AI Policy

Effective 8 October 2026 · Version 1.0

How Indo Mail uses artificial intelligence today, what we plan to build as Indo AI, and the rules that protect your mail, your data and your choices. Written for users, business administrators and regulators.

1. Our approach in one page

Indo Mail is an email and productivity service from INDO CHATS PRIVATE LIMITED. Your mail is private. We use artificial intelligence (AI) only where it clearly helps you, only with data that the task needs, and never to profile you or to read your messages for advertising.

This policy explains what AI features exist today, what we plan to add, which promises apply to every AI feature, and what you can do if you are unhappy with an automated result. It applies to Indo Mail on the web, on Android, on iPhone and as an installed web app.

We do not train any AI model on your mail, attachments, Drive files or contacts. This will not change without your separate, explicit opt-in.
AI features that touch your content are stateless: the content is processed to produce your answer and is not kept by the AI service afterwards.
Data of Indian accounts stays in India.
A person always confirms before anything is sent, deleted, shared or paid.
You can switch optional AI features off.

2. What we call AI

In this policy, AI means software that produces a result by learning patterns from data instead of following only fixed rules. It includes machine translation, language models that write or summarise text, and models that classify messages.

Some protective features of Indo Mail look intelligent but are plain rules written by our engineers, for example the scam warnings described below. We list them here so that you know exactly how each decision is made, and we mark them as rule-based.

3. AI and automated features live today

Feature	How it works	What data it uses	Where it runs
Indo Translate	Amazon Translate, a machine translation service, converts text between 34 languages when you ask for it.	Only the text you paste or choose to translate, plus the language pair.	Amazon Web Services, Mumbai region (ap-south-1).
Contextual ad matching (free plan)	Ads are matched to the placement, the country, the language and the words on the page that shows the ad.	Page context only. No mail content and no personal data.	Indo Mail servers, Mumbai.
Scam and phishing warnings	Rule-based checks run when a message arrives and show a banner with plain reasons.	The message itself: sender, links, attachment names and key phrases.	Indo Mail servers, Mumbai. Nothing leaves the server.
Sender checks (SPF, DKIM, DMARC)	Standard internet checks decide whether a sender is who it claims to be.	Mail headers and DNS records.	Indo Mail servers, Mumbai.
Indo Shield spam filter (release 1.6)	Seven checks score every incoming message from outside: sending-server reputation, sender-domain reputation and age, SPF/DKIM/DMARC, headers, a content model, links and attachments. The message goes to Spam at your chosen strictness, with the reasons shown. The content model is a word-based classifier trained only on example messages written by Indo Mail.	The arriving message. Only the sending server's IP address, the sender's domain and link domains are looked up in public blocklists; no text leaves the server. Your Report spam and Not spam clicks record the sender, its domain, the sending server and link domains, never the words.	Indo Mail servers, Mumbai.
Summary by Indo AI (release 1.6)	When you press Summarise, the most informative sentences of the conversation are picked and key details are listed: amounts, dates, reference numbers, codes, attachments and requests made of you.	The conversation you opened, for that request only. Nothing derived from it is stored.	Indo Mail servers, Mumbai.
Suggested replies (release 1.6)	Rule-based: three short fixed replies in English or Hindi are chosen from the kind of message you received. Nothing is sent until you tap one, edit it and press Send.	The last message you received in the conversation. Not shown for newsletters or automated mail.	Indo Mail servers, Mumbai.
Important marking (release 1.6)	Rule-based: mail is marked Important when you have written to the sender, it was sent directly to you, or it asks you to act. The reason is shown on the message. Your Important and Not important clicks are remembered per sender for your account only.	Sender, recipients, subject and opening lines when the message arrives, and your own sent mail. Mail with failed sender checks or a warning is never marked.	Indo Mail servers, Mumbai.
Inbox tabs and smart views (releases 1.5 and 1.6)	Rule-based sorting into Primary, Promotions, Social, Updates and Forums, and into Purchases, Jobs, Newsletters and Notifications.	Sender, standard list headers, subject and opening lines when the message arrives. Only the tab name is kept.	Indo Mail servers, Mumbai.
Nudges (release 1.6)	Rule-based reminders: mail that asked you a question 3 to 14 days ago without a reply, and mail you sent 4 to 21 days ago without an answer.	Dates, folders and question marks in your own mailbox.	Indo Mail servers, Mumbai.
Indo AI writing help and Ask Indo AI (release 1.9.2)	A large language model on Amazon Bedrock writes or rewrites an email, drafts a reply, writes a summary in its own words, or, in Ask Indo AI, uses Indo Mail tools (search, read, save a draft, archive, star, mark read, snooze) with the person's own access. It never sends email by itself: anything it writes waits in the composer or in Drafts for the person to check and send.	Only what the request needs: the text you wrote or selected, the conversation you opened, or the results of the searches it makes for your question. Nothing is stored after the answer and nothing is used for training.	Amazon Bedrock, AWS Mumbai region (ap-south-1). Started only when you press an Indo AI button; can be turned off in Settings → Advanced. Up to 150 requests per person per day.

