Privacy Policy

Kalmia monitors production AI agents. Teams send us the traces their agents produce, and we analyze them for issues and report what we find. This page says plainly what we collect, what we do with it, who else touches it, and how long we keep it. Effective October 26, 2026.

Who this covers

This policy covers the Kalmia product at kalmia.dev, its API and MCP server, and this website. It is written for two groups of people.

  • Customers and their team members: the people who sign up, create a workspace, and send traces. For your account information and your use of the product, Kalmia decides how data is handled and is the “controller” in the sense privacy laws use.
  • People whose information appears inside traces. A trace can contain whatever the customer's agent saw and said, which may include personal information about the customer's own users. We process that data only on the customer's instructions, as their “processor.” If your information is in a trace, the customer who sent it is the one to ask about it, and we will help them respond.

What we collect

Account data. Your email address and name, the workspaces you belong to, and the role you hold in each. Sign-in uses a one-time link or a password; we never see the password in plain text. Workspace invitations record the inviter and the invited address.

Traces. The core of the product. A trace is the record of one agent run that a customer sends us through our API, SDK, or an import they authorize: the prompts and instructions the agent received, the model outputs, the tool calls it made and their results, timing, token counts, cost, and any metadata or tags the customer attaches. We do not decide what goes into a trace; the customer's integration does. Customers are responsible for having the right to send what they send, and for redacting anything they do not want us to process.

Analyses we derive. From traces we produce audits, verdicts, error records, findings, cluster investigations, digests, and fix suggestions. These are stored with the workspace that owns the traces and are deleted with them.

Integration data. Only when a customer connects a service. A Slack or Microsoft Teams connection stores the channel and the credential needed to post digests there, encrypted; when the Slack bot is invited to a channel, it reads replies to its own digest threads so it can answer them. A GitHub connection stores the app installation and reads repository contents for the repositories the customer selects, to ground fix suggestions in real code. A session-replay connection stores the customer's PostHog project credential and pulls the session recordings the customer links to a trace. A Braintrust import uses the customer's API key to copy traces from their Braintrust account. Disconnecting an integration removes its credential.

Usage data about this website and product. We use PostHog to record page views, clicks, and session recordings of the Kalmia interface, tied to your user id, email, and name once you sign in. Session recordings mask every input field, so text you type is not captured. We use Sentry to collect error reports and Axiom to keep request logs, both of which include technical details like request paths, timestamps, IP addresses, and browser information. API requests made with a workspace key are counted per workspace and route.

Cookies. A session cookie keeps you signed in. PostHog sets cookies to recognize a returning browser. We do not use advertising cookies or trackers.

How we use it

  • To run the service. Storing traces, analyzing them, showing results in the dashboard, serving them over the API and MCP, and delivering digests by email, Slack, or Teams.
  • To operate it safely. Authenticating you, enforcing per-workspace rate limits and spending caps, preventing abuse, and debugging failures.
  • To support you. Answering questions and investigating problems you report, which can mean an engineer looking at your workspace's data.
  • To improve the product. Understanding which features are used and where the analysis is weak. That includes re-running our analysis on stored traces to measure its accuracy, within the retention limits below. We do not use your traces to train models, ours or anyone else's.
  • To send you what you asked for. Sign-in links, invitations, and digests. We do not send marketing email to product users without a separate opt-in.
  • To meet legal obligations and to enforce our terms.

AI processing

Kalmia's analysis is done by large language models. To audit or investigate a trace, we send its contents to a model provider and store the result. Today those providers are Anthropic (through Amazon Web Services Bedrock) and OpenAI. We use their business APIs, under terms that do not allow the provider to train models on the data we send. Model providers may retain inputs briefly for abuse monitoring under their own policies. We do not train models of our own on customer traces.

Who we share it with

We do not sell personal information and we do not share it with advertisers. Data leaves Kalmia only in these cases.

Service providers that run the product for us. Each one sees only what its job requires.

  • Vercel: hosting and serverless compute for the application.
  • Supabase: the database that holds accounts, workspaces, traces, and analyses, and the authentication service.
  • Amazon Web Services (Bedrock), Anthropic, and OpenAI: model inference for the analysis described above.
  • Upstash: rate-limit counters, keyed by workspace id.
  • Resend: transactional email (sign-in links, invitations, digests).
  • PostHog: product analytics and session recordings of the Kalmia interface.
  • Sentry and Axiom: error reports and request logs.

Services you connect. Slack, Microsoft Teams, GitHub, PostHog, and Braintrust receive or provide data only when a customer connects them, and only as described above.

Legal requirements. We will disclose data when the law requires it, to protect the rights and safety of users or the public, or to enforce our terms. Where we are allowed to, we will tell the affected customer first.

A change of ownership. If Kalmia Technologies is acquired or merges with another company, data may transfer to the new owner under this same policy, and we will notify you.

How long we keep it

  • Traces and everything derived from them are deleted automatically 90 days after they are captured, by a daily cleanup. Deleting a workspace deletes its traces, analyses, integrations, and keys immediately. To have specific traces removed sooner, email us and we will delete them.
  • Account data is kept while your account exists and deleted when you ask us to close it.
  • Logs, error reports, and analytics are kept for a limited period for debugging and then expire under each provider's retention settings.
  • Backups of the database are kept by our database provider for disaster recovery and age out on their schedule, so a deleted record can persist in a backup for a short time after deletion.

How we protect it

All traffic to Kalmia is encrypted in transit with TLS, and our hosting and database providers encrypt data at rest. Every workspace's data is isolated by database row-level security, so one customer's traces are never readable by another. API keys are stored only as SHA-256 hashes; we cannot recover a key after it is shown once. Integration credentials are encrypted before they are stored. Access to production is limited to the people who operate the service, and sensitive settings changes are recorded in an audit log. No system is perfectly secure; if we learn of a breach affecting your data, we will notify you without undue delay. To report a vulnerability, email team@kalmia.dev.

Your choices and rights

You can see and change your name and email in account settings, delete traces and workspaces from the dashboard, and disconnect integrations at any time. To export your data, close your account, or ask anything else about your information, email team@kalmia.dev and we will respond within 30 days.

Depending on where you live, you may have legal rights to access, correct, delete, or receive a copy of your personal information, to object to or restrict some processing, and to complain to a supervisory authority. We honor those requests regardless of where you live. If your information was sent to us inside a customer's trace, we will refer your request to that customer and assist them, because they decide how that data is used.

You can opt out of PostHog analytics and session recording in your browser by blocking the PostHog domain; the product keeps working.

Where data is processed

Kalmia Technologies operates from the United States and our providers store and process data there. If you use Kalmia from outside the United States, your data will be transferred to and processed in the United States. For customers who need them, we make available standard contractual terms for international transfers and a data processing agreement; email team@kalmia.dev.

Children

Kalmia is a business tool and is not directed at children. We do not knowingly collect personal information from anyone under 16. If you believe a child has given us information, contact us and we will delete it.

Changes to this policy

When we change this policy in a way that matters, we will update the effective date at the top, and for significant changes we will email workspace owners before the change takes effect. Earlier versions are available on request.

Contact

Privacy questions, data requests, and security reports: team@kalmia.dev. Kalmia Technologies, United States.