Privacy Policy

Kaer Labs Ltd (“Kaer”, “we”, “us”, or “our”) provides an autonomous AI agent platform through kaer.ai, app.kaer.ai, computer.kaer.ai, chat.kaer.ai, and our APIs and SDKs.

Controller: Kaer Labs Ltd, company no. 16674310, registered in England and Wales. Registered office: 167–169 Great Portland Street, Fifth Floor, London W1W 5PF. Email: [email protected].

For most account, security, billing, website, and product-improvement processing, Kaer is the controller of your personal data. Where a business customer uses Kaer to process Customer Data on its behalf, Kaer acts as a processor or service provider for that Customer Data under our Data Processing Addendum, which applies automatically to every business customer.

Contact us at [email protected] (privacy), [email protected] (security), or [email protected] (legal). If you are in the UK or the EEA, you have the right to lodge a complaint with your local data-protection supervisory authority; in the UK that is the Information Commissioner’s Office. We ask that you contact us first where possible so we can try to resolve the issue.

This Privacy Policy explains how Kaer collects, uses, stores, shares, and protects personal data when you visit our websites; create or use an account; use Kaer chat, assistants, workflows, automations, APIs, SDKs, Mail AI, voice mode, travel/maps features, computer-use agents, or paired-device features (PaerIDE/SuperWork); connect third-party accounts, mailboxes, calendars, drives, repositories, devices, or APIs; or contact us.

This policy does not replace the Data Processing Addendum or an enterprise agreement. If there is a conflict for Customer Data processed on behalf of an organisation, the Data Processing Addendum (or a signed agreement) controls for that Customer Data.

Account & identity

Email address; optional name or display name; OAuth identifiers and profile data (e.g. Google id, email, name, picture; GitHub id, login, name, avatar); login timestamps; timezone; country derived from timezone during verification; waitlist email; contact-form name, email, and message; and optional onboarding details such as role, industry, use cases, and experience level. If you use email-and-password login we store an Argon2id password hash, not your plaintext password. Magic-link tokens, one-time passcodes, and refresh tokens are stored as hashes.

Authentication & session

Session cookies, JWT claims, token identifiers, hashed refresh tokens, OAuth state cookies, hashed API keys, and similar authentication data. Session JWTs expire after one hour; refresh tokens use a 30-day sliding period with a 90-day absolute cap.

Usage, security & telemetry

Agent runs, model usage, compute charges, daily usage, workflow activity, API usage, timestamps, error events, and security audit events. For security audit events we persist IP addresses. Infrastructure and edge systems may also process request metadata needed to operate, secure, and troubleshoot the Service. We also record which parts of the app you use and what you do in each (for example opening an area, sending a message, changing a setting), with no message content and no cookies; while “Help improve our services” is on these records are linked to your account (see Improving Kaer).

Customer Data

Prompts, task descriptions, chat messages, prior transcript history, tool calls and outputs, file paths and contents, uploaded files, project files, workflow configuration, assistant instructions, sandbox artifacts, agent-generated browser activity, and API inputs and outputs. Customer Data may include personal data about you, your employees, your customers, your correspondents, or other third parties, depending on what you provide or connect.

Mail AI

If you connect a mailbox, we may ingest sender name, sender email, subject, message id, and full message body (attachments are not stored — only an attachment count). This may include personal data about third-party correspondents. Mail credentials and OAuth tokens are encrypted at rest.

Voice & audio

If you use voice mode, your browser captures microphone audio with your permission. Depending on the feature, speech is turned into text either by your browser’s own speech recognition (which may send the audio to your browser’s maker — Google, in Chrome — under its terms) or by a Kaer-operated speech-to-text system. We do not store raw audio. Voice-activity logs may store non-content metrics only.

Location & maps

We do not automatically prompt for precise GPS. If you explicitly choose to share precise location, we process those coordinates for the requested feature. We also process typed addresses, place queries, route inputs, viewport coordinates, and timezone-derived country.

During turn-by-turn navigation in the Kaer app, your device sends its position about every 15 seconds. For that trip we keep the route and only your latest position — never a trail — and delete them when you arrive or end the trip, or after 12 hours without an update.

While you navigate by car, Kaer also measures how quickly traffic is moving on the stretches you drive, and uses that to estimate congestion on other Kaer drivers’ routes. Each reading holds the approximately 100-metre map squares covered, the direction of travel and road name, how much slower than free-flowing traffic it was, the time, and a pseudonym that changes every day and is never stored with your account. Readings are deleted within an hour. Other drivers only ever see added delay and slow stretches on their own route — never who, how many, or when.

Outside navigation, location is not stored in a dedicated table but may appear as free text inside conversations or task records.

