Most of what ships as "agentic AI" right now is an automation with AI sprinkled on it, and I think "agentic workflow" is a contradiction in terms. A workflow is deterministic. Known inputs, known outputs. If you've already drawn the path, the thing walking it isn't making decisions.  I went on the @ainativedev podcast with @sjmaple to argue that, and a few other things I've been chewing on. That the line between autonomous agents and supervised ones is a trust line, not a technical one. That you haven't needed to hire junior developers for their coding skills for about a year now, and that you should absolutely keep hiring them anyway, because otherwise nobody becomes an architect. And that software agencies are heading towards something closer to a law firm than a SaaS startup. Also, the story about how I don't know my wife's phone number 🫣 (link to full video in the first comment)

Simon Maple meets @Snyk's Krzysztof Huszcza at @aiDotEngineer for a live, unscripted skill review — and the bet that follows involves a hat, some Tessl swag, and a real production skill on the line. Find out if it holds up to @Anthropic's best practices. Watch the full video at tessl.co/c89

Turns out, agent skills can "fail" before they're ever used. This happens when they're suboptimal in structure and intent. I went through a @tessl_io Skills Clinic with @sjmaple and we put @QodoAI's PR Resolver skill through an automated quality review! A few targeted changes moved the quality score from 78% to 89%. We also got into an interesting question: should AI write the skills that tell AI how to work? Watch the clinic: piped.video/watch?v=Imw9nfui…

Claude Code: The Six-Reaction Origin Story It started as a side project that got six reactions on Slack. A year later, it's writing the majority of Anthropic's product code. Lamis Mukta from Anthropic tells the origin story of Claude Code, and why the "worst" launches sometimes become the biggest ones. Watch the full episode at tessl.co/00r or listen wherever you get your podcasts.

Most teams have code reviews, CI pipelines, security checks, and deployment processes. Then they hand their coding agents a collection of prompts and skills with none of that. Join Simon Maple at AI Engineer SF for a session on why skills are becoming a critical part of the software stack, and why they deserve the same engineering discipline we've spent decades building around code. June 30, 2026 • 10:30am - 11:30am at Booth L-G48 Learn what happens when agent skills are versioned, reviewed, evaluated, and managed like production code, and why that shift is becoming essential for teams building with AI. See you at the Tessl booth.

Almost everyone wrote his own agent skill at this point, so no wonder we arrived here: a package manager for skills - tessl. Might be worth checking out, thanks @sjmaple #devoxxPL #goodtobehere

Skills are the new code... 🧠 I've always thought that the more context you give your coding agents, the better. Turns out there is a sweet spot, a "goldilocks zone" of context, as @sjmaple put it at @DevoxxPL. Pretty wild to see the difference a well-written skills file can make on an agent's output. Shout out @tessl_io! too bad the room was so small.

"𝐂𝐨𝐝𝐞 𝐩𝐫𝐨𝐝𝐮𝐜𝐭𝐢𝐨𝐧 𝐢𝐬 𝐧𝐨 𝐥𝐨𝐧𝐠𝐞𝐫 𝐭𝐡𝐞 𝐛𝐨𝐭𝐭𝐥𝐞𝐧𝐞𝐜𝐤." In our latest AI Native Dev conversation, Ryan Lopopolo (@_lopopolo) from @OpenAI explains why engineering teams need to rethink where they spend their time. As coding agents get better, the highest leverage work shifts from implementation to shaping the environment around it: defining clear interfaces, capturing team knowledge, creating feedback loops, and building systems that help agents make good decisions consistently. Ryan shared how his team went from roughly 3.5 PRs per engineer per week to around 70 as models and workflows improved. The hard part is no longer generating code. It's creating the context, guardrails, and workflows that turn code generation into reliable software delivery. Watch the full conversation or listen on YouTube, Spotify, or Apple Podcasts. Links in the comments.

The most expensive model in the benchmark wasn't the best value. Rob Willoughby and Simon Maple ( @sjmaple ) evaluated 19 model configurations on real agentic tasks and found that DeepSeek V4 Flash scored 82.3 while costing just $0.0236 per task. Claude Haiku 4.5 scored 82.9 at roughly four times the cost, while DeepSeek V4 Pro scored 85.3 at nearly eight times the cost. The interesting part isn't that Flash beat stronger models. It didn't. The interesting part is how little quality was gained for how much additional spend. That becomes a very different conversation once you're running agents at scale. A model that looks marginally better on a benchmark can end up costing dramatically more over the course of a year, especially when agent workloads start growing. The benchmark also surfaced something that many teams probably aren't measuring closely enough. The biggest performance jump didn't come from switching models. It came from adding the right skill. DeepSeek V4 Flash moved from 64.1 to 82.3 with skill context applied, which raises an uncomfortable question about how much of agent performance is actually model selection versus everything built around the model. The full breakdown is worth reading, particularly the sections on points-per-dollar, turn counts, and why the cheapest model in the benchmark ended up being one of the most interesting. Read the full blog here: tessl.io/blog/same-quality-a…

