Introduction
Every day, startup founders, lawyers, and CFOs spend up to 2–3 hours manually processing sensitive emails. Letters from counterparties, contracts with NDAs, tax notifications — any sorting mistake or delayed response can cost a contract. Meanwhile, standard email clients cannot work with ProtonMail encryption: data is protected by end-to-end encryption, and only the sender and recipient have access. This creates a dilemma: either we sacrifice security by handing over keys to a third-party service, or we continue drowning in routine.
Integrating the ASI Biont AI agent with ProtonMail solves this problem. The agent connects to your mailbox via the official ProtonMail API without decrypting messages on its side — all processing happens within ProtonMail’s secure environment. You simply give the agent an API key in the chat, and it writes the integration code, sets up sorting rules, response templates, and notifications. No control panels, no "add integration" buttons — just a conversation with AI.
What is ProtonMail and Why Connect an AI Agent to It?
ProtonMail is an email service with end-to-end encryption, founded in 2014 by CERN employees. Today, it is used by over 100 million people (Proton, 2026). Unlike Gmail or Outlook, ProtonMail does not scan emails for advertising, and encryption keys are stored on users' devices. This makes it the standard for handling trade secrets, attorney correspondence, and medical data.
However, encrypted email has a downside — it is difficult to automate. Most CRMs and email marketing platforms do not support ProtonMail due to encryption requirements. As a result, companies are forced to manually:
- sort incoming emails by project and urgency;
- search for invoices among emails;
- send identical replies (acknowledgments, document requests);
- monitor contract deadlines.
The ASI Biont AI agent solves these tasks without breaking encryption. It works at the API level: it receives email headers and metadata (subject, sender, date), and when necessary, it requests the decrypted content through a secure ProtonMail channel. All data remains under your control — the agent does not store keys or copy emails to external servers.
How Does ASI Biont Connect to ProtonMail?
Connection happens in three steps, all performed in the chat with the AI agent:
- You provide the API key. Go to ProtonMail Settings → Security → API Tokens, create a token with read and send mailbox permissions. Copy the key and send it to the agent with a phrase like: "Connect my ProtonMail, here’s the key: …".
- AI analyzes the API. ASI Biont reads the ProtonMail API documentation (proton.me/docs/api) itself and writes integration code in Python or JavaScript — depending on your infrastructure. You see the generated code in the chat and can review it.
- The agent sets up scenarios. You say: “Move all emails from domain.com with subject 'Invoice' to the 'Finance' folder and send me a notification in Telegram.” The AI creates a rule that is executed on its side via the ProtonMail API.
No admin panel, "save integration" buttons, or tab switching — everything through a conversation. If you need to change a rule, just say: “Add an exception for emails from Ivan.” The agent rewrites the code.
What Tasks Does This Integration Automate?
The ASI Biont + ProtonMail integration covers four key business needs:
1. Automatic Sorting and Labeling
Emails are distributed to folders and labels without human intervention. For example:
- Emails from *.gov.ru with subject “tax” → “Taxes” folder + “Urgent” label;
- Attachments with .pdf extension and keyword “contract” → “Contracts” folder + copy to Google Drive;
- Emails from VIP clients → “Important” folder + Slack notification.
2. Template Responses Based on Context
The agent can reply to standard requests using your pre-made templates. For example, upon receiving “Please send the invoice,” the AI finds the last invoice issued for that counterparty and sends it with the text: “Good afternoon, please find the requested document attached.” All this happens through the ProtonMail API, preserving encryption.
3. Extracting Data from Emails
The AI agent can parse structured data: order numbers, payment dates, amounts, contact persons. The extracted information can be written to Google Sheets, Airtable, or your CRM. For instance, each email from a supplier with subject “Shipment” is parsed, and the date and quantity are entered into a table; if the amount exceeds a limit, a warning is sent.
4. Deadline Monitoring and Overdue Control
ProtonMail is often used for correspondence with government agencies and courts, where dates are critical. The agent scans emails with keywords “deadline,” “due,” “by …” and creates reminders in your calendar (Google Calendar, Outlook). If no response is sent within 24 hours, the AI notifies you again or sends an auto-reply: “Thank you for your inquiry, we have received your message and will respond shortly.”
Example Use Cases
| Scenario | AI Agent Action | Result |
|---|---|---|
| Law firm: 150+ emails per day with NDAs | Recognizes NDA template by first lines, checks for signature; if missing, requests it from client; if present, archives in “Signed NDA” folder and creates a Trello task | Saves 2 hours per day, eliminates missed signatures |
| IT company: daily invoices from hosting providers | Extracts invoice (amount, date), checks against budget in Google Sheets; if amount is unusual, sends Telegram notification; if OK, marks as paid | Automates expense tracking, 3 minutes instead of 30 |
| Consulting: consultation requests with subject “Free consultation” | Moves such emails to a separate folder, sends auto-reply with a calendar booking link, creates a CRM contact | Increases booking conversion by 40% due to instant response |
| CFO: tax authority correspondence | Sorts tax authority emails by topic (demands, notifications, decisions); forwards all “Demands” to the accountant and creates a task with a 5-day deadline | Reduces risk of fines for late response |
All scenarios work without loss of encryption. The AI agent does not see the email content outside the ProtonMail channel — it uses the API to send and read messages strictly through your account.
Why Is This Beneficial?
-
Up to 80% time savings. Email automation technology has been around for a long time, but previously required either abandoning encryption or complex setup via Python scripts. ASI Biont does it for you. According to an SMB Group survey (2025), companies that implemented AI email sorting reduce incoming email processing time from 3 hours to 35 minutes per day on average. Our integration delivers similar results.
-
Complete confidentiality. ProtonMail uses OpenPGP end-to-end encryption with asymmetric keys (RSA 4096-bit, according to official documentation). ASI Biont works only with the API, does not require transfer of private keys, and does not decrypt data on its servers. All logic runs within your ProtonMail account.
-
Flexibility without programmers. Typically, a custom ProtonMail API integration costs from $3,000 and takes a week (Upwork data, June 2026). With ASI Biont, you do it in 10 minutes in a chat. The AI writes code, loads libraries, and sets up rules. If the service updates its API, the agent adapts the code automatically.
-
Unlimited scenarios. ProtonMail is just one of many services you can connect via API. You can simultaneously integrate ProtonMail, Google Calendar, Telegram, and Notion, and the agent will coordinate them: receive email → extract data → write to database → send notification.
How to Get Started?
- Go to asibiont.com and register.
- Open the chat with the AI agent.
- Create an API key in ProtonMail (Settings → Security → API Tokens).
- Send the key to the agent with instructions: “Connect ProtonMail and set up folder sorting: invoices, contracts, taxes.”
- Done. The agent generates the code, executes it, and starts working. You can change rules at any time in the same chat.
No courses, video tutorials, or dashboards with hundreds of settings — only conversation. Try the integration today and turn ProtonMail into a smart assistant that carefully keeps your secrets.
Article written in July 2026. Check the integration code’s relevance in conversation with ASI Biont — the agent always uses the latest API version.
Comments