Speed-to-Lead Automation Workflow: A Personal Reply in Under Two Minutes
Leads contacted within five minutes convert many times more often than leads contacted the next morning, and almost nobody who works alone can reply within five minutes all day. The workflow takes each new lead from the portal, the site or the CRM, looks up what is already known, qualifies from the enquiry text, drafts a two-sentence reply in the agent's voice with one question, holds it until quiet hours end, sends it from the agent's own address, logs the exchange to the CRM, and texts the agent a three-line summary when the lead answers. A quiet lead gets one nudge; a reply stops the automation and hands the conversation to the agent.
- Built and run for you$197/month
- Custom build$2,000–3,500 one-time
- TriggerNew lead (portal email, site form, CRM webhook)
- Reference stackClaude + n8n
Per agent or small team, up to 500 leads a month, one CRM. Conversational qualification over several turns by SMS, a voice agent for missed calls, or lead routing across a team are a custom build.
Written by Max Zeshut
Founder at Agentmelt · Last updated Sep 14, 2026
The first reply goes out between appointments, hours later, and a third of leads never get one. The ones who do get a template. By the time the agent is free, the lead has heard from someone faster.
Every lead gets a personal first reply in under two minutes, day or night within your hours; the answers land in the CRM; the agent only picks up conversations that have started.
How the workflow runs
Trigger, then 8 steps.
Trigger — New lead (portal email, site form, CRM webhook). A lead email from Zillow or Realtor.com, a website form submission, or a new-person webhook from Follow Up Boss or kvCORE starts the run.
- 01Receive the leadIntake & triggersWebhookName, email, phone, source, the property or page they enquired from, and their message. Portal emails are parsed for the same fields.
- 02Look up the lead in the CRMEnrichment & researchHTTP RequestExisting record, past conversations, stage, and any notes — so the reply does not treat a past client like a stranger.
- 03QualifyScoring & detectionAI AgentTimeline, pre-approval, price band and intent from the enquiry text; 'hot' only when the timeline is near and something specific is named. Nothing is inferred beyond the text.
- 04Draft the first replyDrafting & generationAI AgentTwo to four sentences in the agent's voice, referencing the property or page, with exactly one question. Fair-housing wordlist applied.
- 05Check quiet hours and consentClassification & triageCodeHolds the message until quiet hours end in the lead's time zone; checks that a phone number came with consent before any SMS; skips leads who opted out.
- 06Send the replyWrite-back & actionsGmailFrom the agent's own address, threaded to the lead's email where there is one; SMS as an optional second channel.
- 07Log and hand offNotifications & escalationHTTP RequestNote and stage written to the CRM; hot leads text the agent a three-line summary with the number to call.
- 08Follow up on silenceScheduling & remindersWaitTwo days without a reply: one nudge that offers two showing slots or asks what they are after. A reply at any point stops everything and forwards the thread to the agent.
Each capability page explains the pattern and lists every automation that uses it.
The template is built on n8n. The same steps run on Make, Zapier or plain code; the tool is chosen to fit your accounts, and you own the result.
- 01 Receive the lead —
Webhook - 02 Look up the lead in the CRM —
HTTP Request - 03 Qualify —
AI Agent - 04 Draft the first reply —
AI Agent - 05 Check quiet hours and consent —
Code - 06 Send the reply —
Gmail - 07 Log and hand off —
HTTP Request - 08 Follow up on silence —
Wait
Data it touches
- Portal lead emails or site form
- CRM (Follow Up Boss, kvCORE, HubSpot)
- Agent voice samples
- Calendar for showing slots
- Consent and opt-out records
Guardrails
- Quiet hours: nothing sent 9 pm–8 am in the lead's time zone; messages queue until morning.
- One question per message; the second message asks the next question, never both at once.
- SMS only with a number the lead gave on a form; STOP honoured on the first word and written to the CRM.
- Fair-housing language check on every draft; flagged drafts cannot be sent.
- Approval mode for the first week: drafts go to the agent for a tap before sending.
Standard or custom
When this becomes a custom build.
The subscription covers the standard version — $197/month, setup and maintenance included. Three things move it to a custom build ($2,000–3,500 one-time), delivered into your accounts to keep. If one of them is you, say so on the audit and the plan prices both.
A voice agent that answers missed calls during showings and books the appointment.
Conversational qualification over several turns by SMS, with the consent records that requires.
A team or brokerage with lead routing, round-robin and a voice per agent.
