Engineers use AI to explain code, chase bugs and write queries. BeeSensible highlights API keys, tokens and customer data in the prompt. One click turns them into placeholders, and the answer is still about your code.
Who this is for
65
kinds of sensitive data recognised while you type, in AI tools, email and chat
At once
Checked text is discarded straight away
EU
Our own servers at Scaleway and Hetzner, European companies with no US parent
A stack trace, a log line, a snippet from production: AI helps you work through them quickly. Sometimes there is an access token, a connection string or a customer email in there too. BeeSensible highlights those details as you type. One click masks them, so the prompt holds a placeholder such as [NAME1], and you ask your question as usual.
From the field
An engineer is stuck on an error and pastes the code fragment into an AI tool. The fragment holds a connection string with a password and an API key from the staging config. BeeSensible highlights both. Once they are masked, the question goes out and the answer still deals with the same code.
A developer asks AI to read a trace that breaks on an edge case. Buried in it are an access token, a customer's email address and an account number from the payload. They are highlighted and become placeholders. The explanation of the edge case is just as useful.
The engineering lead wants to know which AI coding tools are in use and how often keys and customer data turn up in prompts. BeeSensible shows this as totals per tool and per type of data, along with how often developers adjust highlighted details before sending.
The catalogue holds over 1,000 AI tools, including tools built for engineering, from AI code editors to autonomous coding agents. Each is scored on six dimensions, including data governance, jurisdiction and vendor trust. When a new one comes into use, it shows up in your overview. Then the organisation decides: allow it, allow it with a condition of your own, or disallow it. Decide, say, that DeepSeek is not allowed and Microsoft 365 Copilot is, and anyone opening DeepSeek sees that decision in the browser, with Copilot offered first as the alternative.
Engineers can remove credentials before asking an AI tool or a teammate for help.
A stack trace, an incident report, a sample data export: open it in the dashboard or the extension, or open a PDF in the desktop app for Mac or Windows. API keys, tokens, customer emails and account numbers are highlighted. You choose what goes and share an anonymised version.
Reported by customer orders@veldkampgroep.nl, account svc-checkout-prod, originating from 145.94.28.117.
The attached configuration still carried the key sk_live_9fQ2mTb4Xr and a session token eyJhbGciOiJIUzI1NiJ9.
An impression of the anonymisation screen. Detection runs on the document you open yourself.
Why this is hard
Engineers use Copilot, ChatGPT and Claude to fix bugs, explain code and write queries. Often with a snippet from production close at hand. Clear rules on which tool is fine for what make that work predictable.
An access token or connection string pasted into an AI tool nobody has approved has to be rotated and reported, even when everyone acts in good faith. With the key highlighted in the prompt, you see it before you send.
Logs, traces and query results contain names, email addresses and account numbers of real users. In a hundred lines of log they are easy to miss.
Existing controls tend to look at files, network traffic and endpoints. Someone who types or pastes text into a prompt in a browser tab needs help there and then, in the tool they are using.
Across engineering
The same question comes up in every workflow, from debugging to data migrations.
Debugging
Snippets and error messages you paste into an AI tool, sometimes with a key or password mixed in.
Code review
Diffs and pull requests you ask AI to summarise, config and credentials included.
Logs and traces
Stack traces and log lines with access tokens and real users' data.
Incidents
Postmortems and reports where payloads with customer data get pasted in.
Data migrations
Sample rows and queries with names, email addresses and account numbers.
How BeeSensible helps
Sensitive details are highlighted as people write. They decide what to mask or remove.
Highlights API keys, passwords, tokens and other credentials as you type, as well as personal data such as names and email addresses.
Runs in the browser, in AI tools, email and chat. Plus a desktop app for macOS and Windows, for Outlook and the desktop versions of ChatGPT, Claude and Copilot. No IDE plugin and no proxy.
When something is highlighted, you replace it with a placeholder such as [NAME1], or remove it. The question still makes sense to the AI tool. Your organisation chooses how warnings and masking are set up.
Security sees patterns by tool and category. What a developer writes is never stored and cannot be read.
For engineering leads, security and CISOs
BeeSensible gives you figures for the conversation with security and audit: how many highlights developers adjusted before sending, what happened after a tool notice, and what share of AI use runs through approved tools. All at organisation level, never per developer.
Total detections
12,438
Top apps
Example dashboard. Counts and types only, never content.
Security officer
Show auditors and the organisation that people get help while they write a prompt. With figures for each period: highlights adjusted or sent anyway, a recorded decision per AI tool, and what people did with the notice.
Engineering lead
The dashboard shows no text and no individuals. Only totals, with no per-employee drill-down anywhere. Insight into patterns, not surveillance of people.
CISO and IT
No proxy and no new application. The extension runs in Chrome and Edge. Detection runs on our own servers, hosted by European providers, with all traffic encrypted in transit.
Honest answers
Straight answers to the questions people actually ask before rolling this out.
The extension checks text in the input fields of supported tools for sensitive data. That text goes to a BeeSensible server at a European provider, is checked in working memory and discarded straight away. The content never reaches the dashboard and cannot be read by an administrator.
Not on its own. You see a highlight in the text and choose: mask or remove. Your organisation chooses how warnings and masking are set up, and whether someone can still go ahead with a tool that is not allowed.
BeeSensible supports your measures with insight and reports. The assessment stays with your organisation. You are the controller, BeeSensible is the processor and signs a data processing agreement (DPA) with you.
Yes. Detection works in Dutch and English and recognises technical secrets and personal data: names, email addresses and account numbers.
You get information for IT, privacy and security, a rollout plan and an explanation with videos for your team. No proxy is needed and nothing changes in your network. The extension runs in Chrome and Edge, the desktop app is optional. Central rollout through Microsoft Intune or Group Policy is available on every plan.
Compliance
GDPR
Supports your accountability and helps protect personal data as it is entered.
Secrets and credentials
Helps recognise API keys, tokens and connection strings before they reach an AI tool or log.
Data processing agreement
We sign a DPA with every customer. A product DPIA is available on request.
EU processing
Detection runs on our own servers at Scaleway in the Netherlands and Hetzner in Germany. European companies with no US parent, both ISO 27001 certified.
Try BeeSensible for 14 days with a few colleagues, or see it in a demo first.
Rens, BeeSensible
I'm happy to give you a twenty-minute tour. Leave your details and I'll find a time that suits you. You're also welcome to call or email.
ChatGPT