It reads every email that hits support@, looks up the customer before it writes a word, replies in the thread within minutes, and holds the ones that are yours.
It lives in your support inbox — looking up the customer before it writes, replying in the thread in their language, filing every ticket to the CRM, and holding only the emails your rules say are yours.
Email is still where most support actually happens, and it's the channel the AI chat widgets ignore. A customer writes to support@ at 7am, and until someone opens the inbox the ticket is aging. Fasrad's email support agent lives in that inbox. When a message arrives it reads the whole thread, pulls up who the sender is — their orders, their plan, the last time they wrote in — and answers in the same thread, from the same address, usually within a minute. Not a canned acknowledgement: the answer, with the tracking number, the reissued invoice, or the refund already done.
It runs in whichever mode fits you. Give it a dedicated support address and it answers on its own, inside limits you set. Point it at your own inbox instead and it drafts every reply in your voice and sends only when you tap approve — nothing leaves your address without your say-so. Both modes read your mail without touching it: nothing is marked read, moved, or deleted. Customers write in Spanish, German, or Japanese; it replies in kind.
Rules you write once, applied to every email:
Every email becomes a record. The sender is filed in the built-in CRM as a contact, the ticket gets a category and a one-line summary, and the agent remembers what it learns — that this customer had a late delivery in May and was given a credit, that they prefer short replies. When it needs you, it asks with the draft attached and buttons to tap, on Telegram or in chat. And because it keeps a searchable copy of the mailbox, 'how many people complained about the linen shirt this month?' is a question it can answer in seconds, by meaning, not by subject line.
These pages cover the adjacent jobs buyers usually compare before choosing an AI agent.
Your choice, per inbox. On a dedicated support address it replies on its own within the rules you set — refund limits, topics that always come to you, phrases it must never use. On your own inbox it drafts in your voice and sends only when you tap approve. Many teams run both: the agent answers support@, and drafts the sensitive ones for you.
In the thread, always — proper reply headers, 'Re:' subject, from the same address the customer wrote to. To the customer it looks like a person answered their email, because one effectively did.
It looks them up before it writes. Orders, plans, and past tickets can live in its built-in records, in a sheet you keep, or behind your own API — it reads all three. And it remembers what it learns across emails, so the credit you issued in May comes up when the same customer writes in September.
It replies in the language they wrote in — no rules to set, no separate templates. Your instructions stay in your language; the customer gets theirs.
No. It reads mail read-only. Nothing is marked read, moved, or deleted, and nothing is filed unless a rule you wrote says to. If you look at the inbox after it has answered a ticket, the original email is exactly where it was.
They come to you, untouched, with the context attached — who the sender is, what they've asked before, and a draft if you want one. You define what 'shouldn't' means in plain English: legal threats, chargebacks, press, anyone on a list of accounts. It tells the customer who has it and when they'll hear back, so nothing sits silent.
$49/month or $490/year — cancel anytime. That covers the inbox, the CRM, the searchable mail history, and the rules — unlimited emails, no per-ticket or per-resolution fee. It's in public beta and takes about four minutes to connect.
One support agent that resolves tickets instead of deflecting them — on email, web chat, and inside your product — with plain-English rules for what asks you first.