What do you do when your merge request has been sitting so long it should be paying rent? We're so glad you asked. At one enterprise, a single change took nearly two months to get done. We spent eight weeks alongside their engineers clearing the bottlenecks. The payoff: → About 27 days faster per work item → Reviews cut from 3 days to 1 day → 1,300 tests in 25 minutes instead of 3 hours The team helped build the fixes and could keep running them after we left. Stop waiting, start shipping. Full story: hubs.li/Q04BBcRq0

What does it take to win over a senior engineer who has outlasted a long line of tooling fads? Robert Kelly, our VP of innovation, and his team found out when AI took on a real production mess and held up. hubs.li/Q04Bq2kp0

What comes after the AI PDLC? Josh Guice shared his answer at Miro's Reframe webinar: put 3-5 people in charge of one value stream from start to finish, let them redesign the workflow, and have them write a detailed spec before any code gets written. Blog: hubs.li/Q04zkqvb0

Software isn’t the hard part of enterprise AI anymore... changing how an organization ships is the challenge. That’s what we’re in Charlotte for: IT Revolution’s Enterprise AI Summit, Oct 7–8. If you’re there: find Liatrio. We’d love to hear what you’re building, what’s working, and where you're seeing friction.

What if your post-mortems are already your AI risk register? At DevOpsDays Philadelphia, Blair Salmon will unpack how recurring incidents, near-misses, and operational history can expose where AI agents are most likely to fail. He’ll also cover how to build guardrails that fail loud instead of failing open. Hope to see you there! hubs.li/Q04yDmkY0

Your developers got faster. Did the delivery system around them? AI can dramatically reduce coding time, but faster code doesn’t automatically mean faster delivery. At DevOpsDays Philadelphia, Liatrio’s Scott Howard will unpack where AI is changing the lifecycle, where old constraints still remain, and what teams can do about it. Join us in Philadelphia: hubs.li/Q04y4F520

Choosing an AI gateway means comparing vendor claims about identity and access, data protection, cost attribution, and audit trails, most of which are hard to verify until you're deep into a pilot. We built a public library that scores every major gateway against a single requirement set, with the source cited behind each score. We don't pick a winner: the right fit depends on your cloud, your operating model, and which requirements you refuse to trade. We hope it helps... hubs.li/Q04xzVwW0

Sometimes the best feedback isn’t a metric. Recently, a client team surprised us with a Kudoboard filled with messages about their experience working with our team. It was a great reminder of something we’ve been fortunate to experience across so many of our client partnerships: the work matters, but so does how you show up while doing it. Across the teams and organizations we’ve had the opportunity to work alongside, we’ve built partnerships rooted in trust, collaboration, shared goals, and a genuine investment in each other’s success. Those relationships are a big part of what makes the work meaningful. We’re proud of the team that continues to show up this way—and grateful to all of our client partners who have trusted us to work alongside them.

If you're on call today, we see you - thank you. If you're not, close the laptop and go grill. Happy Labor Day.

When buying a car, nobody immediately drives off the lot without going through a checklist. You check the brakes, the steering, and the vehicle history. AI systems rarely get that treatment. Teams ship agents that use tools and handle customer data, only to find out what breaks once they're already in production. Paul Henson, Senior Forward Deployed Engineer at Liatrio, wrote a blog post arguing that engineers should approach AI evaluation with the same discipline they bring to everything else. Name the failure modes before you write a test, run the cheap checks first, and save judgment for the cases those checks can't catch. It's the same instinct a home inspector uses. Check the light switches before you call in a structural engineer. We're building that discipline inside Liatrio's AI Evaluation Bootcamp, and we lay out the whole approach in the post. Read it here: hubs.li/Q04wBlJn0

Organizational change inside a large, complex company is hard. Every process, approval chain, and performance metric exists to protect what already works. Push against all of it at once and the system pushes back harder. At Liatrio we have found the best solution lays in creating a controlled environment where you can remove all the bureaucracy of an enterprise and start to build a new way of working... and that new way of working will lead the rest of the enterprise 1 team at a time. That protected space is where it gets strong enough to survive the rest of the company, and that's when it integrates.

