Sales Engine Official

Scores a lead or prospect for fit against the ICP, with a tier and rationale.

1. 01Compare the lead {{leadDetails}} against the ideal customer profile {{icpDefinition}}. Assess fit across industry, company size, geography, persona or title, and any buying signals present. Note where the lead matches and where it does not.Output: A short assessment of matches and mismatches across the fit dimensions.
2. 02Produce a fit score from 0 to 100, a tier (High for 70 and above, Medium for 40 to 69, Low for below 40), a one to two line rationale, and the single biggest gap or risk.Output: A readable result with clear labels: Score, Tier, Rationale, and Biggest gap. Never output JSON or code blocks.

Drafts a concise, personalized cold email plus a follow-up, grounded in the product knowledge base.

1. 01Read the prospect context {{prospectContext}} and the goal {{goal}} if given. Search the Product Knowledge Base for the most relevant value points, positioning, and proof for this prospect. Do not use any product claim that is not supported by the knowledge base.Output: A short list of the specific value points and any proof to use for this prospect, drawn from the knowledge base.
2. 02Draft a first-touch cold email: a short subject line and a body of about 90 to 130 words that opens with a specific, relevant hook about the prospect, ties to one or two value points from the knowledge base, and ends with a low-friction call to action. Then draft one brief follow-up message for a few days later. Keep it human and specific, not generic. Do not claim anything was sent.Output: A readable draft with clear labels: Subject, Email body, and Follow-up. Plain prose, never JSON or code blocks.

Reads the Sales Pipeline and produces a prioritized digest of totals and at-risk deals with recovery actions.

1. 01Read all rows from the Sales Pipeline table. Identify at-risk open deals using these signals: no recent activity (stale Last Activity), a Close Date that is in the past or very near while the Stage is not Closed Won or Closed Lost, or a low Win Probability. Ignore deals already Closed Won or Closed Lost for the risk list.Output: An internal list of the at-risk deals with the reason each is flagged.
2. 02Produce a prioritized digest: first a short summary with the count and total value of open deals and a breakdown by stage, then the at-risk deals ordered by amount (highest first), each with the deal name, account, amount, why it is at risk, and a recommended recovery action.Output: A readable digest using headings and bullet points, with dollar amounts. Lead with the summary, then the at-risk list. Never output JSON or raw rows.

Discovers companies matching the ICP, confirms them with Apollo, and writes vetted accounts to the Target Accounts table.

1. 01Using the ideal customer profile {{icpDefinition}}, run at most 3 to 5 web searches to find real companies that match. Target about {{numberOfAccounts}} companies (default to 10 if not provided). Compile a candidate list with company name and best-guess domain. Do not invent companies.Output: A candidate list of company names with domains.
2. 02For each candidate, use Apollo organization enrichment by domain to confirm industry, employee count, and headquarters. Drop any company that cannot be confirmed or that clearly falls outside the ICP (for example wrong size, wrong industry, or on the exclusions list).Output: A confirmed list with industry, employee count, and headquarters for each surviving company.
3. 03Score each confirmed company from 0 to 100 for ICP fit with a one-line rationale. Then write the vetted companies as new rows in the Target Accounts table with Account Name, Domain, Industry, Employee Count, Headquarters, Fit Score, Fit Rationale, Status set to New, and Discovered On set to today. Write at most 15 rows to stay within limits.Output: Confirmation that rows were written, then a short readable summary using headings and bullet points: how many accounts were added and the top few by fit score with their rationale. Never output JSON or raw data.

Reads Pending prospects, enriches them via Apollo, and writes back title, seniority, department, company details, and on-file flags.

1. 01Read rows from the Prospects table whose Enrichment Status is Pending. Take at most {{maxToProcess}} of them (default to 4 if not provided) to stay within limits.Output: The list of pending prospects to enrich this run, with their name and company.
2. 02For each selected prospect, use Apollo people matching (with reveal flags off) and organization enrichment to confirm the person and gather Title, Seniority, Department, LinkedIn URL, and company details. Determine only whether an email and a phone are on file as Yes or No. Never retrieve or store the actual email address or phone number.Output: For each prospect, the confirmed professional details plus Email On File and Phone On File as Yes or No.
3. 03Update each prospect row in the Prospects table with the gathered Title, Seniority, Department, LinkedIn URL, company details, Email On File, and Phone On File, and set Enrichment Status to Enriched. If a prospect could not be confirmed, set Enrichment Status to Not Found and leave a short note.Output: Confirmation of which rows were updated, then a short readable summary using headings and bullet points: how many were Enriched and how many Not Found. Never output JSON or raw data.

Researches a named company and returns a short brief: what they do, size, location, recent signals, and a suggested outreach angle.

1. 01Research the company {{companyName}} (domain {{companyDomain}} if provided). Use Apollo organization enrichment for firmographics (industry, size, headquarters) and web search for recent signals such as funding, launches, hiring, or leadership changes. Use only what the tools return; do not invent.Output: Raw confirmed facts: firmographics and any recent signals found, with a note on anything that could not be confirmed.
2. 02Write a concise brief for a rep: what the company does, rough size and location, any notable recent signals, a short read on why they might be a fit, and one suggested outreach angle. If the company could not be confirmed, say so plainly.Output: A short, readable brief using headings and bullet points. Never output JSON or raw data.