Wild Moose

You’ve been dragged out of bed at 4 a.m. by a production alert, and you're not even sure where to start. Every engineer has been there. AI is going to change this for good.? It’s going to replace engineers as first responders. For years we had to suffer this chaos of critical incidents. This isn’t just frustrating—it’s risky. Every extra minute of downtime costs money and risks impacting major accounts. But - there’s a new standard thanks to LLMs When an alert is triggered, an AI agent can: ?? Analyze logs ?? Identify metric anomalies? ?? Review recent code changes? And - ? Give you actionable insights ? All before an engineer even looks at the alert. That means - first of all, you don’t need to waste time on repetitive checks on alerts that often turn out to be nothing. But that’s not all. With AI that *truly* understands your system, you can convert all that tribal knowledge that burdens your SMEs and blocks engineers from solving issues by themselves—into powerful automations. Soon - every company will rely on AI as the first responder to production issues. That’s exactly what we do at wild moose - we can integrate into your system in minutes. The result is less downtime, happier engineers, and more time for innovation. #AI #AIOps #Reliability #SRE #UptimeMatters

Your team’s tribal knowledge is buried in Slack. How can AI surface the answers you need—right when you need them? In critical incidents, context is everything. Engineers need immediate access to the right information to resolve issues fast, without dragging half the team into the fire. But here’s the reality we know all too well: You’re on-call, unsure who’s familiar with the service. The documentation? Incomplete and outdated.? You end up endlessly scrolling Slack, hunting for clues. This isn’t just frustrating—it’s risky. Every extra minute of downtime risks impacting major accounts, with serious business consequences. We realized there’s a better way. The shift to remote work has unintentionally done us a favor: troubleshooting conversations are now documented in Slack. Enter LLMs— technology that thrives on large, unstructured, human-centric data like this. Wild Moose uses advanced AI to make Slack work for you by turning it into a living knowledge base: ?? Surface past conversations to see who solved similar issues and how. ?? Generate and update incident response playbooks, automatically. ??Transform Slack into a dynamic, searchable resource for your entire team. The result? Less downtime, happier engineers, and more time for innovation.

Yes!! a first time, ai business event with 10 innovative Israeli startups at the AWS re:Invent LV. #coreai #aitools #devtools #dataengineering #cloudfinops #aitechnology #llm #genai AI21 Labs @bria.ai Pelanor PointFive Journeyz Kubiya.ai BigID SQream Wild Moose GigaSpaces Technologies Obviously, you do not want to miss it ?? for an invitation please contact Jack Hickey Israel Export Institute ???? ????? Eva Koritni Omer Fein

Finding a co-founder for a startup is one of toughest things ever. I met mine through Zoom. TL;DR I believe stress-testing the relationship as fast as possible is crucial. What was your experience meeting your co-founders? — When we first started exploring the idea of working together, we were all living in different countries. I was in California completing my MBA, Roei was in Canada doing his postdoc and commuting on weekends to New York to be with his wife, and Tom had just moved to Israel with his new fiancée. To start getting to know each other, we went through a list of 50 questions for co-founders over Zoom calls. It was clear this could be a strong partnership, but still, this is a huge decision: you are deciding who you’re potentially spending the next 10 years of your life with. You are tying your professional destiny with those people. Arguably, it’s the most important decision you’ll make as a founder, far more important than any startup idea that can change a hundred times. To make a decision, we flew for weekend offsites together. We rented Airbnbs in Santa Cruz, then Austin, then New York... Each time, we spent days and nights? brainstorming and, more importantly, getting to know each other. When I was about to graduate, I wanted us to get to a decision - are we going for this? But Roei had to leave his postdoc for the startup, which is a decision with no turning back. We decided to apply the “fail fast” principle to this choice. We moved in together for three months. It was a high-stakes experiment, it could fail miserably but we’d know quickly. We rented an Airbnb in the middle of nowhere with a goal: to secure initial funding by the end of our stay. If we succeeded, we’d continue. If not, we’d walk away. In the first week, a small disagreement escalated to a huge argument, just because we hadn’t yet learned how to have difficult conversations. We could decide to say goodbye, but we chose to work through it. And I think that was the real test. We proved ourselves again and again that we are willing to do A LOT to make this work. Because, as every founder knows, things are going to become tough. If you’re growing the company and working together for years - you can be sure you’ll experience new levels of professional and personal challenges, and you want to make sure as early as possible that your team can face those, support each other, and grow from them. Moving in together was kind of crazy, and definitely isn't the right idea for every team, but I do think that stress-testing the relationship as fast as possible is really important. Of course, while working continuously to build strong communication and trust.

