It answers on email and your site, looks up the order, remembers the customer, closes what it can within your rules, and hands you only the calls that are yours.
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.
It answers on the inbox and website you already have, resolves the ticket within rules you wrote in plain English, remembers the customer next time, and asks you only when the call is yours.
Most AI support tools are a smarter FAQ: they answer the question they can find in your help center and route everything else to a queue that is still yours to clear. Fasrad's customer support agent is built the other way round. It reads the ticket, looks up the customer and the order in your records, decides what the right outcome is, and does it — the refund, the reissued invoice, the tracking link, the plan change — inside rules you wrote in plain English. The customer gets an answer and an action. You get a one-line summary, or nothing at all, because nothing needed you.
It works on the channels you already run. Give it a support inbox and it answers every email in the thread it arrived in, in the customer's language. Drop it on your website and it handles the same questions in chat, captures who it's talking to, and follows up by email if the answer needs a lookup. Either way it is one agent with one memory: the customer who wrote in about a late delivery in May is recognized when they write again in September, and the reply reads like someone who remembers.
What it does with a typical night's tickets:
The approval flow is the part that makes this safe to run unattended. Rules like 'refunds over $50 ask me,' 'never promise a delivery date we can't see,' or 'anything legal goes to me untouched' become standing instructions the agent applies to every ticket. When a rule fires, you get the prepared action and a button — Approve, Offer exchange, Let me reply — on Telegram or in chat. One tap, and the customer has their answer. Every ticket lands in the built-in CRM as a contact, a summary, and a category, so 'why are we getting so many size refunds?' is a question it can actually answer.
These pages cover the adjacent jobs buyers usually compare before choosing an AI agent.
Email and web chat out of the box — a support inbox it runs itself, or your own inbox where it drafts and you approve, plus an embeddable chat for your site. It's one agent across all of them, so the customer who chatted yesterday is recognized when they email today. Telegram and phone are the same agent too.
It acts. It looks up orders and accounts in your records or through your API, issues refunds and credits within the limits you set, reissues invoices, changes plans, books calls, and files bugs to your team. Answering from your help docs is the floor, not the ceiling.
You write the rules in plain English — 'refunds over $50 ask me,' 'never promise dates,' 'legal threats come straight to me.' When a rule fires, the agent prepares the action and holds it; you get a summary and an Approve button on Telegram or in chat. Everything else it closes on its own, and every ticket is logged so you can review what it did.
Yes. Every conversation is filed in the built-in CRM as a contact with a summary, and the agent keeps memories about people — the credit you gave in May, that they prefer short replies, that their last three orders were all size M. The next reply reads like it came from someone who remembers, because it did.
It hands the ticket to you with the full context — who, what, what it already tried — on Telegram, email, or chat, and tells the customer who has it and when they'll hear back. You reply in your own words, or tap the draft it prepared. No hold music, no 'a team member will get back to you' with no one behind it.
It can run alongside Zendesk, Freshdesk, Help Scout, or Gorgias through their APIs — reading tickets, posting replies, updating status — or replace the helpdesk entirely for a small team, since it keeps its own contacts, ticket log, and categories. Most customers start with the inbox and add the connection later.
$49/month or $490/year — cancel anytime. No per-resolution fees, no per-seat charges, no conversation caps — the same price whether it closes ten tickets or a thousand. It's in public beta and takes about four minutes to connect.