Interview, Fireside Chat
Keeping AI Honest: The AI Wave Seen From the Inside | Olivier Pomel, Datadog | RAISE Summit 2026
- Datadog anticipates re-accelerating growth over the past six quarters driven by increased AI adoption scaling and an inflection point in traditional enterprise expansion observed in the last four quarters.
- The company forecasts that difficulties in understanding AI outcomes across businesses will persist for one to two years, while customers are expected to utilize a mix of AI models from providers like OpenAI, Anthropic, and open-source sources rather than committing to a single vendor.
- Datadog plans to evolve its platform strategy from observation into automation and end-to-end problem solving, aiming to build specialized internal models that are 100 times faster and 10 times cheaper by leveraging acquired technology.
- Future development targets include end-to-end world models that ingest telemetry data to predict future issues and recommend actions, with the long-term goal of establishing self-healing software where the system acts as both judge and actor.
- Achieving full autonomy in self-healing software is expected to take significant time due to software environment complexity outpacing road changes in self-driving cars, with customer trust requiring AI-independent fix accuracy well above 90% given lower tolerance for machine errors compared to human errors.
- The company intends to support customer environments by observing, managing, securing, and verifying the correctness of diverse mixed AI models to help customers understand spending, reach outcomes, and improve AI efficiency.
- Strategic focus remains on building the right products, ensuring sales alignment, and acquiring customers to eventually achieve profitability, with a leadership culture driven by a fear of error to foster continuous learning and market adaptation.
- Management expects stock market volatility not to distract from these long-term strategic fundamentals and intends to maintain a disciplined approach to material numbers, ranges, and timeframes despite the dynamic AI landscape.