Deductive AI

Attending #srecon was truly remarkable. As first-time conference sponsors, our booth buzzed with hands-on trials, unscripted demos, tech deep dives, and engaging conversations. Being on call for several mission-critical systems, literally carrying pagers, has given me a profound understanding of the weight of responsibility and stress involved in ensuring 24x7 site reliability and availability. This is the first time in the history of computing, its possible to make 1000x difference in the life of an oncall engineer. Being part of making that happen have been the most gratifying part of my career. #SiteReliability #LifeOfOncall #DeductiveAI

What happens when engineers run a conference booth? You get real, unscripted demos -- no rehearsed fluff. We invited folks to “Bring your favorite bugs,” and they did. We dug into real-world issues on the spot—tracing services, debugging live systems, and finding root causes. Some highlights from Day 1 at #SRECon: 1. Deductive found a memory management bug in the Linux kernel’s transparent huge page subsystem. 2. We built a Cisco ThousandEyes connector on the spot to query and analyze their #alert and #tests endpoints. 3. Discovered service APIs and dependencies in a live production service within seconds. 4. And my favorite... wrote an incident postmortem as an ode to the Linux kernel’s memory management subsystem. This was our first-ever conference booth, and it turned out to be a fantastic way to connect with people who care about building things that actually work. Day 1 was incredible, and we're really grateful to everyone who stopped by, asked tough questions, and shared your stories! If you’re at the conference, come say hello at #Booth15 and see Deductive AI in action -- bring your bugs, and let’s debug together!

Day 1 at #srecon was truly remarkable! Our primary aim entering the conference was to directly confront any doubts surrounding AI SRE tools. The hands-on trial of our product on the demo floor was a big hit. Witnessing users engage with Deductive in real time and present diverse challenges was a pure joy. Here are a few examples from the nearly 100 questions posed: - Developing a tool to access thousand eyes - Identifying a bug in the memory management subsystem of the Linux kernel - Analyzing anomalies across various Datadog charts Excited for Day 2! If you haven't already, make sure to stop by booth #15 to explore Deductive AI. #AI #SRE #deductive

The?Deductive AI?team will be at?#SRECon?this week! ? We're building an agentic?#AGI?for SREs that can supercharge your productivity by helping root cause and mitigate large-scale software outages by reasoning about distributed systems, code, and statistical anomalies in real-time across unprecedented volumes of telemetry data (metrics, logs, traces, and events). ? Stop by our booth to try a live, hands-on demo, where you can take the driver's seat and experience the true power of our agentic code-aware observability platform.

It was a pleasure speaking at the infra.sf meetup last week about the future of AI-driven observability. What made the experience even more meaningful was the opportunity to share the stage with Charity Majors, an engineering leader I greatly respect for her deep insights and compelling articulation. At Deductive AI, we’ve been running AI-powered agents for root cause analysis (RCA) in production for over a year and have gained some good insights along the way. While I couldn’t cover all our learnings in a short talk, I shared three design patterns with the audience that I feel can be applied to most domain-specific agentic workflows: 1. Search as the Core Primitive – Many people may not think of it this way, but RCA is fundamentally a search and query problem across structured, semi-structured, and unstructured data -- often with extremely high cardinality and dimensionality. Effectively searching across this data requires building and maintaining rich search indices that track relationships across metrics, attribute keys, values, log patterns, and code metadata. Without this, even the most advanced AI systems struggle with signal correlation and causal inference at scale. 2. Using Approximations When Possible – Humans naturally excel at eyeballing/approximations; machines do not. This gap is especially evident in observability, where the exact precision of data often matters less than detecting the presence or absence of anomalies. At scale, exhaustively analyzing every log line, metric and trace is impractical. We found that structured sampling -- leveraging entropy, anomaly scores, and cluster similarity -- is a great tool for preserving statistical significance while filtering noise. 3. Code as a First-Class Telemetry Index – We found that leveraging code to search for telemetry data significantly improves RCA. Unlike natural language, code has strict syntax, execution rules, and lower entropy, making it easier for (fine-tuned) language models to learn and predict failures. We often find that starting our investigation by retrieving/understanding code and then using it to correlate logs, traces, and metrics, significantly improves RCA by aggressively pruning the search space. These are just a few of our learnings, but there’s obviously so much more to this. If you’re also building domain-specific agents, I’d love to hear what’s worked for you -- and what hasn’t!

As we wrap up an incredible year at Deductive AI, I want to take a moment to wish everyone a joyful and relaxing holiday season! We’re extremely grateful for an amazing year in which we doubled the size of our team, turned what was initially just a prototype (and a whiteboard idea) into a cutting-edge product defining the future of code-aware observability, and collaborated with some of the best and brightest minds in the industry and academia along the way. To our users -- thank you for putting your trust in us; your support and feedback fuels everything we do. To our investors -- your belief in our vision means the world. And to our growing team -- your passion and dedication inspire us every day! Here’s to an even more exciting and transformative 2025. Happy Holidays!

We're excited to welcome Ezra Stuetzel to Deductive AI as our newest Founding Software Engineer! Ezra brings over a decade of experience in software engineering, including key roles at two successful startups (RiskIQ and Appia-A part of Digital Turbine) and top-tier tech companies (Salesforce and Meta). His expertise spans distributed systems, scalable data solutions, and machine learning applications. Ezra has a proven track record of designing innovative systems, enhancing platform stability, and solving complex engineering challenges. At Deductive, we're building cutting-edge solutions at the intersection of AI, code and observability, and his deep experience and entrepreneurial spirit make him a perfect addition to our team. Welcome, Ezra!

We're extremely excited to welcome Arash Parsa to Deductive AI as our newest Founding Software Engineer! Arash joins us from Google Infra Spanner, where he was a tech lead on critical Spanner serving infrastructure projects and debugging rotations. Before Spanner, Arash spearheaded and led the development of the Cloud Firestore Key Visualizer from inception to launch for some of the largest GCP customers, in addition to other key contributions to GCP’s hotspot detection technologies. Today, the Key Visualizer for Firestore is an invaluable observability feature that is heavily relied upon by GCP customers for debugging and maintenance of their services! Before Google, Arash was a tech lead manager at Radius Intelligence working in the predictive data and ML space. He spent his undergraduate and graduate years at the University of California, Berkeley. At Deductive AI, we are solving some very hard/interesting problems in the area of code-aware observability, with a focus on building multi-agentic flows that can reason on extremely large amounts of data and code. If you are exceptional at what you do and are passionate about problems at the intersection of data and AI, please drop us a note!

Excited to share we are on a lookout for rockstar interns at Deductive AI. The role is based out of the London office in Camden. You will be working directly with me, while collaborating with the rest of the founding team in Mountain View. Apply here if you would like to work on Foundational AI, Agentic Systems and advanced data processing. https://lnkd.in/esHSd-xs

Happy Halloween and a Happy Diwali from team Deductive AI! Bonus points if you can spot our old and new logos. Aditya B., Karan Abrol, Krushnakant Bhattad, Madeline Cripps, Mustafa Mirza, Norm Zhou, Pratyush Verma, Rakesh Kothari, Shauna Blake, Yating Zhou