A recipe is a proven automation, installed in one click: your AI runs it on schedule with your real data — identified visitors, CRM, revenue, ads, SEO — and reports back. No config, no code, and anything that sends waits for your approval until you say otherwise.
Each morning, scan new and contacted leads and draft follow-ups for the ones that have gone quiet.
List my leads in the "new" and "contacted" stages. For each lead with no recent activity, draft a short, specific follow-up email with send_email (it will wait for my approval). Keep it human — reference why they reached out. Log a note on each lead you touch.
Twice a week, surface qualified leads that have stalled and suggest next steps.
List leads in the "qualified" stage. For any that haven't moved recently, suggest a concrete next step (call, demo, tailored email) and draft the outreach with send_email for my approval. Update stages if any are clearly dead.
Every morning, research fresh inbound leads and prep a first-touch.
List leads in the "new" stage. For each, use web_search to learn about their company, add a lead note with what you found, and draft a personalized first-touch email with send_email for my approval. Move them to "contacted" once drafted.
The core loop as a recipe: when an ICP-fit company visits twice, enrich it, find the decision-maker, and draft warm outreach the same day.
Use get_lead_scores and get_company_enrichment to find companies that visited at least twice in the last 7 days and fit our ideal customer profile. For the top 3 accounts: review their journey (which pages, how often), enrich the company, and draft a short, specific warm outreach email with send_email for my approval — reference what they looked at without being creepy (mention the topic, not the exact pages). Log a note on each account.
Post a Slack message whenever a high-intent company (pricing page, repeat visits) shows up — so sales sees it while the visit is still warm.
Check get_analytics and get_lead_scores for yesterday: which identified companies hit high-intent pages (pricing, demo, comparison) or returned for a 2nd+ visit? Post a concise Slack digest with slack_post_message: company, industry, what they viewed, intent score, and a suggested next step for each. Skip days with nothing notable.
Weekly scan for customers whose engagement dropped — flagged with reasons and a drafted check-in.
Use get_lead_scores to list customers with elevated churn risk and get_contact_timeline to understand why (engagement drop, no recent sessions, support friction). For the top 3, draft a personal check-in email with send_email for my approval — helpful tone, no upsell. Add a contact note summarizing the risk reasons.
Each morning, a one-pager per booked meeting: who they are, their journey on your site, and talking points.
Check contacts with a meeting_booked event today or tomorrow (get_contact_timeline). For each: summarize who they are (company, role, enrichment), their full journey on our site (pages, recordings if any), how they were acquired, and 3 tailored talking points. Email me the brief before the workday starts.
Every Monday, review Search Console performance and summarize the top 3 optimization moves for you to act on.
Run an SEO review for this week with get_seo_insights (and web_search for competitor context if useful). Summarize the 3 highest-leverage optimizations (low-CTR pages, striking-distance queries, content gaps) in this conversation with a concrete brief for each. If GitHub is connected and a fix is code (titles, meta, content), offer to open a PR. Then draft a one-paragraph email summary with send_email for my approval.
Every Monday: spend, ROAS by campaign, and the one change worth making — from real ad-platform + revenue data.
Pull get_ads_stats for the last 7 days and tie spend to revenue with get_analytics (by utm_campaign). Report: total spend, ROAS per campaign, best and worst performer, and ONE concrete recommendation (pause, budget shift, audience change). Email me the summary.
Watch Search Console for queries or pages that lost meaningful clicks week-over-week and explain what happened.
Compare this week vs last week in get_seo_insights. Identify queries or pages with a meaningful drop in clicks or position. For each, hypothesize why (SERP change, seasonality, content issue, technical) — use web_research if a SERP check helps — and suggest a fix. Email me only if something actually dropped.
Track what Smithery, Composio, and other MCP platforms shipped this week.
Use web_search to check what these MCP/agent-tool platforms announced or changed in the last week: Smithery, Composio, Glama, Zapier MCP, Pipedream. Summarize anything notable (pricing, features, positioning) and flag any move we should respond to. Email me the summary.
Ask your happiest customers for a review at the right moment — after real product engagement, not on a random timer.
Use get_lead_scores and list_contacts to find engaged, healthy customers (high engagement, customer stage, no recent support friction) we have not asked for a review. Draft a short personal review request with send_email for my approval — one clear link, genuine thanks, easy no. Track who was asked with a contact note so nobody is asked twice.
People who signed up but never activated get one helpful, human nudge — with context from what they did (and didn’t) do.
Find leads who signed up 2–5 days ago but show no activation activity (get_contact_timeline: no return sessions, key goals not fired). For each, draft ONE helpful follow-up with send_email for my approval — reference where they stopped, offer the single next step, no guilt-tripping. Skip anyone already contacted about this.
On the 1st-of-month cadence, compile revenue, traffic, and pipeline into a report.
Pull together a business snapshot: revenue and customers (from connected metrics), site visitors, lead pipeline by stage, and agent activity/spend this period. Write it as a note titled "Business report" and email me the highlights.
Politely chase unpaid invoices on a weekly rhythm — drafted for approval, never sent blind.
Check Stripe for unpaid or past-due invoices. For each overdue customer, look up their contact and draft a short, friendly payment reminder with send_email for my approval — escalate tone gently by how overdue it is (first nudge vs second reminder). Log a note on the contact.
A Friday recap of what the agents accomplished across the business.
Review the Progress Log and completed tasks from this week across all departments. Write a concise Friday recap — what shipped, what's in review, what's blocked — and email it to me.
A first-of-month draft of your investor update — revenue, growth, pipeline, and highlights pulled from real data.
If this is the first week of a month: compile last month's investor update draft — revenue and growth (Stripe + get_analytics), visitors and conversion trend, pipeline (list_contacts by stage), top 3 highlights and 1 honest lowlight from the Progress Log. Structure: TL;DR, metrics, wins, challenges, asks. Email me the draft to edit.
New here? Create a free workspace — the flagship ICP visitor → outreach recipe is where most teams start.