Blog articles
Notes on automation systems, n8n workflows, AI agents, integrations, content pipelines, and the logic behind making messy processes work reliably.
The accounting mess starts at the payment instrument
May 14, 2026
Credit card transactions don't auto-import into Moneybird — not because nobody built it, but because the card provider controls the data.
Bank feeds are only as good as the person reviewing them
May 13, 2026
The question isn't whether to use bank feeds. It's whether the person running them knows enough to catch what the automation misses.
Small firm, no budget, all manual
May 12, 2026
Small accounting firms have the most to gain from automation and the least resources to start — the fix is bridging gaps, not replacing tools.
Scale changes the tools you need
May 11, 2026
A 30-line bank reconciliation is a different problem than a 3,000-line one. The logic is the same. The approach that works for one breaks on the other.
One engineer, four AI models, one finance SaaS
May 10, 2026
He used Claude for debugging, Gemini for UI, free models for prototyping, and paid OCR when accuracy actually mattered. Each model for the job it's good at.
Shopify's accounting integration works fine until it doesn't
May 9, 2026
Partial refunds, netted payouts, transaction fees, currency conversion. Native integrations handle the clean case. E-commerce rarely is.
When everything is automated, what's left?
May 8, 2026
Once the transactions are flowing and the reports run automatically, the next layer isn't more automation. It's knowing when something's wrong.
She learned Power Query. Then she quit.
May 7, 2026
The most underrated outcome of automating your own accounting work isn't efficiency. It's what you learn about data — and what that opens up.
The extraction took a day. The edge cases took a month.
May 6, 2026
PDF invoice extraction is the easy part. The hard part is what happens when the invoice doesn't look like an invoice.
Month-end close doesn't slow down where you think
May 5, 2026
Most close automation focuses on pulling data faster. The real bottleneck is almost always the approval routing — the back-and-forth that nobody mapped.