The work. Already written.
Each template is a full agent spec — system prompt, tool picks, schedule suggestions. Clone one, connect your tools, and it runs on its own.
Scans your outbound Gmail daily, finds sent cold emails with no reply past 4 days, and drafts a contextual follow-up — not the same generic "just checking in".
Watches target companies' LinkedIn hiring for buying signals — a new head of growth, a new VP of engineering, a 3× expansion in sales hires.
Finds trial users in Stripe about to expire, drafts a personalised "before you decide" email referencing what they actually did in the product.
Searches Apollo for contacts matching your ICP, de-dupes against HubSpot, and imports the net-new ones as Contacts with a tag so they're ready to sequence.
Given a prospect + your product context, writes 3 variant subject lines + email bodies that sound like a person, not a sequence.
Reads yesterday's cold-email replies from Gmail, classifies them (interested / not / wrong-person / auto-reply / unsubscribe), and drafts responses to the interested ones.
Builds a sales-ready customer profile from a name, email, or company. Combines LinkedIn, Apollo/Hunter enrichment, Reddit sentiment, and public web research into a structured brief.
Given a HubSpot deal, pulls its contacts, enriches each with Apollo + Hunter, and writes a one-page sales brief back to the deal as a note.
Triages an inbound lead (email or form submission), enriches the sender, scores against ICP, and drafts a tailored response.
Given an email, name, or company domain, pulls firmographic + person data and verifies email deliverability.
Given a list of LinkedIn profile URLs, researches each person and drafts a personalized connection request that references something real they recently wrote or shared.
Given an ICP description, returns a shortlist of likely prospects with LinkedIn profile URLs and personalized talking points.
Given a meeting transcript or notes, drafts a polished follow-up email: what we discussed, what I'll do, what you'll do, proposed next step.
Tracks a competitor's mentions, product launches, hiring activity, and pricing changes. Flags noteworthy signals.
Scans Reddit, HN, and the open web for questions your target audience is actually asking — and turns them into tweet/blog/email angles.
Reads the #help channel for the last few hours, drafts auto-responses for repeat questions, and flags novel ones for human review.
Scans the top 30 HackerNews stories daily for keywords in your space and flags the ones worth a comment or a share.
Daily digest of posts, comments, launches, and mentions of your space from HackerNews and ProductHunt. Use as a routine.
Scans today's Product Hunt launches for competitive products and summarizes each — what they do, how they overlap with us, whether we should respond.
Finds threads in your target subreddits where your product or topic is relevant, and drafts a non-spammy reply for each.
Finds new posts mentioning your space (not just your product) and drafts non-spammy comments that help first, mention you second.
Lightweight Reddit research — top posts, subreddit pulse, mentions of a topic. Great for a quick morning brief.
Watches 3–5 subreddits for new posts matching intent keywords ("looking for", "any recommendations") and ranks them by engagement.
Scans Reddit + HackerNews + X for the fastest-rising topics in your niche this week and returns 5 blog titles with outlines.
Reads this week's shipped features or customer wins from Notion and drafts 3 LinkedIn post variants: punchy / story-driven / contrarian.
Pulls the week's shipments, customer wins, and industry news and drafts a 400-word founder newsletter.
Given 2–3 competitor URLs, finds keywords they rank for that you don't, and writes a one-page content brief per gap.
Takes a blog post URL or long-form piece and condenses it into a 6–10-tweet thread with a strong opener and soft CTA.
Monthly founder-to-investor update, grounded in real Stripe + product data. Warm tone, no bullshit — same format as the best YC monthly updates.
Finds overdue Stripe invoices and drafts a polite-but-firm reminder email per customer.
Reads refund request emails, pulls the customer's Stripe history, classifies reason, and recommends a refund amount + response.
Pulls yesterday's Stripe events (subs created / upgraded / cancelled / refunded), computes net MRR delta, posts a morning digest.
Pulls MRR from Stripe, new signups from the users table (via Sheets), and assembles a 1-page weekly metrics digest.
Reads your git commits since the last release tag and drafts a customer-facing changelog grouped by feature / fix / perf.
Scans recent support tickets for churn signals (frustration, "might switch", repeat issues) and posts a watchlist to Slack.
Classifies your unread Gmail into support / sales / personal / spam, drafts replies to the support ones, and posts a one-line digest to Slack.
Scrapes G2 / Capterra / Product Hunt / App Store for new reviews of your product, summarizes sentiment, top complaints, top praise.
Reads the last week of support messages, identifies repeat questions, and drafts 3–5 FAQ entries for the docs.
Given a topic, produces a publication-ready research brief: target keyword, competitor analysis, outline, sources, angle. Stops short of drafting (that's a human call).
Daily digest of Twitter/X mentions of your brand, with a draft reply for each — supportive for positive, helpful for questions, escalate tag for complaints.
Weekly sweep of your HubSpot: finds stale deals, duplicate contacts, missing required fields, and weird data. Proposes fixes — never writes without review.
On the 1st of each month, pulls traffic + ranking data from Google Search Console and Google Analytics, and writes a 1-page narrative report.
Daily check of tracked keywords in Google Search Console. Flags anything that moved ±3 positions or crossed the page 1 boundary.
Daily summary of new Shopify orders — totals, top products, any issues (failed payments, unusual discounts, first-time vs returning buyers).
Put one on the clock in under 5 minutes.
Pick a template, connect the tools it needs, set a schedule. Your agent starts showing up on its own.