Privacy policy

Eager (“Eager”, “we”, “us”) is a sole proprietorship based in California, USA. This policy explains what information we collect when you use eager.ai (formerly guidinghand.ai), the Eager console and API, and the Eager desktop app for Mac and Windows (together, the “Service”), and what we do with it.

Questions or requests: support@guidinghand.ai.

Two kinds of people use Eager. Account holders are businesses and their support staff who sign up, make codes and run tasks. End users are the people whose computers connect to a session, usually the account holder’s customers.

For information about end users that passes through a task (for example, what’s on their screen), we act on the account holder’s behalf as its service provider. The account holder decides which tasks to run and is responsible for having the right to run them. End users with questions can contact the business that invited them, or us.

Account information. When you sign in with Google we receive your name and email address. When you sign in with a code we email you, we receive the email address you type in; we keep a one-way hash of the code (never the code itself), with the network address it was asked from, only to check it and to limit abuse. We note whether you first signed up with Google or by email, and the network you signed up from, so the free credit goes to each person once. We also store your team name, your balance, top-ups and usage, your auto top-up settings, and, if you create one, a one-way hash of your API key (never the key itself). Your account is stored with our payment processor, Stripe. When you add funds, Stripe collects your payment details and keeps your card on file so it can be charged for later top-ups, including automatic ones if you turn them on. We only see the card’s brand, last four digits and expiry date; we never receive your full card number.

Session and device information. For each connected computer: the pairing code, the computer’s name, its operating system, screen size, the app version, and whether the permissions the app needs are granted.

Task content. The instructions you give the agent; its progress updates, questions, and your answers and approvals; screenshots of the connected computer’s screen, taken while a task runs; and the actions the agent takes, such as where it clicks and the text it types.

Conversations. What people write to an agent in a chat, the files they attach, the agent’s replies, and the files its code makes. We keep them as long as the account keeps its history (90 days unless it sets otherwise); a file uploaded but never sent is deleted after a day.

Diagnostics. A record of each step of a task: the kind of action, screen coordinates, key names, how many characters were typed (not what they were), how long each step took, whether the screen changed, and any errors.

Network information. Like most websites, our servers see IP addresses and browser details, which we use for security and to limit abuse. When someone downloads the app from an invite page, we briefly remember their network address, operating system type and time zone (for up to 30 minutes) so the app can fill in their code when it opens.

Cookies and local storage. We use one cookie to keep you signed in (it lasts 30 days) and a short-lived one while you sign in, which lasts 10 minutes: with Google, it checks that the sign-in you return from is the one you started; with an emailed code, it makes the code work only in the browser that asked for it. The console keeps its current session and your preferences in your browser’s local storage, and while you wait for an emailed code, the sign-in page keeps the address in that tab’s storage. When you first visit our website, it sets one more cookie of ours recording how you found us: the page you arrived on, the site that sent you (without what you searched for there), any campaign tags in the link and, if you came from an ad, the ad’s click ID. A later visit from an ad or a campaign link is added to it. It lasts 90 days from the last time it changed. If you sign up, or write to us through the pricing page’s form, we save it with your account or your message (our sales team sees it with the message), to learn which of our pages, posts and ads bring people to Eager. We do not pass this attribution cookie to Google Analytics or PostHog. Separately, if you accept analytics, we use Google Analytics and PostHog to measure page visits, sessions, engagement, browser and device information, approximate location and selected actions such as requesting a demo or publishing an agent. These providers receive browser network information, including your IP address, when your browser connects to them. We send fixed page and action labels, not conversation content, uploaded files, form answers, names, email addresses, organization names or page query strings. PostHog links activity to an opaque account ID when you sign in; Google Analytics does not receive that account ID. With your consent, PostHog also records how you move through public website pages, including clicks and scrolling. All page text and form inputs are masked; forms, customer trials and embedded content are blocked from recordings. Sign-in pages, the console, customer chats and the desktop app are not recorded. Page queries and fragments are removed from replay URLs. Automatic analytics click capture, AI conversation tracing and application-log collection remain disabled. Advertising features and Google signals are disabled. Analytics cookies and your analytics choice last up to 180 days. PostHog also keeps browser identifiers in local storage while analytics is enabled. Data already received by each provider follows the retention settings of its analytics project. You can accept or decline in the Analytics preferences box that opens on your first visit, and change your choice at any time from Analytics preferences in our website’s footer or in the console’s account menu. Declining stops future collection and clears our analytics cookies and PostHog browser identifiers; it does not erase data already collected. We keep optional analytics off when your browser sends Do Not Track or Global Privacy Control. Our existing sign-in and first-party attribution work separately from this optional service.

On the connected computer. The desktop app saves its connection details on that computer. It doesn’t read files or look at the screen except to carry out a task.

• To run the Service: connect computers, carry out tasks, show progress, and let you approve or stop steps.
• To manage accounts, balances, free credit, top-ups and billing.
• To keep the Service secure, prevent abuse and fraud, and enforce our terms.
• To troubleshoot problems and fix bugs.
• To contact you about your account, billing, and important changes to the Service.
• To learn which pages, posts and ads bring people to Eager (from our attribution cookie), and, with your consent, to understand visits and selected feature use through Google Analytics and PostHog.
• To comply with the law.

We don’t sell personal information, share it for advertising, or use task content to train AI models.

