Score every demo request against your ICP, update HubSpot and ping the right rep in Slack within a minute of the form submit.
Reps spend the first hour of the day opening demo requests from students, competitors and two-person shops that will never buy. Real buyers wait half a day for a reply while the queue gets sorted by hand. By the time someone has checked the company website, the hot lead has booked a call with a competitor.
In HubSpot, add three contact properties: ai_fit_score (number), ai_fit_tier (dropdown: hot, warm, cold) and ai_fit_reason (multi-line text). Reps will sort and filter on these, so create them before building the workflow.
In n8n, add a HubSpot Trigger node on contact creation, or a Webhook node called from a HubSpot workflow when the demo form is submitted. Pass email, company domain, job title, team size and the "what are you trying to solve" answer.
Add an HTTP Request node that fetches the homepage of the email domain, then an HTML node to extract the visible text. Skip the fetch for free-mail domains like gmail.com and flag those leads for manual review.
Add a Basic LLM Chain with an Anthropic Chat Model node (Claude Sonnet), the scoring prompt below and a Structured Output Parser for score, tier and reason. Set temperature to 0 so the same lead gets the same score twice.
Use a HubSpot node to update the three properties, then an IF node: a score of 70 or above posts to #inbound-hot and assigns an owner by round robin, 40–69 goes into a nurture list, and anything below 40 gets no alert.
Export the scores after seven days and compare them with the meetings reps actually booked. Adjust the ICP rules in the prompt whenever you see two or more misfires of the same kind.
You are a sales development lead for {{company_name}}, which sells {{product_one_liner}}.
Our ideal customer: {{icp_description}} (for example: B2B software companies, 20-500 employees, with a dedicated RevOps or sales ops person).
Score this inbound lead from 0 to 100 for fit.
Form answers: {{form_answers}}
Job title: {{job_title}}
Company website text: {{website_text}}
Rules:
- Students, job seekers, resellers and competitors score below 20.
- Free email domains cap at 50 unless the answers show clear buying intent.
- Do not guess company size if the website does not state it.
Return only JSON: {"score": number, "tier": "hot" | "warm" | "cold", "reason": "one sentence a rep can read in five seconds"}Write a Slack message telling a sales rep about a new hot lead. Keep it under 60 words.
Line 1: {{first_name}} {{last_name}}, {{job_title}} at {{company}} ({{employee_count}} people).
Line 2: Why they scored {{score}}: {{reason}}
Line 3: What they asked for, quoted from the form: {{problem_answer}}
Line 4: One suggested opening question for the first call, based on their answer.
No greetings, no emoji, no bold text.An hour before each discovery call, get a one-page brief on the account, the attendee and recent news, saved to Salesforce.
Every new Zendesk ticket gets a category, priority and sentiment tag plus a draft reply in an internal note before an agent opens it.
Copilot and Claude review each PR for bugs, missing tests and risky changes before a human looks, so reviewers can focus on design.