Production Automations, Built on n8n

Describe an ideal customer in plain English and get back a sheet of verified, enriched leads — hands-off.

1. 1A reviewer describes the ideal customer profile in a form
2. 2Claude converts it into a structured prospect-search filter
3. 3Pull matching people from the data provider, page by page
4. 4Enrich and verify every email before keeping the row
5. 5Write only verified leads to a sheet and send a summary

Combines Claude Opus, a self-hosted Ollama model and live email verification.

Turns every newly-assigned sales opportunity into a researched account brief — automatically, overnight.

1. 1Pull open CRM opportunities assigned in the last day
2. 2Keep only accounts with a website but no insights yet
3. 3Recursively crawl the company’s website (sitemap-aware)
4. 4Gemini reads the pages and extracts business insights
5. 5Write the insights straight back onto the CRM record

Batched — a router feeds the engine 20 accounts per run to cut cost and runtime.

Reads everything the CRM knows about an account and writes the rep a clean, formatted narrative brief.

1. 1Pull a batch of 50 accounts from the CRM
2. 2Bulk-query 6 related record types (activities, opps, cases…)
3. 3Gemini summarizes the full history into a narrative
4. 4A second pass formats it as a clean HTML brief
5. 5Write the summary and HTML back onto the account

Re-architected from ~30,000 down to ~600 workflow calls per month.

When a sales call ends, it finds the right deal, pulls the key moments, and drafts the rep’s follow-up.

1. 1Call ends → the transcript is ingested
2. 2Confirm the host is an active, eligible rep (rep gate)
3. 3Match the call to the right opportunity (pipeline → Gemini fuzzy-match)
4. 4Extract the key moments and next steps
5. 5Patch the deal and draft a follow-up email as the rep

No confident match? It routes to Slack for manual review instead of writing.

Turns a list of leads into five on-brand cold-email drafts each, ready for a human to pick.

1. 1Reviewer submits a sheet of leads and a notify address
2. 2Read the verified-lead tab and prepare a drafts tab
3. 3Generate 5 distinct style variants per lead with Claude
4. 4Write every draft back into the sheet
5. 5Email a confirmation when the batch is done

Five voices per lead — the human chooses, not the machine.

Listens to recorded sales calls and writes qualification signals straight into the CRM.

1. 1The call platform fires a webhook on transcript-ready
2. 2Fetch the full transcript via API
3. 3Claude extracts MEDDPICC qualification signals
4. 4Draft proposal and next-step suggestions
5. 5Push signals to the CRM and notify the rep on Slack

Turns every call into structured, comparable deal intelligence.

A content studio in a workflow: drafts strategy and creative, with human approval at every step.

1. 1Reviewer submits a content brief
2. 2A strategy agent proposes on-brand angle options
3. 3A creative agent turns the chosen angle into posts
4. 4Human approves or sends feedback at each stage
5. 5Approved content is published or handed off

Also drafts daily LinkedIn posts and repurposes Reddit threads into social copy.

Verifies a new agent’s professional license against official records the moment they’re activated.

1. 1CRM posts the new agent’s details to a webhook
2. 2Look up the license against the state authority
3. 3Scrape the official record via browser automation
4. 4Compare name, number and expiry
5. 5Return a verified / flagged result to the CRM

Uses Firecrawl and Skyvern browser agents to read sites with no API.

Runs a new client through full onboarding — subaccounts, folders, fields, emails — from one form.

1. 1Intake form collects client data, mints an ID and tracker row
2. 2Create and configure the CRM subaccount
3. 3Provision custom fields, drive folders and a calendar widget
4. 4Send client and internal onboarding emails
5. 5The orchestrator coordinates each step and tracks state

Eleven coordinated sub-workflows behind a single form.

Every Monday, scrapes a venue’s review pages and emails the owner a clean AI summary.

