Dashboard in Catalan and Spanish
What happened today, who was most active, where Claude acts (PC, cloud, internet, connectors, subagents) and the timeline of each session.
Back to products
AI governance
Do you know how your team uses Claude?
This service shows you real Claude usage (Cowork, Office Agents and Claude Code) across your organisation: who uses it and how much, with which tools, what the estimated cost is and what actions it performs on your computers and systems. You will receive alerts when something is risky and a summary by email. Without storing the content of conversations.
The problem
Claude is already in the company. Do you know what it does?
When a team starts working with Claude, usage grows without anyone seeing it: not management, not IT, not the data protection officer.
You do not know who uses it or how much, or what spend each person generates.
Claude can delete files, send emails or open websites and nobody finds out until there is a problem.
Plugins and connectors get connected that nobody has approved.
The AI usage policy is written, but it cannot be checked.
Our answer
A dashboard, alerts and a summary that tell you what is happening
Claude sends an activity log to a server (who, when, which tool, whether it went well, how many tokens). The system filters it, stores it without the content of conversations and turns it into a dashboard in plain language.
Around thirty alerts
Mass deletions, emails to many recipients, websites outside the usual list, unauthorised connectors or plugins, high cost, out-of-hours use, repeated errors. Critical ones arrive immediately by email; the rest in a summary.
Usage per person
Tokens per model and estimated cost per person, to see where spend is growing.
Email summary
A daily summary, in Catalan or Spanish, with the day’s alerts.
Incidents
Open an incident from an alert or a session, assign it, comment on it and close it with a reason.
Privacy by design
The filter removes the text of conversations before anything is stored.
How it works
How Claude monitoring works
1
We agree the goal and the policy
What you want to measure, who sees what and how long data is kept. We prepare with you the information for staff and the impact assessment (DPIA).
2
We enable collection
Your company’s Claude administrator points monitoring at our server (a few minutes) and adds the domain to the allowed network egress. Optionally, you deploy the plugin that lets us also see actions on the PC.
3
1–2 week pilot
We check that data arrives and that no content is stored, with a small group.
4
Dashboard and alerts running
We adjust thresholds to your reality and switch on the email summary.
5
Periodic review
Claude evolves and its data format still changes: we revalidate the privacy filter after each major update.
Requirements
What you need to use it
A Claude Team or Enterprise plan.
An organisation administrator in Claude.
Adding the collection domain to Claude’s allowed network egress.
Information for staff and, where applicable, employee representatives. We help with that.
Roles and responsibilities
Who is involved
eSynapsing administration
Sees companies for support and service maintenance purposes.
Your company’s manager
Only sees your company’s data.
Access log
Every query to the dashboard is recorded: who, when and which screen.
Security and data
How we protect your data
The service stores metadata: who (user email), when, which tool, whether it succeeded or failed, duration, tokens and estimated cost. We do not store the text of conversations or documents: Claude sends it in the activity log and our filter removes it before storing. If you install the plugin, we also see the command, domain or path of some actions (with passwords and keys masked) and only the domains of email recipients. Data is kept for 30 days (incidents, 180). Data intake requires a token specific to each company and requests are rate-limited. The data is personal (user email): you are the controller and eSynapsing the processor, under the corresponding contract. This service is independent; it is not affiliated with Anthropic. Claude is a trademark of Anthropic.
Frequently asked questions
What people ask us about Claude monitoring
If you have a question you don't see here, write to us and we'll reply as soon as possible.
Do you store what my people write to Claude?
No. Claude sends that text in the activity log, but the filter removes it before anything is stored. We only keep metadata. We run an automated test of the filter and repeat it when Claude is updated.
What personal data do you process?
Each user’s email and identifier, and their activity (sessions, tools, tokens, alerts). With the plugin, commands, domains and paths of some actions, with secrets masked. Staff must be informed and a DPIA carried out; we support you.
Do I need special computers or software?
Not for basic logging: it is configured in Claude’s administration (Team or Enterprise plans). The plugin is optional and extends visibility to actions on the PC. It only covers users who install it.
Can I block or suspend users from this service?
No. The service watches and alerts; it does not block. It can note what would have been blocked (simulated mode) so you can assess rules before applying them yourself.
Does it only monitor Claude?
Yes. The service monitors Claude usage (Cowork, Office Agents and Claude Code). It does not cover other AI tools.
How long is the data kept?
Detail, 30 days; incidents, 180. You can request deletion when the service ends.
See it in action
Already using Claude? You will see the dashboard in 30 minutes.
We show you the dashboard with eSynapsing’s own usage data, anonymised. We are in a pilot phase: if you want, we start with your team.