Work

What we've built

Three client systems, one for each kind of automation we sell. Clients are anonymised; the work is real.

Lead gen & sales-ops agentsA B2B beverage brand in Spain

An outbound sales pipeline that runs itself

The sales team was spending its week building lists, researching companies, and chasing replies by hand. We built a pipeline that takes a plain-language description of the ideal customer and carries each lead through to a booked conversation, with people approving anything that goes out.

What we built

  • Lead sourcing from Google Maps and Apollo, with AI query expansion and email finding
  • Contact enrichment that verifies decision-makers before they enter the pipeline
  • A qualifier that grades every company 0 to 10 against the client's own ICP documents
  • Personalised email and LinkedIn sequences across several sales reps, gated by an approval step in Airtable
  • AI triage of every reply into meeting, question, or not interested, with a knowledge-base agent that drafts answers and hands off to a person when unsure
  • A nurturing writer that drafts the next email from inbox history, CRM notes, and fresh web research
n8nOpenAIPerplexityAirtableHubSpotInstantlyApollo
Content & SEO ops automationA German SEO and web agency, and its clients

An AI editorial desk that publishes to WordPress

The agency plans topics for itself and for clients in a pharmacy, a tea shop, a facilities company, and a tech brand. We turned that plan into a production line: the agency fills a sheet, and drafts arrive in each client's WordPress ready for an editor.

What we built

  • Research against the current top Google results for each topic
  • Writing by one model and revision by a second, in the client's tone
  • A generated key visual and infographic for every article
  • Posts created in WordPress as drafts, with an email to the editor to review
  • Migration of an older blog into WordPress
n8nOpenAIClaudeGoogle SheetsWordPress
Reporting & ops agentsA Bordeaux wine merchant

A weekly newsletter drafted by an agent

Writing the weekly newsletter meant reading the wine press, picking stories, and laying out a campaign. We built an agent that does the reading and the first draft, and leaves the send to a person.

What we built

  • Scraping eight wine news sources on a schedule
  • Choosing the five strongest stories and summarising them
  • Adding a product highlight and a historical fact
  • Writing the campaign straight into Mailchimp for review and sending
n8nOpenAIMailchimp

More from the same builds

Content & SEO ops

Tracking whether AI assistants recommend your brand

Prompt lists checked on a schedule across ChatGPT, Perplexity, and Google AI Overviews, with brand-mention scoring over time. Paired with an llms.txt generator and MCP servers for Search Console and keyword data.

Reporting & ops

An executive's assistant on Telegram

Voice note in; calendar booked, email sent, CRM checked, or company documents searched. Backed by a knowledge base that refreshes itself from the company's Drive folders.

Content & SEO ops

WordPress monitoring with AI eyes

Security advisories and uptime checks across a fleet of sites, plus a vision model that spots broken pages a status code would miss.

What every build has in common

A human approves what goes public

Outreach, replies, and articles wait for a person before they send or publish.

Built for volume

Batching, waits for rate limits, duplicate checks, and error alerts, so the system keeps working at month three, not just in the demo.

Inside your existing tools

Your CRM, inbox, sheets, and CMS stay the source of truth. The agents work there, not in another dashboard.

Want one of these running for your team?

Tell us which workflow eats your week. We'll say quickly whether it's worth automating.