Computer-use, sandbox & paired devices

For cloud computer-use, we may process screenshots of the cloud sandbox or browser, browser state, file contents, command output, tool calls, and test artifacts; some screenshots may be sent to a vision model, and transient context may be cached briefly. For paired-device use, file paths, file contents, command output, prompts, results, workspace paths, device metadata, and tool-call telemetry may be processed. Computer-test features may persist intercepted browser HTTP traffic, including headers and bodies.

Payment

Kaer uses Stripe for payments. Stripe receives your email, Kaer user id, tier, amount, and tokenised card details. Card data is tokenised client-side; Kaer does not see or store full card numbers. Kaer stores payment identifiers such as Stripe customer, subscription, and payment ids.

Connectors

If you connect third-party services, we process the account data, tokens, credentials, content, and outputs needed to operate the connector. OAuth tokens, connector tokens, and user-provided API keys are stored in an encrypted per-user vault.

Cookies & local storage

Essential cookies and local storage for authentication, OAuth flows, CSRF protection, session management, and UI preferences run by default — the Service cannot operate without them. We do not set analytics, functional, or advertising cookies at all, so there is nothing to consent to and no cookie banner. See our Cookie Policy.

Where GDPR or UK GDPR applies, we rely on the following legal bases:

We do not intentionally ask you to submit special-category or highly sensitive personal data unless a feature specifically requires it and you are authorised to provide it. If you include such data in prompts, files, mail, screenshots, browser traffic, or connected services, we process it to provide the Service at your direction and subject to applicable law.

• provide, maintain, secure, and troubleshoot the Service;
• authenticate users and manage accounts, sessions, API keys, and connected services;
• run agents, tools, workflows, computer-use sessions, voice mode, Mail AI, maps/travel, and paired-device features;
• generate, store, and return outputs and agent records;
• process payments, billing, usage, taxes, credits, cancellations, and refunds;
• provide support and respond to privacy, legal, and security requests;
• detect and prevent fraud, abuse, malware, unauthorised access, spam, and security incidents;
• fix failures and improve service quality using operational data such as error events and aggregate usage counts;
• evaluate model performance and train Kaer’s own in-house models while “Help improve our services” is on (on by default; you can switch it off at any time), with passwords, keys, tokens, card numbers and email addresses removed;
• comply with law and enforce our agreements; and
• protect the rights, safety, and property of Kaer, users, and the public.

Kaer is an AI agent platform. To provide the Service, we send prompts, messages, context, tool outputs, file contents, mail content, screenshots, and other task-relevant Customer Data to AI model providers when needed for the requested feature.

Every model provider Kaer can send this content to is listed, with what it receives and the country of the company, under Sub-processors. Several of them are companies based in China, and others are in the United States. Which provider handles a request depends on the model you or Kaer’s router picks: for example, by default Alibaba Cloud (Qwen) triages Mail AI and reads screenshots for vision tasks, and Zhipu AI (Z.ai, GLM) is tried first for mail summaries and draft replies.

If the provider handling a request fails or is unavailable, Kaer automatically sends the same request to another provider on the list, without asking you first. DeepSeek, Kimi and GLM models that Alibaba Cloud or Tencent Cloud host for Kaer run on those companies’ services, not on DeepSeek’s or Moonshot’s; Kimi K3.1 Pro is the exception and is served by Moonshot AI directly.

These providers may process prompt content outside the UK and EEA. Where UK GDPR or EU GDPR requires a transfer safeguard, Kaer relies on the UK International Data Transfer Agreement or Addendum, or the EU Standard Contractual Clauses, with supplementary measures. We never use Customer Data to train third-party foundation models.

From 22 October 2026, “Help improve our services” (previously called “Help improve Kaer”) is on by default for everyone who has not made a choice. If you had already switched it off, it stays off. While it is on, Kaer uses records of how you use the Service, including your prompts, messages and coding actions, to fix problems, improve our services and train Kaer’s own models. You can switch it off at any time.

What is collected

Usage of each area of Kaer (for example chat, build, tasks, calendar, mail and settings): which area you opened, what you did there, whether it worked and how long it took. Content samples: your prompts, Kaer’s messages and the coding actions Kaer takes for you (such as commands run and files edited). These records are linked to your account. Private (incognito) chats are never collected.

How secrets and identifiers are removed

Passwords, API keys, access tokens, payment-card numbers and email addresses are removed automatically before a record is stored. Other details you type — names, places, anything else — can remain in the text, so these records are treated as personal data: access is restricted and the safeguards below apply. They stay linked to your account so that we can find and delete them when you ask.

Purposes

Fixing problems: finding and fixing failures and slow or confusing parts of the Service. Improving our services: understanding how each area is used and measuring answer quality and safety. Training models: developing, training and testing Kaer’s own in-house models. Kaer never uses this data to train third-party foundation models.