Translate sends only the text you submit. We keep a short history of your own translations so that you can look back at them; you can delete that history. Translated text is never used to train a model.

Pro, Business and Enterprise plans carry no ads, so contextual ad matching does not apply to them.

None of the features in this table trains a model on your mail. Summaries, suggested replies, Important marking, tabs, nudges and the spam filter send no message text to any AI company.

4. Planned AI features (Indo AI)

Indo AI is our name for assistant features. Summaries, suggested replies, Important marking and nudges are live from release 1.6 (8 October 2026) and are described in section 3. The other features below are planned. When we release them, this policy applies from the first day, and we will update the release notes and this page before launch.

Writing help: draft a reply, adjust tone, shorten or expand a message you are composing.
Summaries: live since release 1.6 (section 3). A later version may write summaries in its own words with a language model running on Amazon Bedrock in the Mumbai region.
Smart search: find mail by meaning, not only by exact words.
Priority inbox and tabs: live since releases 1.5 and 1.6 as rule-based tabs, smart views and Important marking (see our Recommendation System page).
Better scam detection: the Indo Shield filter (release 1.6) added a content model trained only on Indo Mail's own examples and public reputation lists; further learned models are planned (see our AI Moderation page).
Translation inside mail: translate a message you are reading with one click.
Live from release 1.9.2 (9 October 2026): writing help (Write for me, rewrite, shorten, elaborate, fix grammar, bullet points, translate), Reply with Indo AI, written summaries, and the Ask Indo AI assistant, as described in section 3. They are started only by your action and can be turned off in Settings → Advanced.

Indo AI features will be off until you turn them on, or will be started only by an action of yours such as pressing a button. We will not read your mail in the background to produce AI output that you did not ask for.

5. Promises that apply to every AI feature

These promises apply to today's features and to every future feature.

No training on your mail. We do not use your messages, attachments, files, photos, contacts or searches to train or fine-tune any model, ours or anyone else's.
No advertising from content. AI is never used to read your mail to target ads. Ads use page context only.
Stateless processing. Content sent to an AI service for your request is used for that request only. We configure services so that they do not retain it or use it to improve their models.
Data in India. AI processing for Indian accounts runs in India. Summaries, suggested replies, Important marking and the spam filter run on Indo Mail's own servers in Mumbai and send no text to any AI service. We plan to run writing help on Amazon Bedrock in the Mumbai region, so prompts and answers stay inside India.
Least data. We send the smallest amount of text needed. For a summary we send the thread you opened, not your whole mailbox.
No sale of data. We never sell personal data and we do not share it with AI vendors for their own purposes.
Security. AI requests travel over encrypted connections and are tied to your signed-in session. Requests are rate limited.

