Warp

We've stopped buying SaaS and and are moving everything to agents, Skills and just-in-time apps. In some cases, we've canceled subscriptions altogether and saved $10k+/year in the process. Here's our early success stories: Docs migration: We moved our docs site from a hosted solution to Markdown with Astro in a matter of hours, made them better, and saved $10k+/year. Our DevEx team: - Ported over all 266 pages, including video embeds, tab components, callouts, and image formatting - Built a custom Ask AI widget backed by Kapa that also improves the experience - Exposed all docs content as plain Markdown for agent and LLM accessibility - Set up hosting on @vercel, including proper preview deployments and better analytics This was all done by batching parallel cloud agents with Oz. Now, they're using computer use to validate that all pages migrated without regressions. Should be in prod soon. Community monitoring: Our DevEx team also replaced community insight SaaS subscriptions with an agent that monitors activity, suggests replies, and self-improves with agent memory. We use Octolens for a data firehose of Warp mentions from X, LinkedIn, Reddit, etc, and set up a scheduled agent using Oz to review mentions. This agent has a repo of Skills to: - Decide how to engage (skip, like, or reply) - Suggest replies - Review our own feedback to the agent via Slack Then, the agent can PR changes to itself based on our feedback. You don't need agent memory tools. Let agents PR Markdown changes. Self-serve data insights: Our data analytics team made customer data visualizations self-serve with agent Skills + BigQuery CLI access. Instead of pinging the `data` channel to ask questions, we open the `dbt` repo and ask agents what we want to know. The agent tells us about DAU trends, depth of feature usage, platform adoption, etc. Replaced the Metabase UI with agents for a lot of our team. Just-in-time recruiting apps: Our recruiting team tracked everything in the Greenhouse UI and managed offer letters from a google sheet. Now, they've connected to the raw Greenhouse API to build a custom UI to manage candidates. They're also prototyping a generated offer letter to break down compensation. Grafana alert monitoring: There's much more we're doing that's SaaS adjacent too. Our platform team built an agent to catch Grafana alerts, investigate, and suggest action items using the Oz cloud agent with codebase access. Collapses Slack ping → terminal → fix to simply Slack ping → fix

In this week’s Friday Deploy (yes, Dev Interrupted has a dedicated Friday news segment -- get into ittttt!), Ben Lloyd Pearson and I are joined by Warp founder/CEO Zach Lloyd (a past guest and friend of the pod) to talk about the infrastructure gap blocking most teams from capturing the value from agents -- and why he launched Oz to fix it (btw don't miss the great launch demo from Ben Holmes). But first we cover the news...! And yeah, it's been a week: 🔥🧯 On Monday, GitHub had service issues impacting Actions, PRs, notifications, and Copilot, including notification delays on the order of ~50 minutes before things recovered later that day. (Did anyone else's agents get cranky about this?!) -- its a signal that many are moving from single-agent productivity to multi-agent throughput. 🧛♂️ 💸 We also dig into Steve Yegge “AI Vampire” framing: the productivity boost is real BUT the fatigue curve is real-er. What are we going to do about all this new economic pressure to convert 10x tooling into “10x output”... and who really captures the value of that output? And we shout out Ai2 + the work of Tim Dettmers that explores a methodology for training repo-specialized coding agents on brownfield codebases, using a small number of insights to leverage massive gains with minimal fine-tuning. So tune in now to catch Zach's deep dive into Oz, Warp’s new orchestration platform for running agents in the cloud with sharing, scheduling, artifacts, and auditability. Because folks, we're not in Kansas anymore. 🌪️ 🎧 Breaking GitHub, AI vampires & the great Oz: https://lnkd.in/gdREvbXa

In 2026, there are two sets of companies - those who leverage pricing to gain market-share and those who lose market-share. Changing prices used to be verboten. In a seats-based world margins were fat, headcount was growing, and everything cost $9.99. In the modern market, the winners are using pricing as a mechanism to grow faster. Companies like Lovable, Intercom, and Browserbase are taking advantage of the AI-wrought sea change to accelerate growth to historically high levels. If you want to be inspired by ambitious price change pioneers - read our latest whitepaper: Real world pricing pivots from Intercom, Lovable and many other friends. https://lnkd.in/gC8V83Qd

Warp can now create Figma designs and manage design systems end-to-end with the latest Figma MCP server. With the Warp Figma Skill, agents can: ✅ access design tokens ✅ create components ✅ design system in sync with code Enable the MCP server from your settings and apply the Skill

Warp now supports "code to canvas" with the Figma MCP server. Render UI, push it to the Figma canvas, get design feedback, pull it back into code. Tighter design-dev loop without context switching. Here's how it works:

Warp is now a member of the Agentic AI Foundation. Together, we're pushing the frontier of agentic development forward, supporting open standards for AGENTS[md], MCP, and more. The future of agentic development is multi-model, compatible with the full ecosystem of Skills and workflows. Thanks to The Linux Foundation for bringing the agentic coding community together to make open standards a reality.

We’re hosting a webinar on February 24 to talk about Oz helping teams move from individual AI usage to coordinated, agent-driven workflows across the whole organization. If you’re an engineering leader thinking about how AI agents actually get deployed inside real teams, this will be practical and concrete.

As AI agents move into production, visibility and governance matter. Read Fast Company's article to explore how we’re building secure, cloud-based agent workflows with Oz: https://lnkd.in/dx7pVT9U

We’ll be live on OpenAI Build Hours today at 10am PT to talk about prompt caching. The session is a technical deep dive into how we’re leveraging prompt caching in Warp and the results we’re seeing so far. If you’re building production agents or are curious how Warp is built, join the session live: https://lnkd.in/ezgK5DMb