Full-stack AI governance
Purview governs your Microsoft AI. PromptKing governs everything else.
Together, they cover your full AI governance stack.
What Microsoft Purview covers
Microsoft Purview provides deep governance within the Microsoft boundary — excellent for organisations standardised on Microsoft AI.
- Microsoft 365 Copilot
- Copilot Studio agents
- Azure OpenAI
- GitHub Copilot (Microsoft umbrella)
If your organisation runs exclusively on Microsoft AI, Purview is excellent.
What falls outside Purview's scope
| Vendor / deployment | Purview status |
|---|---|
| Anthropic Claude (direct) | No Purview coverage |
| Google Gemini (direct, Vertex AI) | No Purview coverage |
| AWS Bedrock (all models) | No Purview coverage |
| IBM Watsonx | No Purview coverage |
| xAI Grok | No Purview coverage |
| Meta Llama (self-hosted) | No Purview coverage |
| Custom and fine-tuned models | No Purview coverage |
The average enterprise runs 6–8 AI vendors. Purview covers 2–3.
What the gap looks like in practice
A financial services firm using M365 Copilot, Claude for credit assessment, and Llama on Bedrock for document processing has Purview coverage for Copilot and zero governance tooling for the other two — which happen to be Annex III high-risk use cases under the EU AI Act.
How PromptKing fills the gap
Same Article 26 evidence model as Purview. Same FOCUS 1.4 export standard. Covering the vendors Purview cannot reach. Not competing — complementing.
- Cross-vendor seat intelligence — ghost seats, utilisation, rightsizing
- EU AI Act Annex III classification and Article 26 Deployer Evidence Packs
- Connector confidence grading (A–D) — no silent mixing of data quality
- FOCUS 1.4 conformant export — unifies with Purview in Power BI or Snowflake
- GPAI vendor compliance matrix — CoP status for Claude, Gemini, Grok, Llama
The architecture
Microsoft stack
M365 Copilot · Azure OpenAI · Copilot Studio · GitHub Copilot
↓
Microsoft Purview
↓
Compliance Manager · DLP · audit logs
Non-Microsoft AI
Claude · Gemini · Bedrock · Watsonx · Grok · Llama · custom
↓
PromptKing
↓
Article 26 Evidence Pack · FOCUS export · EU registry
PromptKing runs above telemetry, not inside it. LangSmith tells you how a prompt executed. OpenTelemetry tells you how a system performed. PromptKing tells you what it cost, whether you can trust it, and whether you're compliant.
See what PromptKing covers that Purview can't
Who in your organisation actually uses prompt traces to make a business decision? Prompt tracing tells you what happened. PromptKing tells you what it meant.