See, control and prove AI in your company.
Palveron shows where AI is used in your company, checks requests from applications and agents and, in selected AI services, your employees' prompts before data leaves the organization, and records every decision in a verifiable way.
Four tasks, one platform
AI has arrived in companies. Control often has not.
AI is already in use
Employees work with ChatGPT, Claude or Copilot, often without rules and unnoticed by IT.
Data leaves the organization
Customer data, addresses and credentials end up in prompts before anyone can step in.
Agents act on their own
Agents call models and tools without a person approving each action.
Proof is becoming mandatory
The EU AI Act requires you to show how you control AI in your company.
EU AI Act: what applies already and what is still to come
Fines up to €35 million or 7% of worldwide annual turnover, whichever is higher (Art. 99)- Feb 2, 2025AI literacyArt. 4
- Aug 2, 2026Transparency obligationsArt. 50
- Dec 2, 2027High-risk AI systemsAnnex III
- Aug 2, 2028High-risk AI in productsAnnex I
One control point for every use of AI
Applications and agents call AI through Palveron, employees use selected AI services with the Browser Guard. Every request on these paths is checked before it leaves your company.
PalveronSee, control, prove
Know which AI is used
The Browser Guard detects when a known AI service is opened, and with Discover it also finds AI extensions in the browser. You decide for each service what is allowed.
- No setup per service
- One status per service: sanctioned, pending review, unsanctioned
- With Discover, new services appear in the app for review
Rules that act before anything happens
Baseline protection and your own rules apply before a request reaches the model: pass, mask, ask, redirect or block.
- Baseline protection for credentials, payment and contact data
- Your own rules per project
- Human approval where it matters
Verifiable on the blockchain
Every decision is stored as a record with an integrity value, a digital fingerprint. When anchoring is active, the integrity values are anchored on the Flare blockchain. Anyone with a record's ID or integrity value can check, without an account, whether it is there; the check shows no content. This produces a report for the EU AI Act and controls for ten additional frameworks.
- Integrity value for every record
- Anchoring on the Flare blockchain
- Verifiable with ID or integrity value under "Verify evidence"
Integrations and frameworks
Connect
SDKs, a proxy and adapters for the tools your teams already use.
- TypeScript
- Python
- Go
- Java
- Proxy
- MCP
- LangChain
- CrewAI
- n8n
- Make
- Cursor
- Windsurf
- Claude Code
- OpenClaw
Map
Controls for 11 frameworks and a report for the EU AI Act.
- EU AI Act
- GDPR
- DORA
- NIST AI RMF
- ISO 27001
- ISO 42001
- SOC 2
- HIPAA
- OWASP Agentic
- Singapore PDPA
- Colorado AI Act
Trust you can verify
Every decision leaves evidence that you can check with its ID or integrity value.
- Decisionpass, mask, ask, block
- Recordstored with an integrity value
- Batchmany integrity values, one root
- Flare blockchainthe root is anchored on the blockchain
- Verifiable with ID or integrity valueunder "Verify evidence"
Verifiable with ID or integrity value
Every record carries an integrity value. When anchoring is active, it is placed on the Flare blockchain in a batch. Anyone with the record's ID or integrity value can check under "Verify evidence" whether it is anchored there; the check shows no content.
Servers and database in the EU
Servers in Germany, database in Frankfurt. Every service provider is listed in the Trust Center with its domicile and data location.
For data protection officers and works councils
The Browser Guard sends neither the name nor the email address of the person using it. We disclose on its page what it records and what it does not.
Open about limits
We state openly what Palveron does not detect, right in the app.
Frequently asked questions
Your question is not listed? Talk to us
What does the Browser Guard record about employees?
Neither the name nor the email address of the person using it. It records the AI services visited, checked prompts as an encrypted record, and the decision. What exactly is recorded and what is not is set out on the Browser Guard page.
Where is our data stored?
The servers are in Germany, the database is in Frankfurt. You find every service provider with its domicile and data location in the Trust Center.
How does Palveron differ from DLP or Purview?
Palveron is built for the use of AI: it checks prompts before they reach an AI model, governs the tool calls of agents through MCP and stores every decision as a record with an integrity value, which the Compliance Hub maps to the controls of frameworks such as the EU AI Act.
How is Palveron set up?
You connect applications and agents through an SDK, the proxy or MCP; the API key of your project is all it takes. For employees in the browser, you install the Browser Guard and enter the gateway address and your project's API key there, by hand or centrally through managed browser settings.
What does Palveron not detect?
Palveron does not check prompts in browsers without the Browser Guard or in AI services it does not know. The Browser Guard does not check the content of attached files today. If it does not recognize a send, it reports it instead of letting it through silently.
Do we need developers for this?
Not for the Browser Guard and the overview of AI services in use: you set up both in the dashboard and in the extension. Connecting your own applications and agents through an SDK, the proxy or MCP takes someone who adjusts configuration or code; the guides are in the documentation.
Get to know Palveron
In a demo, we walk through your own use cases and show how Palveron makes AI use visible, checks prompts and produces evidence.