Production moves fast—here’s how we built an on-call copilot that can keep up. The problem was obvious: Autonomously solving production issues is an incredibly complex challenge for an AI agent. Why? First, playbooks are outdated. We received this response from every company we spoke with. And it makes sense: production is always shifting, and companies want to move fast. So trying to keep every potential issue and response up to date is practically impossible. When we realized we can’t rely on playbooks, we decided to explore new approaches to make our AI agent fully autonomous. But letting AI make every decision in real-time didn’t work. The debugging search space is massive, with endless checks to run, making it too slow and computationally expensive. So, what’s the solution? It's all about finding a balance between the two extremes. Here’s what we’ve learned: ? Slack Is Gold With so much troubleshooting happening in Slack (especially since WFH became the norm), we can capture this information without adding extra work for teams. Slack conversations show us exactly what steps were taken—and when. ? Coverage Can Grow With Use If a new issue arises and someone runs a query we haven’t seen before—great! We’ll add it to our automated responses for the next time a similar issue shows up. ? Every Company Has Repetitive, Time-Consuming Checks While these checks may differ between organizations, they exist everywhere. Automating them saves critical time—right where teams need it most. The results speak for themselves. ??♀? With this approach, we’re cutting investigation time by over 40% right out of the gate. In the photo, my team demonstrates the best view as we work to nail the perfect balance between two extremes.

Yesterday I met Yasmin Dunsky of Wild Moose. They are building autonomous investigation and debugging agents to help SREs triage and resolve incidents. They made great progress since we last met. Quite a few companies are using them to reduce the load and time to solve production issues. If you are an SRE or manager - Reach out to Yasmin and check Wild Moose out!

We had a big breakthrough today: in just 3.5 minutes, Wild Moose identified the root cause of an issue that took a team of engineers half a day to diagnose. The speed and precision were way beyond our expectations. Being a founder is a constant grind, but a moment like this of true innovation is an energy boost that reminds us why we keep pushing forward. This feels big. And we’re only warming up. ????

Is "looking for new AI tools" just a Solution in Search of a Problem? We’re seeing frustration from employees when management pushes to “add AI”. It is tempting to criticize management here, but their job is not always easy: failing to modernize can stifle progress, while chasing after every shiny new product can waste valuable time and money. So, how do you strike the right balance? It’s about separating real value from hype, which is often easier said than done. The key is to focus on three crucial elements: 1. A new solution 2. To an old, stubborn problem 3. With a compelling “why now” When these three elements align, it often signals a true disruption—a known problem finally being tackled in a way that was previously impossible. My advice for managers: When evaluating new tech or speaking with entrepreneurs, challenge the “why now.” Look beyond superficial enhancements to understand if the technology genuinely addresses a core aspect of your problem in a novel and impactful way.

We've all been there - frantically fixing a bug that somehow slipped through the cracks and made it into production. Was it a rushed deadline? A complex feature? Or maybe something else entirely? ?Vote and share your own war stories in the comments.?

Automation of incident response is where the industry is heading. We're seeing response times slashed from hours to minutes as AI triages issues, analyzes impact, and suggests next steps - all before human intervention is needed. And this is just the beginning. As models continue to improve, managing incidents without these tools will soon become unthinkable. Yet, we often hear misconceptions about companies' readiness to adopt these solutions: ?? “Our logs are messy” - This isn’t a problem. In fact, it's where Generative AI excels. Large Language Models thrive on unstructured data. ?? “We use legacy observability tools and are mid-migration to a new one” - Great automation tools are observability-stack agnostic. They create an abstraction layer, integrating with any obscure tool in the background. ?? “Our runbooks and playbooks are outdated” - Yes, that is the case for 99.999% of companies we talk to. LLMs can leverage data such as Slack history and post-mortems to automatically maintain playbooks with minimal manual intervention. Hit us up if your organization is ready to take incident response to the next level.