Signal #10 in #TheWeeklySignal series: Structural Integrity An #agent doesn't just parse an #OpenAPI doc, it reasons over it. Structural integrity measures whether the surface shapes and schema model makes agent reasoning and planning possible. Three APIs. Very different answers. 👇

Signal #9 in #TheWeeklySignal series: Complexity Comfort Before an agent calls an API, it plans. The bigger and deeper the surface, the harder that planning problem gets. Three APIs. Very different answers on how navigable their surface actually is. 👇 Powered by: github.com/jentic/jenti...

Signal #8 in #TheWeeklySignal series: Summary Coverage Before processing #OpenAPI parameters, schemas, or examples, an AI agent reads the `summary` field. That one-liner is the operation's headline. Three APIs. Very different answers on whether it exists at all. 👇

We’re launching our free and useful API AI-Readiness CLI tools on ProductHunt today. All feedback welcome. www.producthunt.com/products/api...

Signal #7 in #TheWeeklySignal series from @jentic.bsky.social: Example Validity Previously, signal #3 asked: do your APIs have examples? Most don't. This week: the ones that do — are they correct? Three APIs, very different answers. 👇

Signal #6 in #TheWeeklySignal series: Authentication Strength An agent doesn't just need to know auth is required. It needs to understand how to authenticate safely: scoped, renewable, machine-readable. Does your OpenAPI description say enough? Three APIs. Very different answers. 👇

Arazzo Specification v1.1.0 is out from the @openapis.org 🚀 The headline: #AsyncAPI support. For the first time, #Arazzo workflows span both synchronous and async #APIs in a single machine-readable document. 🧵

Signal #5 in #TheWeeklySignal series: Response Coverage When an agent calls your API, it needs to understand what comes back. Three examples, very different results. 👇 This week's question: beyond the success case, how much of your API's response landscape have you actually documented?

As #AI agents gain more capabilities, everyone's paying much closer attention to their API estate. But how do you ensure the #APIs you deliver actually cater for agentic consumers? Here's Erik Wilde's top tip for delivering AI-ready APIs

Signal #4 in #TheWeeklySignal series: OperationId Quality An agent doesn't read paths and infer intent. It looks for operationIds , the stable names tool-calling systems, #Arazzo workflows, and SDK generators depend on. Three APIs. Very different results. 👇

Speaking at #APIDays / Future of Software Technologies NYC next week on: 𝗕𝗲𝘆𝗼𝗻𝗱 𝗢𝗽𝗲𝗻𝗔𝗣𝗜: 𝗪𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀 𝗧𝗵𝗮𝘁 𝗔𝗜 𝗔𝗴𝗲𝗻𝘁𝘀 𝗖𝗮𝗻 𝗔𝗰𝘂𝗮𝗹𝗹𝘆 𝗘𝘅𝗲𝗰𝘂𝘁𝗲 I’ll explore Arazzo 1.1.0, including AsyncAPI support. If you’re around NYC and want to chat OpenAPI/Arazzo, or tooling at @jentic.bsky.social, come say hello 👋

Signal #3 in The Weekly Signal series: Example Density A schema tells you the shape. An example shows you what the data actually means. Using @jentic.bsky.social API AI-Readiness Framework Three APIs. Very different results. 👇

It's very nice to be keynoting at APIOps Helsinki on: 𝘼𝙋𝙄𝙊𝙥𝙨 𝙞𝙣 𝙩𝙝𝙚 𝘼𝙜𝙚 𝙤𝙛 𝘼𝙄 𝘼𝙜𝙚𝙣𝙩𝙨 The thesis: If an agent can call APIs, trigger workflows, or decide what to call, then action-layer control is the capability & security boundary, not the prompt. #apiopshelsinki2026 #AIAgents #OpenAPI #Arazzo

Most #APIs weren't designed for autonomous agents. On May 23rd, join Erik & I for a "Designing AI-ready APIs" masterclass. We'll work through the role #OpenAPI, #Overlay & #Arazzo play in improving AI-readiness for APIs. Learn more & Register: www.eventbrite.co.uk/e/designing-...

Signal #2 in The Weekly Signal series: Error Standardisation #RFC9457 First #API interactions rarely start clean. The question isn’t “did it fail?”, it’s, "Did the error tell you enough to recover?" Three APIs. Very different results. 👇

Nice to be part of this @infoworld.bsky.social feature by @doerrfeldbill.bsky.social on building enterprise-grade agentic systems. Hot take: Determine what's adaptive vs. deterministic, and codify the latter. Not everything needs an LLM in the loop. www.infoworld.com/article/4154...

