRV (Swedish Patent and Registration Office), sole inventor Neda Modirzadeh, applicant Mima Solutions AB: MELA Multi-Engine Legal Architecture (SE 2530870-1), CLR Compliance Logic Registry (SE 2630207-5), LTE Legal Trace Engine (SE 2630208-3), PCT-Gate Procedural Control Gate (SE 2630302-4), RTDG Reverse-Traceable Decision Gate (SE 2630343-8). Together they cover the traceable, reproducible review architecture.

Q: How does Mima differ from generative legal AI tools like Harvey, Casetext, or ChatGPT?

Generative tools produce free-form text and are non-reproducible: the same prompt yields different outputs each run, and findings are not structurally linked to legal sources. Mima is the inverse — deterministic engines produce structured findings linked to specific clauses and statutes, reproducible bit-for-bit, with full audit trail by design. Generative tools are generally not classified for EU AI Act high-risk legal use; Mima is engineered specifically for that regime.

Question 1

Vad är Mima Decision Gate?

Accepted Answer

Mima Decision Gate är en deterministisk granskningsmotor som verifierar, varnar eller blockerar varje myndighets- och juridiskt beslut innan det skickas. Varje fynd kopplas till lagrum och processen är fullt spårbar enligt EU AI Act.

Question 2

Hur skiljer sig Mima från ChatGPT eller andra generativa AI-verktyg?

Accepted Answer

Mima använder deterministiska analysmotorer. Samma dokument ger alltid samma analys. Generativa modeller som ChatGPT producerar olika svar varje gång och saknar spårbarhet till källor. Mima kopplar varje fynd till en specifik klausul eller lagparagraf och är byggd för EU AI Act-kraven på transparens och mänsklig kontroll.

Question 3

Sparar Mima dokumenten som laddas upp?

Accepted Answer

Nej. Mima tillämpar noll datalagring för dokumentinnehåll. Filer bearbetas i minnet och raderas omedelbart efter analys. Endast anonymiserad metadata sparas för spårbarhet och revisionsändamål.

Question 4

Är Mima kompatibel med EU AI Act?

Accepted Answer

Ja. Mima är byggd enligt EU AI Act (Regulation 2024/1689) med full transparens, spårbarhet, mänsklig övervakning och teknisk dokumentation. Plattformen är klassificerad och dokumenterad enligt regelverkets krav för högrisksystem inom juridik och offentlig förvaltning.

Question 5

Vilka använder Mima?

Accepted Answer

Advokatbyråer, kommuner, regioner och statliga myndigheter i Sverige och övriga EU. Plattformen används för granskning av myndighetsbeslut, avtalsanalys, GDPR-pre-audit, ärendehantering och legal due diligence.

Question 6

Hur fungerar spårbarheten i Mimas analyser?

Accepted Answer

Varje fynd Mima rapporterar är länkat till en specifik mening i dokumentet och en specifik lagparagraf eller standardklausul. Detta möjliggör fullständig revision och uppfyller EU AI Act:s krav på förklarbarhet.

Samma ärende.