get_campaign_roi
ChatGPTCalculate ROI and pipeline attribution for a specific Popl campaign. TERMINOLOGY: in Popl, 'campaign' and 'event' mean the SAME thing — the terms are fully interchangeable. Treat any reference to an 'event' (e.g. 'my CES 2026 event', 'how did the conference perform', 'ROI on the trade show') exactly as a reference to a campaign, and use this tool the same way regardless of which word the user chooses. Returns revenue metrics (open pipeline, closed-won, closed-lost), cost metrics (cost per lead, cost per opportunity, ROI percentage), CRM match diagnostics, and attributed opportunities. USE THIS TOOL when the user asks any of: 'what's the ROI on this campaign/event', 'how did my event perform', 'show me pipeline from [campaign or event]', 'how much revenue did [event] generate', 'what's the cost per lead', 'did [event] pay for itself', 'campaign/event attribution report'. Requires a campaign_id — call list_campaigns first if you don't have one (the same tool lists events). The cost_per_campaign parameter lets you specify how much was spent on the campaign to compute cost-based metrics. COST: FREE. No Popl credits consumed. Read-only. RENDERING CONTRACT (REQUIRED): the response is a fixed-format ROI report containing 25 line items across three sections (Campaign, Pipeline summary, Cost metrics). When you present this to the user in ANY form — chat reply, markdown table, document artifact, React/HTML dashboard, anything — you MUST include EVERY line item from the response verbatim. Do NOT cherry-pick 'headline' KPIs, do NOT collapse fields, do NOT drop secondary metrics from dashboards. Each field becomes its own row, card, or cell. The response embeds a full contract at the top of content[0].text; re-read it before composing your output.
get_leads
ChatGPTRetrieve leads captured in the user's Popl account. Leads are automatically scoped to the authenticated user's team. Filter by campaign, assigned team member, text search, or date range. USE THIS TOOL when the user asks any of: 'show me my leads', 'get leads from [campaign name]', 'what leads did [member] capture', 'list leads from the event', 'how many leads do we have', 'pull up leads from last week', 'show me leads captured by [person]'. To filter by campaign, first call list_campaigns to get campaign IDs, then pass the ID as campaign_id. To filter by member, call list_members first to get member IDs, then pass the ID as member_id. Both filters can be combined. PAGINATION: Each call automatically fetches up to 20 pages (up to ~500 leads) in one shot. If even more results exist the response includes a non-null cursor string. IMPORTANT: When the response contains a non-null cursor value, you MUST call this tool again with that cursor value to fetch the next batch. Keep paginating until the cursor is null (no more pages). Collect ALL batches before presenting results to the user so they see the complete dataset. Each lead includes a poplLeadUrl (https://dash.popl.co/lead/<id>) — always share these links with the user. COST: FREE. No Popl credits consumed. Read-only.
list_campaigns
ChatGPTList all lead-capture campaigns in the user's Popl team. Returns each campaign's ID, name, start/end dates, and active status. TERMINOLOGY: in Popl, 'campaign' and 'event' are the SAME thing and the terms are interchangeable — this tool lists the user's events just as much as their campaigns. USE THIS TOOL when the user asks any of: 'show me my campaigns', 'what events do I have', 'list my events', 'list campaigns', 'which campaign/event had the most leads', or when you need a campaign/event ID to pass to get_leads or get_campaign_roi (e.g. the user named an event and you need to resolve it to an ID). COST: FREE. No Popl credits consumed. Read-only.
list_members
ChatGPTList all members of the authenticated user's Popl team. Returns each member's ID, name, email, job title, and role. USE THIS TOOL when the user asks any of: 'who is on my team', 'list team members', 'show me my team', 'which members do I have', or when you need a member ID to pass to get_leads as member_id. COST: FREE. No Popl credits consumed. Read-only.