"The latest shift in the API landscape signals a profound transformation toward AI-native infrastructure, as I could see from last week’s news, where the traditional boundaries between human developers and autonomous agents are rapidly dissolving." apichangelog.substack.com/p/2026-16

What actually goes into building enterprise-grade AI agents? My latest for @infoworld.bsky.social explores the emerging best practices behind agentic systems: www.infoworld.com/article/4154...

Big milestone for #Arazzo 1.1 - #AsyncAPI support has been merged into the release candidate. Arazzo will soon span @openapis.org and AsyncAPI. One workflow language for multiple API styles. Huge thanks to the community. Just 136 review conversations over 7 months! github.com/OAI/Arazzo-S...

How do we bring determinism to AI systems that struggle with context and consistency? The answer lies in well-defined, machine-readable workflows, as Frank Kilcommins explains.

Signal #1 in The Weekly Signal series: Description Coverage. Your #API may be perfectly valid. It may also be almost useless to an AI agent. An agent reads your #OpenAPI Description. What it finds there is what it knows. Nothing more. Three APIs. Very different results. 👇

Have you got a problem? An API problem? A generic 400 Bad Request helps no one, not developers, not AI agents. #RFC9457 fixes this. We use it across our @jentic.bsky.social #APIs and just open-sourced reusable schemas + Python/TypeScript packages. #Apache2.0 github.com/jentic/api-problem-details

#LLMs figure things out. But for known capabilities, don't leave business logic to chance. #Arazzo codifies exactly what #agents should do: open standard, precise execution, no vendor lock-in. Discoverability without determinism is just controlled chaos. 👉 nordicapis.com/why-ai-agent... #APIs

"A workflow that can't be executed remotely feels like a loss of potential. I mean, you can still trigger it manually, or even execute it periodically. However, there's no way for it to receive programmatic requests from the outside world." apichangelog.substack.com/p/exposing-w...

Agents are getting more capable. The question now: does it work correctly, safely, predictably — every time? #AgentSkills, #MCP, #Arazzo. Three approaches at different points on the flexibility-to-control spectrum. Join Erik and I April 22nd to learn more. Register 👉 jentic.com/webinar/agen...

"APIs are no longer just passive data pipes, but active, auditable, and intelligent frameworks that bridge the gap between human intent and machine execution." Thanks to @jentic.bsky.social for their support, helping keep an independent API publication alive. apichangelog.substack.com/p/2026-14

Your AI agent nearly gets the API workflow right. Nearly. Vague API docs = wrong tool calls, bad arguments, silent failures. Arazzo is the open standard giving agents deterministic, step-by-step execution so "nearly right" becomes "always right." Watch the full talk 👇 youtu.be/XjbsAdREHxA

#Arazzo Editor is live! Build #API workflows with form-based editing + live diagrams. No YAML required (unless you want it). Below, I cover why tooling matters for standards, and how Arazzo fits in the agent-ready orchestration picture. jentic.com/blog/arazzo-editor #OpenAPI @jentic.bsky.social

Jentic Mini: open-source API execution layer built by OpenClaw for OpenClaws https://github.com/jentic/jentic-mini/discussions/129 (https://news.ycombinator.com/item?id=47602193)

We had a great time hosting and helping 100+ people in Dublin to get set up on OpenClaw. Looking forward to the next one and getting even more people started on OpenClaw with Jentic Mini 🦞🔒

Raising Lobsters 🦞 in Dublin! Packed house at the #Jentic event in partnership with #Give(a)Go and #AICollective! There’s a ground shift! Standing room only to get your #OpenClaw setup securely with #JenticMini Less apps! One #universalagent #aiagents #apis @jentic.bsky.social #raisinglobsters

Dublin startup Jentic launches a free, open source permission firewall for OpenClaw and other AI agents, giving developers credential control, fine-grained permissions, and a single killswitch.

Raising Lobsters 🦞 in Dublin! Packed house at the #Jentic event in partnership with #Give(a)Go and #AICollective! There’s a ground shift! Standing room only to get your #OpenClaw setup securely with #JenticMini Less apps! One #universalagent #aiagents #apis @jentic.bsky.social #raisinglobsters

#ArazzoUI now supports local file upload and drag & drop. Remote URL loading was there from day one, but this was the most requested community feature since launch. Try it out: arazzo-ui.jentic.com Learn more: jentic.com/product/araz... #arazzo #openapi #apiworkflows #jentic

Universal agents are here, and #OpenClaw proves it. These agents eagerly want unfiltered direct access to #APIs. That changes what #API-First means. Can your APIs can be reasoned over by machines, not just called by them? Join our live session next week to learn how to prepare: shorturl.at/rQlc8

Announcing Arazzo UI: Visualize API Workflows jentic.com/blog/announc...

"A workflow that can't be executed remotely feels like a loss of potential. I mean, you can still trigger it manually, or even execute it periodically. However, there's no way for it to receive programmatic requests from the outside world." apichangelog.substack.com/p/exposing-w...