To decide each step of a task, we send the task’s instructions, the conversation so far and the latest screenshots, through its API, to the maker of the AI model the agent runs on: OpenAI, Anthropic, Google, DeepSeek or Moonshot AI (Kimi), depending on the model the agent uses. Each processes this data under its API terms and policies. DeepSeek processes and stores it on servers in China, and Moonshot AI on servers in Singapore; Moonshot AI’s terms let it use API data to improve its models. An agent uses a DeepSeek or Kimi model only when your team picks one for it. OpenAI says it doesn’t use API data to train its models by default, and that it may keep API data for up to 30 days, including to link the steps of a task together and to monitor for abuse. Anthropic’s and Google’s API terms say how they use and keep API data. Files attached to a chat go to the same maker so the agent can read them. A long answer in a chat also goes to Anthropic (Claude), whatever model the agent runs on, to lay it out as cards and tables, unless your team turns visual answers off for that agent; the layout is kept with the conversation. When an agent has Run code on, or your team talks with it in Coach, its code runs in the sandbox of the maker of its AI model, one per conversation and with no internet access. On a Claude model, its code, those files and the agent’s own files run in Anthropic’s sandbox, which Anthropic keeps for up to 30 days; the copies we give Anthropic are deleted with their conversation or file, and expire within 30 days. On a GPT model, they run in OpenAI’s sandbox, which OpenAI deletes with the files in it once it has gone unused for 20 minutes. On a Gemini model, the code runs in Google’s sandbox, with no files but what the conversation already sends Google. DeepSeek and Kimi models have no sandbox, so Run code is off on them. When your team talks in Coach with an agent on a Gemini model, Coach runs on Anthropic’s Claude, so that conversation and its files go to Anthropic. When your team talks with an agent out loud in Coach, the audio and what’s said go to OpenAI (GPT-Live), or to ElevenLabs for an agent your team gives an ElevenLabs voice, which listens and speaks for it, whatever model the agent runs on; on ElevenLabs, its own model (Anthropic’s Claude, which it hosts) decides when to hand what’s asked to the agent. We don’t keep the audio; we keep how long each voice conversation lasted, to bill it, and what was asked, as the Coach conversation. When someone calls an agent’s phone number, Twilio carries the call and sends its audio straight to OpenAI (GPT-Live), or to ElevenLabs for an agent with an ElevenLabs voice, which listens and speaks for the agent, whatever model the agent runs on; the audio doesn’t pass through our servers and we don’t keep it. ElevenLabs doesn’t record the audio for us, and keeps the transcript and how long each conversation lasted for 30 days. What the caller asks goes to the agent’s model like any chat. We keep the call as one of the agent’s conversations (what was asked and answered, as text), the caller’s phone number and the number they called, and how long the call lasted, to bill it.

We share information only as needed to run the Service:

• The account holder’s team. Task content, including screenshots of a connected computer, is visible to the account that ran the task, in the console and through the API.
• Service providers that process data for us: Stripe (accounts, billing and payments, and storing your card for top-ups), Google (sign-in, optional Google Analytics with your consent, and the AI model for agents that run on Gemini), PostHog (optional product analytics with your consent), Resend (sending the emails with sign-in codes), OpenAI and Anthropic (the AI model, depending on the model the agent runs on; OpenAI also for spoken Coach conversations and phone calls), DeepSeek and Moonshot AI (the AI model, for agents your team puts on a DeepSeek or Kimi model), ElevenLabs (listening and speaking in spoken Coach conversations and phone calls, for agents your team gives an ElevenLabs voice), Twilio (agents’ phone numbers, and carrying calls to them), Render (hosting our servers and logs), Browserbase (hosting cloud browsers, and recording each as a video for its task’s replay, kept 30 days; not when the account turns screen recording off), Cloudflare (network delivery and security) and GitHub (hosting app downloads).
• When required by law, or to protect the rights, safety and security of our users, the public or Eager.
• If Eager is sold or merged, to the new owner, who will be bound by this policy for information collected under it.

• Tasks, their updates and screenshots are held in our server’s memory while a session is in use, and deleted within 72 hours of its last activity or when our server restarts, whichever comes first.
• Diagnostic step records are kept in our hosting provider’s logs for up to 30 days.
• Emailed sign-in codes (a one-way hash, the address they went to and the network address they were asked from) are deleted within 48 hours. The emails themselves stay in your inbox and, for a limited time, with Resend.
• Account information is kept until you ask us to delete your account.
• Billing records are kept by Stripe, and by us for as long as tax and accounting laws require.

Connections to the Service are encrypted. API keys are stored only as hashes, sign-in cookies are signed so they can’t be forged, and access to our systems is limited. No system is perfectly secure, so we can’t guarantee absolute security. If we learn of a breach that affects your information, we’ll tell you as the law requires.

You can ask us to show you, correct, export or delete the personal information we hold about you, or object to how we use it, by writing to support@guidinghand.ai from the email address on your account. We’ll respond within 30 days (45 days where California law applies) and may need to confirm your identity first. An owner can turn off auto top-up at any time on the console’s Usage & billing page, and you can ask us to remove your saved card.

California residents have these rights under the California Consumer Privacy Act where it applies, including the right to know what we collect (the categories above: identifiers, account and commercial information, internet activity, and the content of tasks), the right to delete and correct it, and the right not to be treated differently for using these rights. We don’t sell or share personal information as those laws define it. You can use an authorized agent to make a request.

If you’re in the European Economic Area or the United Kingdom, we process your information to provide the Service you asked for (performance of a contract), for our legitimate interests in keeping it secure and working and in learning how people find us, and to meet legal obligations. Your information is processed in the United States. You can also complain to your local data protection authority.

End users can make these requests through the business that invited them, or through us; we’ll work with that business to respond.

The Service is for businesses and isn’t directed at children. We don’t knowingly collect information from anyone under 16. If you believe a child has given us information, contact us and we’ll delete it.

We may update this policy as the Service changes. We’ll post the new version here with a new date, and for significant changes we’ll email account holders before they take effect.

Eager, California, USA. Email support@guidinghand.ai.