— DRAFT · ENGINEERING
Why our flagship SaaS runs on a Laravel monolith
Three years and 312 active integrations later. The "you should split this up" timeline didn’t materialise — and we’d still pick the monolith today.
— SOON07 · JOURNAL
Short writing.
Tight thoughts.
Engineering teardowns, opinions, postmortems. Some hand-written by the studio, some shipped by FeedPulse's blog automation against the same brand voice rules.
SECTION 01JOURNAL · LIVE POSTS
01 / WRITING
What we’ve been thinking about.
Newest first. Hand-written essays sit alongside posts FeedPulse ships from the same brand-voice ruleset — flagged AUTO so you can tell them apart.
TEARDOWN12 MIN
The 312-feed problem
How we got Google Merchant Center to stop mass-disapproving products at 3am.
2025-11-04READ →
OPINION6 MIN
Why we still build on WooCommerce
It's deeply unfashionable. It's also the right tool for two of our biggest clients.
2025-09-22READ →
POSTMORTEM18 MIN
Operating a small SaaS, badly, for two years
Lessons from running FeedPulse before it was a real product.
2025-07-10READ →
SECTION 02JOURNAL · DRAFTS IN FLIGHT
02 / ON THE WAY
Writing that's nearly ready.
The four below are the next four to land. We post when a piece earns its place, not on a schedule.
— DRAFT · FEEDPULSE
Anomaly detection that only pages humans
Most "anomaly detection" is alert spam dressed up. The discipline isn’t the model — it’s the threshold and the routing.
— SOON— DRAFT · PROCESS
Friday demos as the only meeting we run
A 3-minute Loom replaces an hour of standups. Three years in, no client has asked for the standups back.
— SOON— DRAFT · STACK
When we say no to a tech we like
Solid, Bun, Hono — all good, none in production. The community-size filter, expanded.
— SOONWork with us
Show us your stack.
We’ll tell you exactly what we’d ship.
One call. We'll dig into what you've got, where it's friction, and whether we're the right people to help. No deck, no pitch.
We reply within 4 working hours · Currently booked out 14 days