A year ago, I shared my thoughts at re:Invent, explaining why I joined Sumo Logic as CEO and laid out the importance of logs as a key differentiator. A year later, the atomic level of logs is even more paramount.
It’s not just because Sumo Logic is years ahead in technology when it comes to ingesting and analyzing structured and unstructured logs. We’re seeing a paradigm shift in the world of observability, often referred to as Observability 2.0, in large part because applications and infrastructure are more chaotic than ever.
In fact, as much as things have changed in recent years, I believe that with the growth of generative AI methodology in the coming two to three years, we will see a drastic realignment to log data as a foundation for insights.
The roots of dynamic observability
Organizations still need monitoring built on metrics and traces. If the world was static, we could theoretically instrument everything, getting the full context of your applications, user experience and the infrastructure supporting them.
But that’s simply not the nature of the world. Technology is constantly changing and evolving, and even if you somehow had perfect code, your third-party partners or integrations will update and evolve, driving even more change. There is simply no way to instrument full observability with traces or expand metrics usage for comprehensive visibility without an exponential budget.
Ultimately, when there’s an issue in an application or infrastructure, just rebooting isn’t an option anymore. Observability teams need to find and fix the root cause, being held to the same standards of investigation and forensics as security teams.
It’s not about the trace or the metric, it’s about the root cause, and that’s in the logs. Using logs as a system of record was the beginning, and is the DNA for Sumo Logic. But it’s what happens when you add in AI for insights where things get really exciting and pave the way for Dynamic Observability.
Bringing productized AI to market
This year proved that creating an AI widget was relatively easy, but turning that into a product is much more complicated. While others were focused on tinkering, we instead decided to focus on productizing these new approaches for our customers. That’s why I’m excited that we are revealing Sumo Logic Mo Copilot for general availability, announced at AWS re:Invent.
Copilot is unique in how it’s built and offered to customers. When ChatGPT had its moment this year, plenty of people asked it to write Sumo Logic queries. Sadly, those queries were inefficient or often didn’t even run.
Built on Amazon Bedrock to keep data private and secure, Mo was trained on over two thousand custom queries by Sumo Logic experts and can contextualize results with visualizations that would normally take even a power user some time to build. Faster dashboarding and insights help even non-technical users or junior front-line developers and security analysts get the information they need in real time.
We didn't want to restrict the ability to ask natural language questions of your data to only enterprise-level customers. That’s why Mo Copilot is available to all Sumo Logic customers without any extra charge. I said this could be unlocked when I talked about our Flex Licensing model earlier this year—you pay for the insights from the data, not for ingest, and get access to a wide range of new innovations and AI-powered insights.
The future of AI for Dynamic Observability and DevSecOps
AI innovation adds to the dynamism of technology; it’s much more than using LLMs and generative AI to ask your data questions in natural language.
When I’ve talked with customers and colleagues in this market, I often ask how they plan to solve the challenges we face thanks to the chaos and change in technology. The response I hear is DevSecOps. So of course I have to ask the next questions. What is that? Is it a team? A tool? A philosophy?
Ultimately, it comes down to collaboration. Development teams will still need their application and infrastructure logs, as will security and operations teams. But by agreeing on the base set of information in the logs, the various teams can parse and interpret it for greater understanding and shared workflows. In an incident, you need all of that information and teamwork to answer every question.
Humans will always be a vital part of this experience, but AI is taking it to new levels, and adding collaboration of its own. This isn’t about a singular AI that can do everything. Instead, it’s about various AI models that can criticize and collaborate, creating greater accuracy together. Composite or multi-agent AI is something I believe everyone will build in time because of the improved accuracy, but you can only get truly accurate results with the right data foundation, the structured and unstructured logs.
That’s why we’re embracing multi-agent generative AI now, particularly in our Dynamic Observability product development. Building new predictive insights and letting the AI judges criticize each other has brought us a surprisingly high level of accuracy, and each AI keeps getting smarter and more insightful.
Using these approaches, we’re building a real-time, instant snapshot of your dynamic application that drills directly into root cause even as your services evolve. We’re able to recreate the precision of traces with what we call AI-generated Traces directly from log data, no more instrumenting code.
Our Dynamic Observability solution does the work of query and investigation for you, too. It returns easy-to-understand natural language log summaries and goes even further to AI-generated recommended solutions. From there, it plugs directly into our automated playbooks. The combination reduces response times and downtime exponentially. This is Dynamic Observability.
Come by our booth at re:Invent and check out the prototype for Dynamic Observability for yourself. We can’t wait to show you how we are able to generate service maps, summarize what is happening in your applications and services, and ultimately automatically find the root cause for your errors - all without instrumentation. Not at re:Invent this year? Please reach out to us so we can show it to you.
We’re excited about everything we’re going to show at re:Invent, and I want to hear from you. This space is rapidly evolving, and we will build something better with your thoughts and feedback, so be sure to reach out to us so we can build this future together.
Complete visibility for DevSecOps
Reduce downtime and move from reactive to proactive monitoring.
