The problem this solves
Inbound emails from prospects arrive in the one place with the least context: an inbox. The sender's name and signature say who they are; nothing says how big their company is, whether it fits, what it runs or whether the request is worth an hour of a founder's day. The reply is written from instinct, and the instinct is right about half the time.
Forwarding the email to an agent address turns the inbox into a research desk. The Personalization Agent identifies the sender and their company, enriches both, scores the fit against your ICP, and replies to you - never to the prospect - with a short brief and a draft response written in your voice that engages what the prospect actually asked. You read both, edit the draft if you like, and send it from your own account. The whole loop happens in email, with nothing to log into.
Because the reply goes back to you, the prospect never receives anything you did not send, and the brief gives you the grounds to say no politely to the requests that do not fit.
How the mission runs
- Receive the forwarded email. The Personalization Agent receives the email at your workspace agent address, separates the forwarded message from your note, and identifies the original sender's name, address and signature details.
- Research the sender and the company. The sender's domain is resolved to a company and enriched with headcount, industry, HQ, stack and recent signals; people search confirms the sender's role and seniority. Personal-domain senders are resolved from the name and signature where possible.
- Score the fit. The company and the sender's role are scored against your ICP from 0 to 100 with reasons, and the request in the email is classified - a demo, a partnership, a support question, a vendor pitch - so the brief says what kind of reply is warranted.
- Draft the reply in your voice. A response is written that answers the prospect's actual question, references one relevant fact about their company where it helps, proposes the next step that fits the score - a call for a strong fit, a resource for a weak one - and reads as you.
- Reply to you with the brief and the draft. An email comes back to you, in the same thread, with a five-line brief - who they are, company size and fit score, what they asked, the suggested next step - and the draft reply ready to copy or forward, with the sources for the brief linked.
The prompt
This is the exact objective the agent receives. Swap the obvious placeholders for your own domain, segment or channel and run it as-is from the console, Slack, or the API.
What comes back
A reply to your forwarded email containing a brief on the sender - role, company, size, industry, stack, recent signals - a fit score with reasons, a classification of the request, and a ready-to-send response written in your voice that engages what the prospect asked. Nothing is sent to the prospect; you send the reply from your own account after reading both.
Make it yours
- Have the brief and draft posted to a Slack DM instead of emailed back.
- Log the sender as a contact in HubSpot with the score and the brief as a note.
- Ask for two drafts - one for a strong fit, one for a polite decline - and choose.
- Extend it to a shared inbox so every inbound lead email gets the same treatment.
Frequently asked questions
Could the agent accidentally reply to the prospect?
No. Replies go only to the address that forwarded the email; the prospect's address is never used as a recipient by the agent.
What if the sender used a personal email?
The agent resolves the person from the name and signature where it can, states its confidence in the brief, and writes the draft in a way that does not assume facts it could not confirm.
Does it work with Zoho Mail or other clients?
Yes. Forwarding is client-independent; the agent address receives from any mailbox, and the reply arrives wherever you forwarded from.