Lawful basis

Our legitimate interests (Art. 6(1)(f) UK GDPR and EU GDPR) in fixing problems, improving the Service and training our own models. We have weighed these interests against yours: secrets are removed, private chats are excluded, access is restricted, and you can stop it at any time. You have the right to object to this processing at any time, and switching the setting off does exactly that.

How to turn it off

Go to Settings → AI → “Help improve our services”, or Settings → Tracking & Analytics, and switch it off. You can also email [email protected]. Switching it off stops collection at once and deletes the content samples already kept. With it off, we still count use of each area to operate and secure the Service, but without your account ID. Your choice does not affect the Service or your billing.

Retention

Usage events are deleted 400 days after they are collected. Content samples are deleted 180 days after they are collected, or at once when you switch the setting off or delete your account. Prompt-injection screening logs keep only a hash of the screened text, the verdict and a timestamp — never the text itself or who sent it — and are deleted after 90 days.

Special categories and children

We do not intentionally ask you to submit special-category data (race, religion, health, biometric data, etc.). If you include such data in prompts, Kaer processes it at your direction. The Service is not directed to children under 16; if we learn we collected data from someone under 16, we delete it promptly.

Sale of data

Kaer does not sell personal data, does not share personal data for cross-context behavioural advertising, and never uses Customer Data to train third-party foundation models. While “Help improve our services” is on, our training pipeline may move eligible content samples, with secrets removed, into training storage and process them on external GPU infrastructure (Modal), with training metadata logged to Weights & Biases. We treat these providers as sub-processors and apply contractual, access-control, and transfer safeguards.

Mail AI can read, classify, summarise, draft, and help handle email from a mailbox you connect. When enabled, it may ingest full message bodies and metadata on a polling or webhook basis. This includes personal data about people who send email to you or receive email from you. Mail content is sent to AI model providers: by default Alibaba Cloud (Qwen) for triage and auto-handling, and Zhipu AI (Z.ai, GLM) first for summaries and draft replies. If a provider fails, Kaer automatically sends the request to another provider listed under Sub-processors.

You are responsible for connecting only mailboxes you have authority to connect and for configuring Mail AI in a way that is lawful for your organisation and correspondents. You can limit Mail AI by disconnecting a mailbox, disabling mail features, deleting mail resources, or contacting [email protected].

Kaer agents can operate in cloud sandboxes and hardened container environments, capturing screenshots, browser state, files, command output, tool calls, and test artifacts needed to perform a task. Some screenshot image bytes may be sent to a vision model, and transient context may be cached for up to 24 hours.

Paired-device features let an agent execute approved tools on your own machine, protected by a hard human consent gate: delegation requires a human-set, single-use approval key that expires after five minutes, and the model cannot grant it itself. Mail-class tools cannot route to a paired device. Pair only devices and workspace paths you are authorised to expose. Computer-test features may persist full intercepted browser HTTP traffic, including headers and bodies — do not use them on systems or accounts unless you are authorised to capture that traffic and have considered whether cookies, credentials, or personal data may be present.

Kaer’s own servers and primary storage are in the EU (Germany and Finland). We use the third-party sub-processors below to provide the Service. “Where enabled” means Kaer uses the provider only while it has switched that provider on; every row can receive the data shown. Country is where the company is based.

If an AI model provider fails or is unavailable, Kaer automatically sends the same request to another AI model provider in this table (never OpenRouter, which receives none of your content), so your content may be processed by a provider other than the one first picked.

AI model providers

Other sub-processors

Where transfers leave the UK or EEA, we rely on Data Processing Addenda, the UK International Data Transfer Agreement or Addendum, the EU Standard Contractual Clauses, and supplementary measures. Business customers get at least 14 days’ notice of a new sub-processor under the Data Processing Addendum. Some connectors are user-directed services rather than Kaer-controlled processing — for example Home Assistant, Telegram, Twilio SMS, Spotify, Todoist, Linear, Notion, Strava, Duffel, CoinGecko, and Alpha Vantage. When you connect or instruct an agent to use such a service, that service’s own terms and privacy policy also apply.

We disclose personal data only as needed:

• to sub-processors and user-directed third-party services that help provide the Service;
• to model providers to generate outputs and perform requested AI tasks;
• to Stripe to process payments;
• to connector providers you authorise;
• to comply with law, subpoenas, court orders, government requests, or legal process;
• to enforce our agreements, investigate abuse, or protect the rights, safety, and property of Kaer, users, and the public;
• to professional advisers under duties of confidentiality; and
• in connection with a merger, acquisition, financing, or similar corporate transaction, subject to confidentiality and data-minimisation safeguards.

We do not sell personal data and do not share personal data for cross-context behavioural advertising.