Running the standard version first is the usual path: the exceptions it logs in the first month are the scope of the custom build, with your own numbers. How custom builds work.
Two minutes is the whole product
Everything else in the workflow exists to make a two-minute personal reply safe: the CRM lookup so past clients are recognised, the quiet-hours hold so nobody gets a message at 2 am, the consent check so no text goes to someone who did not give a number, the one-question rule so the reply reads like a person. Speed without those is a spam machine; those without speed is what the agent already has.
The reply is written from the agent's own messages
Three past messages the agent actually sent are the voice sample. The draft borrows their length, their greeting, whether they use the lead's first name, and whether they write 'Hi' or 'Hello'. The result is not indistinguishable from the agent — it does not need to be — but it is recognisably theirs, which is what a lead is comparing it against.
Hand off on reply, not on schedule
The workflow's last job is to get out of the way. Any reply from the lead stops the sequence, forwards the thread and texts the agent a three-line summary — timeline, pre-approval, what they asked. From there the agent is talking to a person who has already answered the first question, which is a better start than most first calls.
Tools in the stack
What it connects to.
| Tool | Role in this workflow |
|---|---|
| n8n | Intake, lookup, timing, sending, logging |
| Claude | Qualification and the reply in the agent's voice |
| Follow Up Boss / kvCORE / HubSpot | Lead record, notes, stage |
| Gmail / Outlook | Sending from the agent's address; reply detection |
| Twilio | Optional SMS to the lead; the hand-off text to the agent |
Building it in Zapier instead? The step-by-step recipe.
Zapier has no importable file, so this is the build guide: the same process as the n8n template and the Make blueprint, one Zap step per line. Expect 10–14 steps; the AI steps use the Anthropic (Claude) app, rules go in Code by Zapier, and every AI step is followed by a Filter that stops the Zap when Claude flags the item for a person.
Before you start
– Keep the values from "Config (edit me)" somewhere the steps can read them — a Storage by Zapier key each, or a Formatter → Utilities step at the top of the Zap: your name, your email, company, time zone, quiet hours (21–8), approver (channel or email), test mode.
– Connect the apps once: Anthropic (an API key), plus Follow Up Boss / kvCORE / HubSpot, Gmail / Outlook, Twilio.
– Turn the Zap on with test mode on; read the first runs in Zap History before switching it off.
Steps
- Trigger — Webhooks by Zapier — Catch Hook (trigger). A lead email from Zillow or Realtor.com, a website form submission, or a new-person webhook from Follow Up Boss or kvCORE starts the run.
- Webhooks by Zapier — Custom Request. GET
https://api.followupboss.com/v1/people?email=…&includeTrash=false(Follow Up Boss API; add the API key header). Existing record, past conversations, stage, and any notes — so the reply does not treat a past client like a stranger. - Anthropic (Claude) — Send Message. Model
claude-sonnet-5, max tokens 2000, temperature 0.2. Paste the system prompt below; the user message is the fields from the previous step.You are the "Qualify" step of the Speed-to-Lead Automation Workflow workflow run by Agentmelt. Task: Timeline, pre-approval, price band and intent from the enquiry text; 'hot' only when the timeline is near and something specific is named. Nothing is inferred beyond the text. Rules: - Quiet hours: nothing sent 9 pm–8 am in the lead's time zone; messages queue until morning. - One question per message; the second message asks the next question, never both at once. - SMS only with a number the lead gave on a form; STOP honoured on the first word and written to the CRM. - Fair-housing language check on every draft; flagged drafts cannot be sent. - Approval mode for the first week: drafts go to the agent for a tap before sending. - Answer only from the data provided in the input. If something is missing, say so in the output instead of guessing. - If this step produces a message or a document, put the full text in `draft` (and a `subject` if it is an email); put a one-line summary in `summary`. - Set `needs_human` true whenever you are unsure, the input is unusual, or a rule above applies; a person will look at it. - Return valid JSON matching the output schema; no prose outside the JSON. Reply with a single JSON object and nothing else: the fields the task needs, plus "needs_human" (true when unsure), "confidence" (0–1) and "summary" (one line).
