Rootly

Engineers hate talking to sales. Honestly, so do I. So why does Rootly ask you to "book a demo" still? Even after getting roasted plenty of times on Hacker News, I insist you talk to us before you buy. Because our sales team is expected to act like advisors, not sellers running a script. Before anyone speaks with a customer, they must pass an intensive demo certification that I personally administer. Every seller has to meet my bar. They are tested on more than product knowledge. They need to understand incident response deeply enough to challenge your process, identify gaps, and bring strong opinionation to the table — so you’re not just hearing what you want, but what you need. Their advice is grounded in what we have learned from millions of production incidents — not generic sales playbooks. We have a team whose full-time job is just internal enablement. Our sellers know the space well enough to speak at SREcon and have built relationships so strong they get invited to customer weddings (happened today, again). There are real tradeoffs: — Some people do not pass. — We will regularly tell prospects not to buy when Rootly is not the right fit. — It takes an unreasonable amount of my time (but worth it). But the upside is just as real: — No wasted calls. Solve the problem and let the product do the selling. — Even when you do not buy, you leave knowing more about incident management. — Our sellers have never missed quota and our win rates reflect that. It started with sales but has expanded to everyone in the company. From legal to design engineering, we hold everyone to the same standards. Just like how incidents are everyones job, so is customer empathy.

Wise built money without borders. Rootly keeps it running without downtime.

What a great night hosting the Replit Community Meetup last week! Thanks to Alexandre Cuoci (Replit), Om Patel (Amazon), and everyone who came out to learn, build, and connect. It's always inspiring to see so many people excited about building with AI. Until the next one!

AI agents don’t just need more context. If you give them a graph, their efficiency jumps by at least 30%. Anthony Alcaraz, GTM Agentic Engineering Lead at AWS and co-author of O'Reilly's Agentic GraphRAG, makes a useful distinction: a large context window gives a model more information, while a graph gives it structure it can navigate. That changes how an agent works. It can move through connected information, observe what happened, check whether it succeeded, and decide where to go next. The graph can also encode relationships and rules, adding a reasoning layer around the LLM. It raises a bigger question: if enterprise knowledge remains fragmented, how much can smarter models really fix? In conversation with Sylvain Kalache, Anthony explores: • Where vector RAG starts to hit its limits • How graphs support retrieval, memory, planning, and enforcement • Why better context could make smaller models more effective • How to preserve the tribal knowledge hidden across an organization • What happens when small teams begin managing fleets of agents Full conversation: https://lnkd.in/e_aWkKPJ

A Slack channel and the word “help.” That was Canary Technologies incident management process when Z Lee joined to lead platform engineering. But the engineering 4x’ed in 18 months and ad-hoc processes wouldn’t work anymore. Canary needed more than a new tool. They needed to build an incident management practice from the ground up: clear severity levels, defined ownership, reliable escalation paths, automated customer communication, and a learning loop to stops the same incidents from coming back. They built it on Rootly. Today, every alert is captured, responders can page the right team directly from Slack, incident timelines feed into RCAs, and more than 20,000 hotel customers around 125 countries can hear about degraded service from Canary first. Best of all, the system scaled without workarounds. And they’re still growing, so check out Canary’s careers page! Read the full story: https://lnkd.in/exBcqg7N

Security teams cannot fight AI-speed attacks with human-speed workflows. As attackers use AI to scale campaigns and operate more autonomously, hiring more analysts to process more alerts will not solve the problem. Nir Soudry, Head of R&D at 7AI, argues that the answer is agentic cybersecurity: AI investigates and remediates threats, while humans define policy, retain accountability, and decide when to intervene. Full episode here: https://root.ly/nir-7ai

Slice handles everything so pizzerias can focus on making great pizza. Rootly handles reliability so Slice can focus on the pizzerias. 

Nothing bonds engineers more than AI… or a good World Cup match. Rootly hosted a happy hour with Tailscale, Aikido Security, Docker, Inc, and Inngest to help folks in San Francisco for the AI Engineer World Fair wind down. Thanks to the 400+ engineers who stopped by to share your passion for AI and soccer with us.