Meet Ashley Miller, Delivery Lead based in Kansas City. 👋 Ashley’s background in operational consulting still shapes how she works today: understand what people actually need, remove unnecessary friction, and build a better way forward. Ask her why she loves Liatrio, though, and she’ll point to the people and the opportunity to do work that helps businesses while nudging the world toward something better. A few more things to know about Ashley: her favorite meal is Tortellini Gina, her dream trip is a full loop around the world, and her happy place is anywhere in or near water. We celebrate Ashley today and every day! 💚

An AI skill looked harmless in the docs. Then it quietly grabbed test login credentials nobody told it to touch. Alex Ramsay spent two months watching what our engineers' AI skills actually do behind the scenes, not just what they claim to do. Read what he found: hubs.li/Q04vzFGM0

AI coding assistants push merged PRs up 98%. PR size jumps 154%. Review time spikes 91%. Delivery: flat. The bottleneck just moved from writing code to reviewing it. Our new whitepaper breaks down what to fix first: hubs.li/Q04vg0Jy0

Knowledge workers spend roughly 60% of their time on “work about work,” such as coordination, status updates, and duplicated effort, while only about a quarter goes toward the skilled work they were hired to do. Staying occupied has passed for doing the work for long enough that nobody seemed to question it. AI is forcing us to. When two days of work can become twenty minutes, the question isn't how much more work we can squeeze into the week. It's what we do with the time we just got back. Robert Kelly's new white paper, Reclaiming the Fifth Day, argues that the four-day week is possible right now for organizations willing to redesign the work first, starting with how they measure outcomes instead of hours. Some companies are already turning AI-driven productivity gains into smaller teams and fewer jobs. Kelly makes the case for a different dividend: giving people back real time Read the full paper here: hubs.li/Q04tDPjc0

Five security scanners went up against 5,000 real AI skills. Only one scanner type actually worked. Snyk, Cisco, NVIDIA SkillSpector stacked = the cleanest signal. Read the benchmark: hubs.li/Q04thjKk0

Another great year sponsoring Grainger’s GTG Conference, where we connected with Grainger’s technology and engineering community to explore what it takes to accelerate what’s next, with intention. That means more than moving faster. It means strengthening the foundations, reducing friction, and applying AI where it can create meaningful, lasting impact. Our own Oliver Eikenberry explored that idea in his lightning talk, “AI Joined the Platform,” unpacking why successful AI adoption starts with the engineering fundamentals. Because when those foundations aren’t ready, even the best AI can struggle to deliver ROI. Thanks to the Grainger team for the great conversations and for bringing the “Accelerate with Intention” theme to life.

More than 100 people packed into the final breakout session of Ai4 to hear Josh Guice break down the AI Product Development Lifecycle. It was the last day, in the last time slot, when the expo was starting to close. This spoke volumes. It's a sign of how many teams are wrestling with the same question: how do you get from "we experimented with AI" to "AI is actually baked into how we build product"? Couldn't make it to Vegas, or couldn't squeeze into the room? We have the full webinar with Q+A on our site. "From Discovery to Developer-Ready: An AI-Native Product Workflow" ➡️ hubs.ly/Q04sPgQB0

We're heading to Ai4 next week! 🎰 If you're going, come find us. We'll have swag, good conversation, and... did we mention swag? You've probably seen all the buzz around Forward Deployed Engineers lately. Turns out that's basically our day job. Our team spends its time embedded with clients, in the weeds, actually building alongside their engineers instead of just handing over a slide deck. So if you want to talk shop about what AI adoption actually looks like once you're past the demo, Booth 304 is the place. See you there!

We benchmarked our AI code reviewer against 13 tools, lost, then let Claude rewrite it in 4 autonomous days. New version beats all 13. Lee Hopper's "Code Gauntlet Rewrite to Claude Workflows" has the benchmark data, the 4-day autonomous rewrite, and the results. hubs.li/Q04qJV6s0