- Code by Zapier (JavaScript) — parse Claude's JSON. Input
text= Claude's reply; codereturn JSON.parse(inputData.text);so the fields become mappable in later steps. - Filter by Zapier. Only continue if
needs_human(from the previous step) *does not exactly match*true. Everything Claude is unsure about stops here — check Zap History for those, or add a Paths step whose second path posts the item to your approver. - Anthropic (Claude) — Send Message. Model
claude-sonnet-5, max tokens 2000, temperature 0.2. Paste the system prompt below; the user message is the fields from the previous step.You are the "Draft the first reply" step of the Speed-to-Lead Automation Workflow workflow run by Agentmelt. Task: Two to four sentences in the agent's voice, referencing the property or page, with exactly one question. Fair-housing wordlist applied. Rules: - Quiet hours: nothing sent 9 pm–8 am in the lead's time zone; messages queue until morning. - One question per message; the second message asks the next question, never both at once. - SMS only with a number the lead gave on a form; STOP honoured on the first word and written to the CRM. - Fair-housing language check on every draft; flagged drafts cannot be sent. - Approval mode for the first week: drafts go to the agent for a tap before sending. - Answer only from the data provided in the input. If something is missing, say so in the output instead of guessing. - If this step produces a message or a document, put the full text in `draft` (and a `subject` if it is an email); put a one-line summary in `summary`. - Set `needs_human` true whenever you are unsure, the input is unusual, or a rule above applies; a person will look at it. - Return valid JSON matching the output schema; no prose outside the JSON. Reply with a single JSON object and nothing else: the fields the task needs, plus "needs_human" (true when unsure), "confidence" (0–1) and "summary" (one line).
- Code by Zapier (JavaScript) — parse Claude's JSON. Input
text= Claude's reply; codereturn JSON.parse(inputData.text);so the fields become mappable in later steps. - Filter by Zapier. Only continue if
needs_human(from the previous step) *does not exactly match*true. Everything Claude is unsure about stops here — check Zap History for those, or add a Paths step whose second path posts the item to your approver. - Code by Zapier (JavaScript) — Check quiet hours and consent. Holds the message until quiet hours end in the lead's time zone; checks that a phone number came with consent before any SMS; skips leads who opted out. Start from
return { ...inputData };and add the rule. - Gmail — Send Email — Send the reply. From the agent's own address, threaded to the lead's email where there is one; SMS as an optional second channel.
- Webhooks by Zapier — Custom Request. POST
https://api.followupboss.com/v1/events(Follow Up Boss API; add the API key header). Note and stage written to the CRM; hot leads text the agent a three-line summary with the number to call. - Delay by Zapier — Delay For — Follow up on silence. Two days without a reply: one nudge that offers two showing slots or asks what they are after.
Guardrails to keep
– Quiet hours: nothing sent 9 pm–8 am in the lead's time zone; messages queue until morning.
– One question per message; the second message asks the next question, never both at once.
– SMS only with a number the lead gave on a form; STOP honoured on the first word and written to the CRM.
– Fair-housing language check on every draft; flagged drafts cannot be sent.
– Approval mode for the first week: drafts go to the agent for a tap before sending.
Full blueprint: https://agentmelt.com/workflows/speed-to-lead-automation/ · n8n template and Make blueprint: https://github.com/zeshutmax/agentmelt-blueprints
Want this running without building it?
Automation workflow
$197/month
We set up, host and maintain this workflow on n8n and connect it to your tools. Setup included, cancel monthly, you keep the JSON.
Custom build
$2,000–3,500 one-time
Your systems, your rules, your edge cases. A one-off build on Claude and n8n, delivered with documentation and a walkthrough.
Per agent or small team, up to 500 leads a month, one CRM. Conversational qualification over several turns by SMS, a voice agent for missed calls, or lead routing across a team are a custom build.
Also sold one-time, installed in your own accounts, as part of the real-estate agents package — kit $49 · install $249 · package $490.
FAQ
Frequently asked questions.
Does it reply from my address?
Yes. Email goes out through your Gmail or Outlook, threaded to the lead's message. SMS, if you switch it on, goes from your Twilio number with your name.
What if the lead replies at night?
Nothing goes back until 8 am in their time zone. The reply itself is forwarded to you immediately, so you can answer by hand if you want to.
How is this different from inbound lead qualification?
That blueprint is for a sales team: it enriches, scores and routes to a rep. This one is for a person who is the rep: it replies to the lead, holds the conversation until they answer, and hands it to you.
Sources
What this blueprint is built on.
Vendor documentation the steps rely on, and the rules the guardrails are written against.
- Follow Up Boss — API documentation docs.followupboss.com/
- FCC — Telemarketing and robocalls (TCPA) www.fcc.gov/general/telemarketing-and-robocalls
- Twilio — Programmable Messaging docs www.twilio.com/docs/messaging