Released: 𝗔𝗿𝗮𝘇𝘇𝗼 𝗨𝗜: 𝗮𝗻 𝗼𝗽𝗲𝗻-𝘀𝗼𝘂𝗿𝗰𝗲 𝘁𝗼𝗼𝗹 𝗳𝗼𝗿 𝘃𝗶𝘀𝘂𝗮𝗹𝗶𝘇𝗶𝗻𝗴 𝗔𝗣𝗜 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀 Renders #Arazzo YAML/JSON as diagrams & docs. No setup required. Review multi-step orchestration logic at a glance. Share workflows without asking teammates to parse raw YAML. See execution paths, not nested objects. #OpenAPI

We've open-sourced the @jentic.bsky.social 𝗔𝗣𝗜 𝗔𝗜-𝗥𝗲𝗮𝗱𝗶𝗻𝗲𝘀𝘀 𝗙𝗿𝗮𝗺𝗲𝘄𝗼𝗿𝗸. Most #APIs were not designs for #AI-systems or agents as the primary consumer. A valid OpenAPI description doesn't mean your API is ready for systems that need to reason, plan, and execute actions safely. youtu.be/mRaVpXCzl30

Here's a little tool called 𝗮𝗿𝗮𝘇𝘇𝗼𝟮𝗼𝗽𝗲𝗻𝗮𝗽𝗶 that converts #Arazzo workflows into #OpenAPI Lets your workflows become deployable to existing #API mgmt tooling 𝘾𝙝𝙚𝙘𝙠 𝙞𝙩 𝙤𝙪𝙩: Playground: frankkilcommins.github.io/arazzo2openapi NPM: www.npmjs.com/package/araz... GitHub: github.com/frankkilcomm...

This week's API Changelog summary includes almost surreal news: - A system that makes it easy for agents to spend online. - Opening anatomical scanning to robots. - Letting agents hire humans. This and more here: apichangelog.substack.com/p/2026-07

Looking forward to speaking at #DeveloperWeek this week (Feb 18th-20th), and participating in the first @openapis.org summit of 2026. “You may have #OpenAPI, but is it AI-Ready?” I'll be exploring what it means to prepare your APIs for agentic and AI-driven uses through 6 key dimensions.

Short walkthrough of our 𝗔𝗜‑𝗥𝗲𝗮𝗱𝗶𝗻𝗲𝘀𝘀 𝘀𝗰𝗼𝗿𝗲𝗰𝗮𝗿𝗱 and diagnostics for #APIs. Most APIs expect devs to fill in gaps in the #OpenAPI description. Expecting AI agents to do that safely is an operational risk. We score six dimensions to show where APIs break for AI consumers. Try at: jentic.com/scorecard

Great round‑up from 𝙏𝙝𝙚 𝘼𝙋𝙄 𝘾𝙝𝙖𝙣𝙜𝙚𝙡𝙤𝙜. The trend of #APIs as the enabling layer for #AI is clear, showing up in real‑time voice, messaging integrations, and driving standardization. @jentic.bsky.social is proud to sponsor and partner with @brunopedro.com. apichangelog.substack.com/p/2026-06

"Looking back at the end of January 2026, it feels like one of those moments where several threads in the API economy quietly tied themselves together." The API Changelog weekly summaries issue is back! Thanks to @jentic.bsky.social for making it possible! apichangelog.substack.com/p/2026-05

Looking forward to speaking at the National Open Source Innovation Summit (Feb 6th), as part of a panel diving into Global Trends in Open Source AI. Great agenda: summit.openirelandnetwork.com/agenda #opensource #jentic #openirelandnetwork #apis4ai

Most APIs are designed for humans, not machines. Here are six dimensions to ensure your systems are interpretable, safe and ready for AI integration. By @jentic.bsky.social Head of Enterprise Architecture @fkilcommins.bsky.social

First post of the year with @thenewstack.io . Having reviews 1000s of APIs, I outline why most APIs break down in AI systems and 6 practical dimensions for making APIs AI-ready. To get more ROI from agents, we have to solve integration (it's the execution layer!). thenewstack.io/why-most-api...

What if you could query the official MCP registry from inside the Arazzo Visual Editor? This demo shows how easy it is to find the Model Context Protocol (MCP) server you want to use on your Arazzo workflow. By using the official MCP registry, you can quickly find the MCP server you're looking for.

For folks straddling #API and #AI lands: we’ve shipped Jentic OpenAPI Tools to #PyPI. A Python toolkit for parsing, validating, and transforming #OpenAPI docs, built with a pluggable backend architecture and strong type safety. Check it out at: pypi.org/project/jent... @jentic.bsky.social