AI and clear processes. Working together.
UNOY combines AI with executable rules, knowledge and approvals. Your workforce can take on work while relevant decisions and results remain open to review.
Workflows preserve processes. Skills and databases hold knowledge. COIA coordinates execution. Approvals and records of work keep results open to review.
Master Formula
Work in. Result out.
§ 02 · The AI Gap
What turns a task into a process.
Work requires suitable information, permissions, responsibilities and defined completion criteria. These requirements are configured for the workforce.
AI Response · no trace
Notice on Data Protection Request
AI does not know who may decide.
No responsibility assigned, no mandatory check, no audit trail. An answer is produced regardless of who is accountable.
! Record missing
Contract Müller GmbH
Handler: ?
Status: ?
AI does not know your files.
Standard models have general knowledge, not your contract, your policy, your member data.
Model Versions
AI changes silently.
Models are replaced, prompts drift, outputs vary. Without a defined process, results are hard to compare.
AI alone responds. Robust expert work decides, documents and hands over.
§ 03 · The Rule Gap
Where AI supports execution.
AI can interpret different documents, prepare information and propose next steps. Workflows define required checks, handovers and approvals.
RPA Workflow · Run #2451
Email Classification
Rules do not understand language.
Classification, meaning extraction and language comprehension are hard to cover with fixed if-then rules alone, for example with unstructured emails.
Notice · #1287
Approved
Reasoning: [empty]
Oversight: ?
Rules justify nothing.
They execute but do not explain. Clients, regulators and audits need justifications, not code paths.
Rule Maintenance · Q2/2026
Claims Classification
Rules break under special cases.
Every new variant requires a new rule. Maintenance effort grows with every exception.
Pure rules execute. Robust expert work reads, writes and reasons.
§ 04 · The UNOY Synthesis
AI alone is too free.
Pure rules are too rigid.
UNOY combines both.
Accountable expert work does not arise from a single tool. It arises when four layers work together: rules carry the logic. Knowledge carries the content. AI supports interpreting, preparing and drafting. Governance defines responsibilities and approvals. UNOY OS connects these four layers in one system.
AI Alone
Free. Fast. Non-binding.
Pure Rules
Stable. Auditable. Mute.
UNOY
Free. Stable. Accountable.
§ 05 · Layer 1 · Rules
Reuse established processes for further cases.
Workflows preserve agreed steps and rules. This supports reuse and review. A deterministic process alone guarantees neither correct inputs nor a correct professional result.
Matter · GDPR Access Request
From letter to verified response.
Step sequence, mandatory fields, sources and approvals are defined in the workflow before a matter starts. Once the access request arrives, it runs against this definition: what must be checked, who approves, what is recorded in the audit trail. AI supports where documents are interpreted and texts are drafted.
Model interchangeable, process defined. A model change requires professional regression tests.
§ 06 · Layer 2 · Knowledge
Skills, RAG and databases: your knowledge.
Standard models know the world. UNOY brings your world into play, in three ways: as reusable skills, as citable RAG sources, as structured facts.
Skill Library
Skills: reusable domain logic.
Defined once, usable in suitable further matters. Versioned and auditable.
RAG Sources · § 4 Access
RAG: citable knowledge base.
Contracts, legal texts, internal memos: directly in the matter with source reference. Answers refer to the sources used.
Member DB · sample data
Databases: structured facts.
Members, policies, deadlines, order data: readable, writable and verifiable inside the workflow. Structured data reduces errors but does not replace professional review.
§ 07 · Layer 3 · AI Within Bounds
AI helps shape the work. COIA coordinates execution.
Claude can operate UNOY and COIA and help build and improve workflows. COIA uses approved capabilities and returns results. People define goals, permissions and required approvals.
Matter · Orchestration
Clearly separated: what AI does, what the workflow does.
AI supports classifying, drafting and reasoning. Mandatory checks, roles, records of work and handovers are defined by the workflow.
A model change does not alter the defined process, but requires professional regression tests.
People define goals, permissions and required approvals.
§ 08 · Layer 4 · Governance
Governance turns AI output into a decision someone can be accountable for.
Traceability is part of the matter. Steps, sources and approvals are recorded within the agreed scope.
Roles · Matter #2451
Roles & approvals.
Who may view, edit, approve? Four-eyes principle can be set up in the workflow.
Audit-Trail
Audit-Trail.
Steps with timestamp, actor, source and model version. Traceable.
Data Residency
EU data residency.
Data in EU data centres. Encrypted at rest and in transit. No training use.
Learning Loop · Q2/2026
Learning loop.
Corrections flow into skills and templates: controlled, with version record.
§ 09 · Proof in the Matter
Evaluate quality through actual results.
Measure accepted completions, errors and total human effort. What matters is whether the workforce demonstrably reduces your workload.
WKO · live
600,000+ businesses
8 applications in production
since 2024
WKO: Many services. One system.
Chamber scale with workflow core. See case study →
FairPlane · Fall #2451
Delay 3h 20min
EU 261/2004 · € 600
✓ Claim confirmedFairPlane: Many cases. Clear rules.
Classification, review, letter. AI within the rule. See case study →
Skribe · Accountability
Skribe Rechtsanwälte
Legal service
where agreed
Skribe: UNOY handles the work. Skribe provides legal services.
Legal services within Service, where required and agreed. See service page →
§ 11 · Next Step
Work in. Result out.
Bring recurring work. Together we review which building blocks apply and which approach fits.