Port.io

Workflows is in open beta! You can now orchestrate your entire agentic SDLC with AI agents, human-in-the-loop steps, and your full Port context built in. Think onboarding, incident response, deployment pipelines, and more, all running with the right people involved at the right time. Try it out and let us know what you build!

Q2 26 is Port.io's all-time best quarter: - Net new ARR : ~30% higher compared to our previous record - We closed our biggest deal so far, almost 8 figures in TCV.  - 70%+ of renewals closed with an ARR upsell - We crossed the 70% ARR in multi-year deals (out of total ARR) These are just a few indications that Agentic SDLC is about to be the biggest AI category in engineering.

Port is proud to be adopted by the team at Via Transportation!

Join us next week where we will share the results of our research on cutting token usage. Well, the results you already know: context lake uses 80% less tokens per query. But if you're curious how you can get the same results, we'll be showing you exactly how it's done.

Cole Medin just broke down Google's new paper on the new agentic SDLC and it's a must watch. We recommend watching the full thing, but to give you a taste, here are some of the main points: - Agent = Model + Harness. The model is ~10% of output quality. The harness is the other 90%. - It's possible to move from outside the Top 30 to Top 5 on Terminal Bench 2.0 just by changing the harness. (That's like getting Opus results for Sonnet prices) - Most agent failures aren't model failures, they’re harness failures. Which means you can harness engineer your way out of them. - At scale, vibe coding can cost 3–10x more per feature than agentic engineering. Lesson: Invest in agentic engineering early and save down the line. Thank you to Cole, and of course to the writers Addy Osmani, Shubham Saboo, and Dr. Sokratis Kartakis for pushing agentic engineering forward https://lnkd.in/ewtMEZkD

Our next Port Community Session is coming up on Tuesday, July 1st led by Hila and Etay from the Port product team. Here's what's in store: A walkthrough of recent releases, including: - workflows which is now in open beta - AI improvements A discussion about skill registry: what it is, how to publish your first skill, and how teams are using it to ground their agents in how their org actually builds. We'll also hear from Romi Hemo, a DevOps Engineer at Honeybook, who will share how their team leverages Port to in their shift from manual to autonomous engineering. And we'll have a look at what's coming up on the roadmap, and which community priorities are shaping the next releases. We hope you join us! Register here: https://lnkd.in/evhmtbxB

Messy context is expensive. But how expensive exactly? How many tokens are agents burning through hopping between data sources and reasoning about them? And can it be solved with structured context? We ran thousands of AI queries to find out! Read the full results here: https://lnkd.in/eYQrYaCf

With AI costs rising, we asked ourselves: How can we save tokens when using AI? And we thought: is messy context the reason AI is so expensive? What if we structure the data according to the kinds of queries AI and humans make? Agents should then reason less and burn fewer tokens. To test our hypothesis, we ran 1000 agent queries across four context conditions and three models. We're excited to share that across all models, context lake used 80% less tokens than direct-to-MCPs. Queries without the context lake cost $1.76/query on average. With the context lake, just $0.35. That's an 80% reduction in token usage. Curious how we ran the experiment and why context lake is more efficient at fetching context? The full results are on our blog Link in comments

Port is proud to be adopted by the team at Jack in the box

Very excited for this one! 🚀 PlatformCon is one of the highlights of my year, and I can't wait to connect with the platform engineering community in both London and NYC. This week, you'll find the Port.io team at: 🔹 London (June 23) — Table #2 🔹 NYC (June 25) — Table #22 And if you're attending in London, make sure to catch a highly anticipated talk from our own Gus Shaw Stewart: 🎤 Stop Being The Human Router – Agentic Provisioning From Your Internal Developer Portal Gus Shaw Stewart will explore: 🤖 The foundations needed for Agentic Engineering 🔐 Why context, governance, and guardrails are critical for AI adoption 📈 How to build the Agentic SDLC Whether you're building an internal developer platform, exploring AI-powered engineering workflows, or figuring out how to safely scale AI across your organization, I'd love to meet you and hear what you're working on. Stop by, say hello, take on our AI Insights dashboard challenge, and let's talk about the future of platform engineering. See you at PlatformCon! 🙌 #PlatformCon #PlatformEngineering #AgenticEngineering #AIEngineering #DeveloperExperience #InternalDeveloperPlatform #DevOps #PlatformCon2026