Automation

How each platform charges you, where each one breaks, where your data sits, how they handle AI steps, and a plain rule for which kind of business should pick which.

What Make is genuinely good at, the three places it breaks, and the building habits that keep a scenario running after the person who built it moves on.

The plain-language version of running n8n on your own small server: what you get, what someone has to maintain, what it costs, and the setup that keeps the business in control.

Concrete automations for sales, operations, finance, support and HR that businesses of 5 to 200 people actually run, and the pattern that makes all 25 safe to leave running.

The escalation ladder, the tone rules, the stop-the-moment-it-lands rule, and the accounting integration that turns chasing money into a process that runs on its own.

How to route every lead source into one queue, send a genuinely useful first reply within minutes, hand off to a person cleanly, and know exactly what to never let the machine say.

How to pull the numbers straight from the real systems, choose the five that matter, let AI narrate what changed, deliver it where you already look, and solve the trust problem.

Retries, idempotency explained plainly, dead-letter queues, alerting, logs, the "one thing changed upstream" failure, and the monthly review that keeps a business's automations alive.

The four jobs AI does well around accounting (categorisation suggestions, receipt extraction, receivables chasing, anomaly flags), the three it must never do without a person, and how to wire it up so the books stay yours.