How we handle your information
Agentic Translator turns a short description of your business into automation ideas. This page explains, in plain language, what we collect, where it goes, how long we keep it, and what you can ask us to do with it. It is written to match what the software actually does.
Last updated 11 September 2026 · Legion AI LLC, Texas, USA, is responsible for this site and the information it collects.
1. What we collect
Everything below is information you type or attach yourself. Nothing is collected from you before you submit a form.
The analysis form
- The industry you pick and your description of the work (required).
- Whether your team already uses AI, and which tools (required question, free-text optional).
- Optional details: business name, location, average hourly pay, number of people, and hours per week spent on the work.
- Up to five attached files (see section 2).
- On the results: your thumbs-up or thumbs-down on each idea, the two answers if you tell a card how often a task happens and how long it takes, any feedback comment, and, if you ask for help, your email address and the ideas you chose.
Other forms
- Contact form: your name or company, email and message.
- Developer network application: name, work email, LinkedIn, portfolio or GitHub link, company website, industry, and the examples you write about your work.
- Industry request: the industry you want added, an optional business name and description, and your email.
- Sign-in (optional): your email address and a password, or the email, name and profile picture Google shares when you sign in with Google. One-time codes are emailed to you to confirm the address.
Collected automatically
- Your IP address is used to apply request limits, and is recorded when you unlock a report or sign out. Your browser type is recorded when you sign out.
- Hosting and error logs record the pages and endpoints requested, not what you typed.
- We do not run advertising trackers. A product-analytics tool (PostHog) is built into the site but is switched off during the pilot; if we turn it on we will update this page and section 7.
2. Files you attach
- We accept PDF, Word, text, CSV and image files up to 10 MB each. The original file is kept with your run in private storage so the team can review the analysis against what you attached. Only staff can open it, through a link that expires after a minute.
- Text is extracted from documents. Images and scanned PDFs are sent whole to our AI provider so it can read them.
- If the file contains no recognisable description of a work process, we tell you so and do not use it.
- If your typed description points at the attachment, or does not describe a task on its own, up to about 3,000 characters of the extracted text become part of your description and are kept with it (section 6).
Please leave out passwords, account numbers, card details, medical or patient records, and documents you are obliged to keep confidential. The analysis does not need them, and the tool cannot tell whether you had the right to share them.
3. What we use it for
- To produce your analysis: your description, industry, attached file text and the optional details are sent to an AI model, which decides which automations fit and writes the summaries you see.
- To size the estimates: team size and hours per week, when given, shape the hour figures. The terms of use (opens in a new tab) explain the arithmetic.
- To train our matching system: we keep descriptions, the recommendations they produced and your thumbs responses. Phrases from real descriptions are used to teach the system which automations they point to, and people at Legion AI review wrong or missing matches. This does not train the third-party AI models in section 4, and the data is not sold or shared with anyone else.
- To respond to you: your email and message are used to answer contact requests, help requests, industry requests and developer applications.
- To run the service safely: request limits, abuse prevention, error diagnosis and backups.
We do not sell your information, use it for advertising, or share it with other businesses for their own purposes.
4. Who else processes it
The site runs on services provided by other companies. Each receives only what it needs for its job and handles it under its own terms.
| Provider | What it does | What it receives |
|---|---|---|
| Anthropic | The AI models that read your description and write the analysis | Your description, extracted document text or images, industry, and the optional details (business name, location, pay, people, hours, AI usage) |
| OpenAI | Writes the rollout roadmap, a feature not shown during the pilot | If requested: a short excerpt of your description, industry, and the names and figures of your matched automations |
| Supabase | Database, sign-in, one-time-code emails, file storage, and backup storage | Everything stored (section 6) including attached files, account emails and passwords, Google sign-in details |
| Vercel | Hosting, request logs, and page-speed measurement (Speed Insights) | Your IP address, browser details, the pages and endpoints you request, and how quickly each page loaded for you. The speed measurement sets no cookies. |
| Optional sign-in; the Inter web font on every page | Sign-in: your Google account email, name and picture. Fonts: your IP address and browser details when the font loads | |
| jsDelivr | Delivers one script library used on every page | Your IP address and browser details when the script loads |
| Our email provider | Sends the emails in section 8 | Recipient address and the email content |
These providers process data in the United States. We do not currently have separate data-processing or business-associate agreements with them beyond their standard terms, which is one reason we ask you not to submit regulated records (section 10).