Kaer’s own servers and primary storage are in the EU, in Hetzner data centres: Falkenstein, Germany (the app, backend and database servers) and Helsinki, Finland (the machines that run users’ computers, the search index, and backups). Kaer’s mail server, which holds Kaer Mail mailboxes and mail sent to kaer.ai addresses, is a Contabo server in Germany.

This does not mean all processing stays in the UK or EEA. Some sub-processors — notably AI model providers, several of which are companies based in China or the United States, and training-pipeline providers — may process prompt content, Customer Data, or related metadata outside the UK and EEA, and certain connector, payment, search, map, CDN, edge, and support services may also process data in other regions. The Sub-processors table names each provider and its country.

Where UK GDPR or EU GDPR requires a transfer safeguard, Kaer uses the UK International Data Transfer Agreement or the UK Addendum to the EU Standard Contractual Clauses, or the EU Standard Contractual Clauses, together with transfer risk assessments, access controls, encryption in transit, encryption for secrets and credentials, provider due diligence, and other supplementary measures. Contact [email protected] for more information about transfer safeguards.

We keep personal data only as long as needed for the purposes in this policy, unless a longer period is required for legal, tax, accounting, security, dispute-resolution, or compliance reasons.

Before deleting a dormant account we send an email notice; you can keep your account simply by signing in before the date in that notice. You can export or delete your data at any time by emailing [email protected]. When we make material changes to this policy, we notify registered users by email in advance as described in Changes to this policy.

We use technical, organisational, and contractual safeguards including encryption in transit with TLS 1.2 or later; AES-256-GCM application-layer encryption for service secrets, connector/OAuth tokens, user secrets, and mail credentials; Argon2id password hashing; hashed refresh tokens and API keys; access controls; environment-based key management; audit logging for security-sensitive events; and least-privilege production access.

Not all Customer Data is application-layer encrypted: email bodies, chat transcripts, sandbox output, captured HTTP bodies, and project file blobs are stored in protected systems but are not separately encrypted by the application layer in the current architecture. No system can be perfectly secure. If we become aware of a personal-data breach requiring notice, we will notify affected users and regulators as required by law, and business customers as set out in the Data Processing Addendum. Report security issues to [email protected].

Depending on where you live and how you use the Service, you may have rights over your personal data.

GDPR & UK GDPR

You may request access; correction; erasure; restriction; portability/export in a structured, commonly used, machine-readable format; objection to processing based on legitimate interests; withdrawal of consent at any time without affecting prior processing; objection to service improvement and model training (by switching off “Help improve our services”); information about transfer safeguards; and review of automated decision-making where applicable.

California (CCPA / CPRA)

You may request to know/access the categories and specific pieces of personal information we hold; to know sources, purposes, disclosures, and recipients; deletion; correction; opt-out of sale or sharing; limitation of sensitive personal information where applicable; portability; and non-discrimination for exercising your rights. Kaer does not sell personal information and does not share it for cross-context behavioural advertising.

How to exercise rights

Use account settings where available, or email [email protected] — including for account deletion or data export. We may verify your identity or authority before fulfilling a request. We generally respond within one month for GDPR/UK GDPR requests and within 45 days for CCPA requests to know, delete, or correct, unless an extension is permitted. You may use an authorised agent where the law allows.

Kaer provides autonomous agents that act on your instructions — drafting, summarising, classifying, searching, navigating, running tools, operating cloud sandboxes, interacting with connected services, or, with your approval, running tools on a paired device.

Kaer does not intend to make solely automated decisions about you that produce legal or similarly significant effects. Agents are tools under your control, and you are responsible for reviewing agent plans, outputs, and actions — especially before sending messages, modifying systems, committing code, making purchases, booking travel, or taking actions that affect other people. If you believe Kaer has made an automated decision about you with legal or similarly significant effect, contact [email protected] and request human review.

The Service is not directed to children under 16, and you must be at least 16 to use Kaer. If we learn that we collected personal data from someone under 16, we will delete it promptly unless we are legally required to retain it.

Kaer uses essential cookies and local storage for authentication, OAuth flows, CSRF protection, session management, and UI preferences; these always run because the Service cannot operate without them. Kaer does not set analytics, functional, or advertising cookies, so no consent banner is shown. See our Cookie Policy for cookie names, purposes, and browser controls.

We may update this Privacy Policy as the Service, laws, or our practices change. For material changes, we will provide notice by email at least 14 days before they take effect, unless a shorter period is required for legal, security, or operational reasons.

Privacy and data-rights requests: [email protected]. Security matters: [email protected]. Legal notices: [email protected]. Kaer Labs Ltd, 167–169 Great Portland Street, Fifth Floor, London W1W 5PF, is responsible for this policy.