1. 1List the venue’s review pages across platforms
2. 2Scrape the latest reviews from each
3. 3Summarize sentiment and themes with an LLM
4. 4Format a tidy weekly digest
5. 5Email it to the owner

Deployed per venue across hotels and restaurants.

Cleans, classifies and email-verifies raw lead lists before they ever enter outbound.

1. 1Accept an uploaded list or watch a sheet for new rows
2. 2Claude cleans and classifies each record
3. 3Verify emails through two independent providers
4. 4Merge results and flag invalid rows
5. 5Write the clean list back and alert on Slack

Double email verification keeps bounce rates low.

Analyzes pitch decks and builds investor-ready research briefs on demand.

1. 1Receive a pitch deck or a company URL
2. 2Scrape and extract the key content with Firecrawl
3. 3An LLM analyzes deck, market and funding signals
4. 4Assemble a structured brief
5. 5Return it via webhook

Built for fast first-pass diligence.

Spins up sending domains and mailboxes for a new outbound campaign — from one form.

1. 1Form collects website, sender name and domain count
2. 2An LLM suggests on-brand domain variants
3. 3Register domains and create two mailboxes each
4. 4Configure records and warm-up
5. 5Track everything in a data table

Two mailboxes per domain, provisioned and tracked automatically.

Groups leads by market and timezone, then launches them into the right sending tool.

1. 1Read campaign-ready leads from sheets
2. 2Group by industry, region and timezone
3. 3Batch and route to the right sending platform
4. 4Push into Smartlead / GetSales sequences
5. 5Report the launch on Slack

A manual test path dry-runs the logic without touching live sends.

Researches each prospect and writes a genuinely personal, multi-line icebreaker.

1. 1Pull prospects from a sheet
2. 2Scrape each prospect’s public profile and site
3. 3An LLM writes a tailored multi-line icebreaker
4. 4De-duplicate and quality-filter
5. 5Write the icebreakers back to the sheet

Personalization at list scale, not template mail-merge.

Generates short-form videos from a brief and prepares them for multi-platform publishing.

1. 1A form captures the video brief
2. 2Generate scenes and clips via AI video APIs
3. 3Stitch and render the final video
4. 4Save to Drive and notify
5. 5Hand off for multi-platform publishing

Brief in, finished short-form video out.

Continuously curates relevant, on-topic news into a clean, readable feed.

1. 1Pull from RSS feeds and sources every 6 hours
2. 2An LLM scores and filters each article
3. 3Store curated items in a data table
4. 4Dedupe against what’s already been seen
5. 5Serve a reader page via webhook

A self-updating reader page, no manual editing.

Posts scheduled, data-driven comments and updates into project tasks automatically.

1. 1Read the schedule and config from a sheet
2. 2Filter which tasks are due an update
3. 3Compose the comment content
4. 4Post the comments to the project tool
5. 5Log and respond with the result

Keeps boards current without anyone touching them.

Lets an AI phone agent book, check and cancel appointments on a live calendar in real time.

1. 1The voice agent posts the caller’s request
2. 2Branch on book / check / cancel
3. 3Read or write Google Calendar
4. 4Confirm by email
5. 5Respond to the agent in real time

Turns a natural phone conversation into real calendar actions.

A queue-based recursive crawler that maps a company’s entire site — the engine behind enrichment.

1. 1Read the site’s sitemap and seed the crawl queue
2. 2Recursively discover and fetch pages, breadth-first
3. 3Fall back to a homepage crawl when no sitemap exists
4. 4Hand the cleaned pages to an LLM for analysis

Costed and tuned at scale — ~$0.036 per account across 30,000 accounts.

The matching brain: reliably links any call to the correct opportunity through a confidence waterfall.

1. 1Match on a calendar event in a ±30-min window (highest confidence)
2. 2Fall back to the host rep’s open pipeline
3. 3Gemini extracts the prospect name and fuzzy-matches the account
4. 4Shadow-log the result for accuracy comparison — no CRM writes

Runs in shadow beside the live pipeline to prove match accuracy.