Assumptions Log
Project: Pickles GmbH — AI Governance Framework Version: v1 Date: 2026-02-28
About This Document
This framework was built using a fictional company — Pickles GmbH — as the demonstration vehicle. Because no real company data was available during the build, every claim about Pickles GmbH's systems, operations, or characteristics is an assumption, not a verified fact.
This log records all of those assumptions in one place. Every document in the framework also flags its own assumptions inline using the marker [ASSUMPTION].
If you are adapting this framework for a real organisation: this log is your starting point. Work through every assumption listed here and validate each one against your actual company data before treating any framework document as complete.
Status Key
| Status | Meaning |
|---|---|
| 🔴 Unverified | Not confirmed against real data — placeholder only |
| 🟡 Partial | Partially confirmed — still needs full verification |
| 🟢 Confirmed | Verified against real company data |
Company Profile Assumptions
| # | Assumption | Source Document | Status |
|---|---|---|---|
| A-001 | The company provides legal AI tools covering drafting, research, summarisation, and analysis | Framework outline | 🔴 Unverified |
| A-002 | Clients are German legal professionals or in-house legal departments | Framework outline | 🔴 Unverified |
| A-003 | The company operates under the EU AI Act, GDPR, and German professional confidentiality standards (BRAK) | Framework outline | 🔴 Unverified |
| A-004 | Third-party AI model providers may be in use | Framework outline | 🔴 Unverified |
| A-005 | Hosting location is EU-based | Framework outline | 🔴 Unverified |
System-Level Assumptions (Phase 2 — Vertrag.AI)
These assumptions relate to the fictional Vertrag.AI system used in the Phase 2 worked example.
| # | Assumption | Source Document | Status |
|---|---|---|---|
| A-006 | EU hosting confirmed; specific cloud provider not named | P2-Worked-Example-System-Profile-v1.md | 🔴 Unverified |
| A-007 | Anthropic DPA in place; prompts not used for model training | P2-Worked-Example-System-Profile-v1.md | 🔴 Unverified |
| A-008 | No fine-tuning applied to base model | P2-Worked-Example-System-Profile-v1.md | 🔴 Unverified |
| A-009 | RAG corpus quarterly review cycle | P2-Worked-Example-System-Profile-v1.md | 🔴 Unverified |
| A-010 | Output labelling and UI disclaimer behaviour as described | P2-Worked-Example-System-Profile-v1.md | 🔴 Unverified |
| A-011 | Audit trail logging per session | P2-Worked-Example-System-Profile-v1.md | 🔴 Unverified |
| A-012 | EU AI Act classification confirmed as limited risk | P2-Worked-Example-System-Profile-v1.md | 🔴 Unverified — requires qualified lawyer review |
| A-013 | BRAK obligations assessed and addressed in product design | P2-Worked-Example-System-Profile-v1.md | 🔴 Unverified |
How to Resolve Assumptions
For each assumption, the validation process requires:
- Identifying the relevant data source (CTO, Head of Engineering, Legal Counsel, etc.)
- Obtaining written confirmation of the actual position
- Updating the relevant framework document to replace the assumed value with the confirmed value
- Removing the
[ASSUMPTION]flag from that field - Updating this log with status 🟢 Confirmed and the date confirmed
Some assumptions (particularly A-012 regarding EU AI Act risk classification) require review by a qualified lawyer, not just internal verification.
Resolved Assumptions
None — this framework was built entirely on fictional company data. All assumptions remain unverified. Confirm against real company data before operational use.
Last updated: 2026-02-28. This log should be maintained and updated whenever the framework is adapted for a real organisation.