Blog Posts
How long can you actually stay in the sun? It depends on your skin type and today's UV
"Wear sunscreen when the UV index is high" is the advice everyone gives. It's also not a real answer to the question most people actually have: how long can I be outside right now before I need to…
read more
Choosing a product categorization API: five questions that separate grounded from guessing
- ACJ Labs
- Engineering
- 26 Jul, 2026
If you're mapping a product feed to the Google Product Taxonomy, you have more options than you did a year ago — a raw LLM call, a handful of categorization APIs, a feed tool with categorization…
read moreChoosing a receipt extraction API: six questions that separate tax-aware from just-OCR
- ACJ Labs
- Engineering,Tax
- 26 Jul, 2026
Here's the honest version of the pitch first: if all you need is vendor, date, and total off a receipt, you should probably build it yourself. A single vision-model call — GPT-4o-mini, Gemini Flash…
read more
A wrong ABN and a smudged one look the same to OCR. We made sure ours doesn't guess.
- ACJ Labs
- Tax,Engineering
- 25 Jul, 2026
Every Australian Business Number carries a checksum — a modulus-89 formula baked into the digits themselves, so a wrong ABN is usually mathematically detectable, not just "looks a bit off." Which…
read moreBuild vs. buy: should your bookkeeping tool parse receipts with a raw LLM call?
- ACJ Labs
- Engineering,Tax
- 25 Jul, 2026
Here's the honest version of the pitch first: if all you need is vendor, date, and total off a receipt, you should probably build it yourself. A single vision-model call — GPT-4o-mini, Gemini Flash…
read moreHow accurate is "just ask the model to categorize it"? We measured — 1–2 of 12 checks.
- ACJ Labs
- Engineering
- 25 Jul, 2026
If you're mapping a product catalog into Google's Product Taxonomy, the obvious first move is to hand the title and description to a capable model and ask for the category. It reads well in a demo.…
read more