Privacy Policy

Lieva is a B2B developer tool. Our customers are engineering teams at AI startups, and the people who use Lieva are their engineers, founders, and operators. We do not build consumer products, and we do not knowingly process data about individual end users except as it appears inside the traces a customer chooses to send us.

This document covers uselieva.com, the Lieva web app and dashboard, and the Lieva SDKs. It is written to be read by the engineers who integrate us. It is informational and is not a substitute for your own legal review.

Two things matter more than anything else in this policy, so they go first.

We collect two categories of data: account data that we need to run a SaaS, and trace data that you send us to debug your agents.

Account data is the standard set: name, work email, company, password hash (or your SSO identity), team membership, billing details handled by our payment processor, and product telemetry such as which pages you open and which features you use. We use this to operate your account, bill you, support you, and improve the product.

Trace data is whatever your agent run produces and your SDK configuration permits: model calls, tool calls, state transitions, retries, token usage, cost, latency, retrieved chunks, error and failure details, and the prompts, completions, and tool input/output attached to them. Trace payloads can contain sensitive content. You control what goes into them.

When you fork a run from step N and replay the fix, that replay executes in your environment and uses your OpenAI, Anthropic, or other provider credentials directly. The keys are read locally and the model calls go out locally.

Lieva is never in the path of those credentials. We do not receive them, proxy them, cache them, or store them. If a key ever shows up inside a trace payload you sent us, that means it was written into the trace content itself, not handed to us as a credential; you should redact it on the way out (see redaction below) and rotate it. Our design assumption is that keys live and die in your environment.

We use the trace data you send only to provide the product to you: storing and displaying your runs inbox, rendering the timeline, diffing an original run against a replay, generating root-cause labels, producing a regression eval from a trace, and firing the alerts and GitHub issue drafts you configure.

Root-cause labeling sends the relevant trace content to a Claude- or Sonnet-class model. We use these providers under their business or enterprise terms, which we select to keep your content from being used to train their foundation models; if that posture changes for a given provider, we will update this policy and our sub-processor list. We do not sell trace data, we do not use your trace content to train models of our own, and we do not share it with third parties for advertising. We may use aggregated, de-identified operational metrics (for example, error-class frequencies that cannot be tied back to you) to improve detection.

You decide what Lieva sees. The SDK lets you drop or mask fields, scrub prompts and completions, and exclude specific tool calls before any payload leaves your machine. Redaction happens client-side, so content you mask never reaches us in the first place.

You are responsible for not sending data you are not permitted to share, including personal data, secrets, or regulated content that your agreements with your own users or with third parties prohibit you from disclosing. If you process personal data on behalf of others and need a data processing agreement, email legal@uselieva.com and we will put one in place.

We rely on a small set of vendors to run the service, and they only touch the data needed for their function. We describe them by role rather than locking specific vendor names into this page, because vendors change and we would rather keep this accurate than impressive.

Current sub-processor roles are listed below. We keep a current vendor list available on request, and we notify account owners before adding a sub-processor that materially changes how trace data is handled.

Data is encrypted in transit with TLS and encrypted at rest. Access to production systems is restricted to the people who need it, authentication is required, and we log administrative access. No method of transmission or storage is perfectly secure, so we cannot promise absolute security, but we work to protect your data and to keep our posture honest.

On certifications, we tell you where we actually are: a formal SOC 2 audit is on our roadmap and is not complete. We do not claim certifications we do not hold. If you need our current security posture, our sub-processor list, or a security questionnaire answered before you send real trace data, email security@uselieva.com and we will give you the real state of things rather than a badge.

We keep trace data for as long as you keep the project that holds it, plus a short operational buffer for backups. You can delete individual runs, projects, or your whole workspace from the dashboard at any time. Deletion removes the data from active systems immediately and from backups as they age out on our normal backup rotation.

Account data is kept while your account is active and for a limited period afterward to meet legal, tax, and billing obligations, after which it is deleted or anonymized. If you want a full export or a full deletion handled by us, email privacy@uselieva.com. Closing your account triggers deletion of your trace data on our standard schedule.

If you are in the EU, EEA, or UK, you have the right to access, correct, delete, restrict, port, and object to the processing of your personal data, and to lodge a complaint with your supervisory authority. For trace data that contains personal data of your own users, you are typically the controller and we act as your processor; we will help you meet requests that reach you.

If you are a California resident, you have the right to know what personal information we hold, to delete it, to correct it, and to opt out of any sale or sharing. We do not sell your personal information and we do not share it for cross-context behavioral advertising. We will not discriminate against you for exercising any of these rights.

Send any rights request to privacy@uselieva.com. We will verify it and respond within the timeframe the applicable law requires.

Lieva is a tool for engineering teams and is not directed at anyone under 18. We do not knowingly collect personal data from children. If you believe a minor has created an account, or that a trace you sent contains a child's personal data, email privacy@uselieva.com and we will remove it.

When we change this policy in a way that affects how we handle your data, we will update the date at the top and, for material changes, notify account owners by email before the change takes effect.

Reach us at privacy@uselieva.com for privacy and data-rights questions, legal@uselieva.com for data processing agreements and legal matters, and security@uselieva.com for security questions or to report an issue.