6. Human confirmation and human review

AI may suggest. You decide. Indo AI will never send a message, accept an invitation, delete mail, share a file, change a setting or create a payment request on its own. Every such step needs your click or tap, after you have seen the exact text or action.

Automated decisions that affect your account, such as suspending an account, are not taken by AI alone. A trained member of our team reviews the evidence first. If you disagree, you can appeal as described in the AI Moderation page, and you can write to grievance@indoemail.in.

7. Labelling of AI-generated content

Indian law requires clear identification of synthetically generated information under the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021 (as amended). We follow the same idea in the product:

Text drafted by Indo AI is marked in the editor as AI-generated until you edit and accept it.
Summaries and translations carry a visible label such as "Summary by Indo AI" or "Machine translation".
Suggested replies are fixed phrases chosen by rules, shown under the heading "Suggested replies". They are placed in your reply only when you tap one, and you can edit them before sending.
If we ever create or edit images, audio or video with AI, the result will carry a visible label and, where the file format allows it, a metadata marker.
Messages you send are your own. We do not add hidden AI marks to your outgoing mail, but the draft label stays with the text in the editor until you accept it.
We do not allow anyone to use Indo Mail to spread synthetic media that deceives people, as set out in section 9.

8. Your controls

Turn suggested replies, Important marking and nudges on or off in Settings → Indo Mail AI, and set spam filter strictness in Settings → Spam protection. Summaries appear only when you press Summarise. Turning features off does not reduce any other part of the service.
Use Indo Mail without Indo AI. Core mail, Drive and every other app work without it.
Delete your translation history at any time.
Block images and keep remote content off in Settings.
Ask what we hold about you, ask for correction or erasure, or withdraw consent, by writing to privacy@indoemail.in, as provided by the Digital Personal Data Protection Act, 2023 and the Digital Personal Data Protection Rules, 2025.
Report a wrong or harmful AI result with "Report a problem" on every summary, "Not spam" or "Report spam" on filter results and "Not important" on Important marking, or write to support@indoemail.in.

9. Prohibited uses of AI on Indo Mail

You must not use Indo Mail, including any AI feature, to:

write or send scams, phishing, fake bank, KYC, OTP, lottery or "digital arrest" messages;
produce bulk unsolicited mail or evade our sending limits;
create deepfakes or other synthetic content that impersonates a real person, or that sexualises or harms a child, which we report as required by law;
generate malware, or content that helps hack another person's account;
produce material that is unlawful in India, including content that promotes violence or hatred against any group;
take decisions with legal or similar effect on a person, such as hiring, credit or insurance, without human review;
try to extract the instructions of an AI feature, bypass its safeguards, or use it to build a competing model.
Indo Mail does not accept political advertising, and AI may not be used to create it.

Breaking these rules can lead to removal of content, restriction of a feature or closure of the account. Serious cases may be reported to the police or to CERT-In.

10. Our own conduct and governance

Every AI feature is reviewed for privacy, security and fairness before launch, including how well it works in Indian languages.
We test for wrong, biased or unsafe output and fix problems before release.
We keep a record of the AI features we run, the services behind them and the data each one receives.
We list our sub-processors in the Privacy Policy. Today, Amazon Web Services is the only one that processes AI-related text (Translate).
We do not claim certifications we do not hold. Independent security and privacy certifications are planned, not yet held.
If an AI feature causes a personal data breach, we follow our incident process and inform you and the Data Protection Board of India as the law requires.

11. Changes to this policy

We will update this policy when we add or change an AI feature. For a material change that affects your data, we will tell you in the app or by email before it starts. Old versions are available on request from legal@indoemail.in.

12. Contact

Questions about AI: privacy@indoemail.in. Complaints: grievance@indoemail.in (Salena, Grievance Officer, INDO CHATS PRIVATE LIMITED, 168, GT Rd, near Aloka Cinema, Shibpur, Howrah, West Bengal 711102, India). We acknowledge within 24 hours and resolve within 7 days.

This document is governed by the laws of India.