A short piece of fiction, written by API Evangelist. The reader is imagined; the post, and every number and quotation from it, is real: "A Quick Read On Engineering Maturity From The Outside," Kin Lane, November 11, 2026, https://apievangelist.com/2026/11/11/a-quick-read-on-engineering-maturity-from-the-outside/
Marisol ran vendor reviews for a regional freight cooperative, and she read the post on a Wednesday morning with a sales call booked for two o'clock. The vendor was a dispatch-software company whose deck she had skimmed twice. Everything in it was adjectives.
She bristled at the first paragraph. Someone had looked up their own company, their competitors and the platforms from their old jobs, and decided a score could stand in for engineering maturity. It sounded like the sort of shortcut that lets people feel clever without doing the work. She had spent eleven years learning that vendors who demo well and vendors who ship well are often different companies.
The post seemed to know that, though. It said plainly that the history alone does not decide it, and that Oracle and IBM sit in the strong band at 64.5 and 66.4. It also said Oracle's Agent Readiness of 61 beats everyone in the table except Cloudflare. That took some air out of her suspicion. The author was calling it a pattern and not a rule, and she noticed that he kept saying so.
Then she hit the part that made her uncomfortable. The list of things a high score requires was machine-readable contracts, a changelog, a status page, stated rate limits. It was unglamorous, and her own cooperative's internal tools had none of it. She thought about the portal her team maintained and how little of it anyone outside could see. By the post's logic, her own people would barely register. She sat with that for a minute before reading on, and was relieved to find the author had anticipated her. A company can have excellent engineering and share almost nothing, because its APIs sit behind a customer login. The post called a low score a question, in the words "A low score is a question." It was not a verdict on the people who work there.
She still did not fully buy it. A changelog can be tidy and the software underneath can be rotten, and the post admitted it measures APIs and not the product. She wrote that objection on a sticky note and left it on her monitor, because it seemed the honest thing to keep.
But one line was cheap enough to act on. The post said to take "two minutes on its provider page before the first sales call." She had two hours. She opened the vendor's page, found the facet breakdown and the date of the last changelog, and copied both into her call notes without deciding what they meant. The score looked low to her, but she resisted treating it as an answer.
Instead she used the post's advice. Under the number she typed a single line for the call: ask them to show me what I could not see. At two o'clock, when the salesperson finished the deck, she would ask for a walkthrough of their changelog and their published contracts, then see how they responded.
She peeled the sticky note off her monitor and pressed it beside the notes, so both were in front of her.