5. Who at Legion AI can see it
The owner and authorised staff use a private portal to read what testers submit: descriptions and attached-file text, the optional details, the results shown, thumbs responses, feedback comments, help requests and the email addresses given with them, contact messages, industry requests and developer applications. We also keep a log of the exact text sent to and returned by the AI models for each run, used to diagnose wrong matches; it contains the same information.
Access to the portal needs a staff account with a role the owner grants in the portal. Nothing in the portal is reachable from a tester's browser.
6. How long we keep it
| What | How long |
|---|---|
| Original attached files | Kept with the run in private storage, for as long as the run itself (next row), or until you ask us to delete them. |
| Working copy of a run (all inputs, extracted document text, and results while you use the page) | 24 hours, then removed automatically. |
| Your description, industry, optional details, results, thumbs, feedback comments, and the AI request log | Kept to train and improve our matching system for as long as it is useful, or until you ask us to delete them (section 9). There is no automatic deletion period. |
| Help requests, contact messages, industry requests, developer applications | Kept while we deal with them and for our records, or until you ask us to delete them. |
| Saved reports in an account | Until you delete them or close the account. Deleting a report hides it and disables its share link; the underlying data is removed on request. |
| IP address records (report unlocks, sign-outs) | Kept with the record they belong to. |
| Backups | A copy of the database is taken daily and stored privately. Backups are kept for disaster recovery and are not used for anything else; a deletion request removes live data immediately and leaves backups to age out. |
7. Cookies and browser storage
We use only cookies the site needs to work. There is no advertising or cross-site tracking, so we do not show a cookie banner.
| Name | Purpose | Lasts |
|---|---|---|
| lg_preview | Remembers that you entered the preview password | 7 days |
| lg_tid | An anonymous identifier that counts free analyses in a day | 24 hours |
| analysis_owner | A random token proving your browser started an analysis, so only you can reopen it | 30 days |
| sb-access-token and Supabase session storage | Keeps you signed in when you use an account | Until you sign out or the session expires |
| darkMode (local storage) | Your light or dark theme choice | Until you clear it |
| Session storage | Keeps the form you filled in so a page reload does not lose it | Until you close the tab |
Clearing your browser's cookies and site data removes all of these. Without the preview cookie you will be asked for the password again.
8. Emails we send
- To you: one-time sign-in and password codes; your report, if you ask for it by email; replies to anything you send us. We do not send newsletters or marketing email.
- To Legion AI: each contact message, help request, feedback comment, industry request and developer application is emailed to the owner's inbox so it can be answered. Feedback emails include the description you wrote and the ideas you were shown.
9. Your choices and rights
- Every optional field is optional. The analysis works with a description alone.
- You can use the site without an account. An account only adds saved reports.
- You can ask what we hold about you, ask for a copy, ask us to correct it, or ask us to delete it, including analyses you ran without an account, your account, and any application or message you sent. Use the enquiry form and tell us the email address or the approximate date of the run so we can find it. We answer within 30 days.
- Deleting means removing the live records, including the AI request log for those runs. Daily backups age out on their own.
- Depending on where you live, you may have legal rights to access, correct, delete or restrict the use of your information, or to appeal a decision we make about a request. We honour those requests through the same form, and we do not treat you differently for making one.
10. Sensitive information and children
The tool is for describing how work gets done, in general terms. It is not designed to receive protected health information, information covered by legal privilege, payment card data, government identifiers, or personal details about your employees or customers, and we do not want it. If you send such information anyway, we will delete it when we notice it or when you ask, but it will already have reached the providers in section 4.
The site is for adults running or working in a business. It is not directed at children, and we do not knowingly collect information from anyone under 16. If you believe a child has sent us information, contact us and we will delete it.
11. How we protect it
- The site is served over HTTPS and sits behind a preview password during the pilot.
- The database accepts connections only from our own server over an encrypted link, and its tables are locked so that the browser-facing keys cannot read them.
- Access for staff needs a staff account with a role the owner grants in the portal. Sign-in tokens are checked on the server on every request.
- Uploads are checked for type and size and are never executed. Requests are rate-limited to slow down abuse.
No system is perfectly secure. If we learn of a breach that affects your information we will tell you, and any authority we are required to notify, without undue delay.
12. Changes and contact
We will update this page whenever what the site does changes, and change the date at the top. During the pilot that may happen often; if a change affects information you have already given us, we will say so on this page.
Questions or requests about your information go through the enquiry form. Please do not include sensitive information in the message.