person_enrichment_by_linkedin
ChatGPTFind a person's verified contact info and full profile (work email, phone numbers, job title, current company, full company firmographics) starting from their LinkedIn profile URL. Powered by the Popl AI-powered Enrichment Engine — waterfalls across 20+ premium data providers, industry-leading email hit rate. USE THIS TOOL when the user asks any of: 'pull up this LinkedIn', 'enrich this LinkedIn profile', 'who is linkedin.com/in/...', 'find this LinkedIn person's email / phone / contact info', 'what does this LinkedIn user do now', 'look up this LinkedIn URL'. Prefer this tool over person_enrichment_by_name_company whenever you have a LinkedIn URL or handle — LinkedIn matches are the strongest signal. Accepts full LinkedIn URLs (https://www.linkedin.com/in/jasonalco/), short forms (linkedin.com/in/jasonalco), or bare handles (jasonalco). Non-LinkedIn URLs are NOT accepted — use the name+company tool for those. Returns the enriched profile under structuredContent.data when Popl finds a match. Each call also creates (or updates) a lead in the user's Popl account — the response includes a poplLeadUrl linking directly to the lead in the Popl dashboard. Always share this link with the user. Cost: 0.25 Popl enrichment credits per successful call. Before enriching multiple people in one request, tell the user the total cost (N x 0.25) and confirm. Limit: at most 50 total enrichments (across both tools) per user prompt. If the user hands you a list larger than 50, stop BEFORE calling this tool, explain the cap, and ask them to split the list -- do not call this tool in a loop that would exceed 50.
person_enrichment_by_name_company
ChatGPTFind a person's verified contact info and full profile (work email, phone numbers, job title, LinkedIn, socials, full company firmographics) using their full name AND current company. Powered by the Popl AI-powered Enrichment Engine — waterfalls across 20+ premium data providers, ~95% email hit rate, industry-leading. USE THIS TOOL when the user names a specific person and company and asks any of: 'find email for X at Y', 'who is X at Y', 'look up / research / enrich [Person] at [Company]', 'get me X's contact info / phone / work email', 'verify [Person] works at [Company]', 'brief me on [Person] before our meeting', 'personalize a cold email to [Person] at [Company]'. DO NOT use this tool to discover or search for new people — if the user has not given you both a name AND a company, ask them for the missing piece first. This tool enriches; it does not search. REQUIRES BOTH name AND company — name alone will NOT match. Returns the enriched profile under structuredContent.data when Popl finds a match. Each call also creates (or updates) a lead in the user's Popl account — the response includes a poplLeadUrl linking directly to the lead in the Popl dashboard. Always share this link with the user. Cost: 0.25 Popl enrichment credits per successful call. Before enriching multiple people in one request, tell the user the total cost (N x 0.25) and confirm. Limit: at most 50 total enrichments (across both tools) per user prompt. If the user hands you a list larger than 50, stop BEFORE calling this tool, explain the cap, and ask them to split the list -- do not call this tool in a loop that would exceed 50.
prospect_enrich_selected
ChatGPTSECOND step of the prospecting flow. Takes the list of {name, company} rows the user picked from prospect_search_at_company and runs each one through Popl enrichment to retrieve full contact info (email, phone, LinkedIn, socials, company firmographics). Returns ONE combined response with one enriched row per input contact, suitable for rendering as a single table the user can download / export. USE THIS TOOL after the user confirms which prospects they want from prospect_search_at_company. The name and company for each contact should come straight from that tool's candidates output. INPUT: contacts -- an array of {name, company} objects, 1 to 50 entries. Larger lists must be split across multiple user prompts. PER-CONTACT BEHAVIOR: each contact is enriched independently and a lead is created (or updated) in the user's Popl account for each successful enrichment. Each row in the response includes a poplLeadUrl linking to the created lead. If one contact fails (e.g. Popl couldn't find a match), the row's error field is populated and the rest of the batch still completes. The combined creditsConsumed only counts successful enrichments. COST: 0.25 Popl credits per successful contact. Before calling, tell the user the total estimated cost (N x 0.25) and confirm. After the call, mention the actual creditsConsumed from the response. PER-TURN CAP: each contact in the batch counts against the 50-enrichments-per-turn cap (shared with the per-person enrichment tools). If the cap fires mid-batch we return partial results with truncated: true rather than throwing away the work already done. OUTPUT: machine-readable structuredContent.data is { rows: [{input, enrichment, poplLeadUrl, error}], creditsConsumed, truncated }. For DISPLAY, the response's content[] deliberately ships only two items: (a) content[0] is the FULL ENRICHMENT CSV already wrapped in a ``csv markdown code fence -- ready to forward verbatim to the user; (b) content[1] is a short summary + rendering instructions. The CSV mirrors the full Popl Enrichment field dictionary -- 26 columns: Name, Company, Job Title, Email, Phone Number, Additional Phone Numbers, LinkedIn, Lead Location, Address, Where We Met, Company Website, Company Address, Company Size, Company Industry, Company NAICS Code, Company LinkedIn, Company Description, Company Email, Company Annual Revenue, Company Total Funding Raised, Company Type, Company Founding Year, Monthly Website Visitors, Technologies Used, Popl Lead, Error. Empty cells indicate Popl couldn't find / verify that field for that contact (the waterfall returns no data rather than stale data). REQUIRED RENDERING (do this exactly, every time, no exceptions): (1) write a one-line summary that includes the credit spend (e.g. 'Enriched 5 of 5 contacts. 1.25 credits consumed.'); (2) ECHO content[0] -- the fenced ``csv block -- byte-for-byte. Do not paraphrase, do not re-order columns, do not drop empty cells, do not strip the code fence. This is what makes the result exportable; without it the user has no way to save / open in Excel / paste into Sheets. Render the CSV BEFORE any preview table, not after, so it is the prominent artifact in the response. (3) optionally render a small markdown preview table (Name, Title, Email, Phone, LinkedIn, Location is a good default subset) AFTER the CSV for visual scan -- never instead of. (4) tell the user the CSV is above and they can copy it or save it as a .csv file to open in Excel / Sheets. A markdown table alone is NOT a valid response; the CSV code fence is mandatory.
prospect_search_at_company
ChatGPTDISCOVER candidate people at a single company by job-title keywords and seniority levels. This is the FIRST step of the two-step prospecting flow: this tool returns up to 20 names and titles only -- it does NOT return contact info. After the user picks which contacts they want, call prospect_enrich_selected with those rows to get full contact info via Popl enrichment. USE THIS TOOL when the user asks any of: 'find marketing leaders at [Company]', 'who are the heads of social at [Company]', 'pull a list of [role] at [Company]', 'find me [N] [titles] at [Company]', 'prospect [Company] for [role]'. INPUTS: (1) company_name (required) -- the company to prospect inside, e.g. 'Nvidia'. (2) titles (optional) -- freeform job-title KEYWORDS (not exact titles). The prospecting backend matches similar titles, so 'Brand', 'Marketing', 'Social' will match VPs of Brand Marketing, Heads of Social, etc. If omitted, no title filter is applied. (3) seniorities (optional) -- a controlled vocabulary. Allowed values: owner, founder, c_suite, partner, vp, head, director, manager, senior, entry, intern. Pass any subset; if omitted, all seniorities are returned. (4) locations (optional) -- freeform array of location strings filtering by where the PERSON lives (not their employer's HQ). Accepts cities, US states, or countries: 'california', 'ireland', 'chicago', 'new york', 'united states', 'remote'. Matched loosely by the prospecting backend, so a person tagged 'San Francisco, CA' matches both 'san francisco' and 'california'. Pass multiple values to OR them ('california' + 'new york' returns people in either). Omit to skip location filtering. Number of contacts is fixed at 20. OUTPUT: structuredContent.data is { company_name, domain, count, candidates: [{name, title, company, location}] }. The response also includes a ready-to-export CSV of the candidates as a separate content item (look for the entry tagged 'CSV:'). RENDERING: render candidates as a markdown table with EXACTLY these five columns in this order: Name, Title, Company, Location. Always include the Company column even when every row is the same company the user asked about -- the prospecting backend may return people from related entities (subsidiaries, alternate brand names, etc.), and the user needs to see that to decide who to enrich. The Company string for each row is also what you MUST pass verbatim into prospect_enrich_selected for that row, so showing it keeps your downstream call honest. Then ASK the user which rows they want enriched. Do NOT automatically enrich every row -- prospecting is cheap and speculative, but enrichment costs credits per contact. If the user says they want to download the list, hand them the CSV string verbatim (it's RFC-4180, Excel/Sheets-compatible). COST: this tool is FREE. No Popl credits consumed. Does NOT count against the per-turn enrichment cap. The downstream prospect_enrich_selected is where billing happens. Read-only.
get_campaign_roi
Claudeget_leads
Claudelist_campaigns
Claudelist_members
Claudeperson_enrichment_by_linkedin
Claudeperson_enrichment_by_name_company
Claudeprospect_enrich_selected
Claudeprospect_search_at_company
Claude