Ryan Lopopolo tracked PR throughput on his OpenAI team from 3.5 per engineer per week up to 70 — not through adding headcount, but through iterating on the model and the harness together. Every revision of GPT-5 from 5.2 onward compounded on the last, and this clip shows exactly what that felt like from inside the team. Watch the full episode at piped.video/MFQIKbr1IEo or listen wherever you get your podcasts. #AI #agenticcoding #claudecode #codex #AIskills

Developers using AI tools are creating and merging twice as many pull requests — but AI-generated PRs have a 60/40 merge rate compared to 80/20 for humans. That gap reveals something important about how agents are actually being used in the wild: probing, experimenting, spawning throwaway work. Jellyfish's Nick Arcolano breaks down what the data actually says. Watch the full episode at piped.video/GbHfzFcIa0o or listen wherever you get your podcasts #AI #agenticcoding #claudecode #codex #AIskills

Hanging out with @gAmUssA @jbaruch @shelajev @sjmaple at @ainativedev #aidevcon #ainativedevcon

AI Native DevCon is live. After an opening keynote from Simon Maple ( @sjmaple ), Guy Podjarny ( @guypod ) has just taken the stage with "Skills are the New Code". As software development shifts from writing instructions to defining intent, skills, specs, and context are becoming the core building blocks of modern engineering. If you're building AI-native systems, this is exactly the conversation happening right now. Can't join us in London? Watch live from anywhere: piped.video/watch?v=akZ85mG5…

Kicking off @tessl_io AI DevCon with @sjmaple Energy is great and it's a full packed conference audience! ⚡

Your eval leaderboard can change completely depending on which model grades the answers. Simon Maple ( @sjmaple ) reran the same benchmark suite using three different LLM judges: Sonnet, GPT-5.5, and Opus-4-7. Nothing else changed. The tasks, rubrics, scenarios, and model outputs were identical. The scores were not. One model moved by 47 points on a single skill depending on the judge. gpt-5.3 ranked near the top under Sonnet, then dropped sharply under GPT-5.5. Opus consistently scored itself higher than the other judges did. GPT-5.5 turned out to be dramatically stricter overall, averaging almost 7 points lower than Sonnet across the benchmark. What makes this especially interesting is that the instability wasn’t evenly distributed. Tasks with concrete pass/fail conditions stayed relatively consistent across judges. But as soon as the rubric involved interpretation, structure, writing quality, or “best practices”, the variance widened fast. Two judges could look at the exact same output and disagree by double digits on whether the model had actually solved the task properly or just approximated it convincingly. That has pretty major implications for how people read benchmark charts right now. A lot of public evals are presented as if the number is objective, when in reality the scoring model itself is shaping the outcome. In some cases, the judge preference is large enough to reorder the leaderboard entirely. The interesting exception was Opus. It stayed in first place regardless of which model acted as the judge. Everything below it shifted around. Read the full breakdown here: tessl.io/blog/your-benchmark…

The problem is not that AI has not been trained... the problem is that AI has been trained A LOT. @venkat_s shares his thoughts on "Accelerated Inference". Watch the full episode at piped.video/C0OeWkbhiL8 or wherever you get your podcasts. #AI #agenticcoding #claudecode #codex #AIskills

We audit packages. We audit infra. We don’t audit the instructions shaping our AI agents. That’s the problem tessl-audit is tackling. In a new post, Simon Maple ( @sjmaple ) introduces an open-source CLI that scans the skills and plugins loaded into your agent’s context for security findings, quality issues, and actual task uplift. Because agent skills are not just “extra context”. They directly influence how models reason, which patterns they follow, and what they decide to do. The workflow is intentionally simple: npx tessl-audit The tool reads your tessl.json, fetches registry data for installed skills, and produces a posture report showing risky plugins, weak guidance, and skills that have never been evaluated against real tasks. What makes this interesting is how familiar the problem suddenly feels. AI agents are starting to develop their own dependency graphs, except the dependencies are prompts, policies, evals, and reasoning layers instead of libraries. And unlike broken code, bad context often fails quietly. The post also walks through optimizing low-quality skills, generating eval scenarios, and measuring whether a plugin genuinely improves agent performance before trusting it in production. Read the full post here: tessl.io/blog/stop-trusting-…

I’m honestly not sure I could sleep on a pillow that was untitled… this would keep my mind too active #bedSHEETS

Simon Maple’s ( @sjmaple ) benchmark (1,742 tests): 5.5 vs 5.4 → tied with skills (89.4 vs 89.3) $0.49 vs $0.30 → +63% for +0.1 Only win: speed (89s vs 135s) Read more here: dev.to/tessl/gpt-55-is-opena…