In December 2025, an AI startup less than two years old made TechCrunch headlines: Resolve AI raised a Series A at a $1 billion valuation. Just 14 months earlier, it had raised a $35 million seed round led by Greylock.
The more surprising number was ARR, which was reportedly only around $4 million at the time. That implies a valuation multiple above 200x ARR.
This is not only a bubble story. It is a product strategy case about redefining enterprise IT operations with AI at the right moment.
A Very Real Problem
Start with the workflow.
SRE teams at large internet companies receive hundreds or thousands of alerts every day. Each alert might mean users are experiencing slow access, orders cannot be submitted, or data synchronization has failed. SRE engineers have only minutes to decide whether it is a real issue or a false alarm, what the root cause is, and how to fix it.
How painful is this?
- Experienced SRE talent is extremely scarce, with compensation often reaching $300,000 to $500,000 per year.
- Every minute of MTTR, or mean time to repair, can translate into millions of dollars of loss difference for high-traffic businesses.
- Alert fatigue makes engineers numb, and severe incidents can be buried in noise.
For the past decade, the standard answer has been more tools: monitoring tools such as Datadog and Prometheus, alerting tools such as PagerDuty and Opsgenie, log tools such as Splunk, observability platforms, and more. The result is a longer toolchain, while SRE work turns into jumping between systems and assembling fragments.
Resolve AI gives a different answer: do not add another tool. Add an agent.
Redefining SRE
Resolve AI’s positioning is clear: AI for prod, meaning AI for production.
The name itself is interesting. It is not “AI monitoring tool” or “AI alerting platform.” It is “AI for production.” From name to positioning, the message is the same: AI is not only an assistant. It is a participant in production operations.
Resolve AI does three things:
First, AI joins on-call. The AI agent participates in each on-call rotation and automatically handles investigation and initial diagnosis for L1 alerts. Engineers do not need to be awakened by every alert.
Second, AI performs root-cause analysis. For complex incidents, the AI agent does not simply send notifications. It actively pulls data from monitoring, logs, APM, and related systems, correlates signals, and proposes likely root causes.
Third, AI executes remediation. For known issue types, the AI agent can execute fixes directly, such as rolling back configuration, restarting services, or scaling instances. DoorDash’s public case reports an 87% improvement in incident investigation speed and a 5x improvement in MTTR after adopting Resolve AI.
The product logic is this: do not add an AI assistant to SRE. Make the AI agent the execution subject of SRE work, with human engineers moving into supervision and complex decision-making.
This is not incremental improvement. It is job redesign.
Redesign the Product for AI, Not Add AI to the Old Tool
Many B2B AI products make the same mistake: they attach a ChatGPT-style conversation box to an existing product and call it AI plus something.
Resolve AI’s founding team, Spiros Xanthos and Mayank Agarwal, came from Splunk. They understand the operations tooling world deeply. Precisely because of that, they did not extend Splunk’s old boundaries. They rebuilt the workflow from scratch around the AI agent.
That means:
- The interaction is not a dashboard plus alert list. It is an agent-engineer collaboration interface.
- Alert handling is not passive receipt. AI actively participates in every step.
- Knowledge capture is not simply writing runbooks. The agent learns during execution.
This is the core productization difference: are you putting AI into the old process, or using AI to design a new process?
Resolve AI chose the second path.
Validation Signals: The Market Voted With Its Feet
Product design still contains subjective judgment. Market signals are harder to dismiss.
Financing: In October 2024, Resolve AI raised a $35 million seed round led by Greylock, with individual investors including Fei-Fei Li and Google DeepMind scientist Jeff Dean. In December 2025, Lightspeed led the Series A at a $1 billion valuation.
Competition: A strong competitor, Traversal, emerged in the same category and raised a $48 million Series A from Kleiner Perkins and Sequoia. Top-tier venture firms are clearly treating AI SRE as a category worth backing.
Customers: DoorDash was among the first public customers, and the 87% faster incident investigation metric is a verifiable ROI signal.
Exit validation: Moveworks, an AI company in the broader IT service management category, was acquired by ServiceNow for $2.85 billion, further validating the commercial value of AI plus IT operations.
Growth Flywheel: Why Latecomers Struggle to Catch Up
Resolve AI’s deepest moat is not simply technology. It is a data flywheel.
Every real incident handled by the AI agent teaches it more about the system’s behavior patterns: which alerts are noise, which metric changes are precursors, and which remediation steps actually work. The more incidents it handles, the more accurate it becomes. The more accurate it becomes, the more customers trust it. The more customers trust it, the more incidents they let it handle.
This is a classic “better with use” effect.
For latecomers, the biggest barrier is not building a similar AI architecture. With today’s open-source tools, a demo can be built in three months. The barrier is having real incident data to train the agent.
That is not a problem solved by more GPUs. It is a structural advantage created by time.
Three Lessons for AI Product Founders
1. Choose a battlefield with budget, pain, and labor shortage.
SRE talent is scarce, operations cost is high, and enterprises are used to paying for reliability. Few domains satisfy all three conditions. Resolve AI entered a market where customers already know the pain and have money ready, but cannot hire enough people.
If your AI product has to educate the market on why AI agents matter, you may have chosen the wrong direction. The ideal state is that users already know the pain, and you only need to offer the remedy.
2. Do not build “AI plus old product.” Redefine the workflow with AI.
Resolve AI’s team could have built an extension inside the Splunk ecosystem. Instead, they rebuilt from zero. That choice came from understanding the native boundary of AI capability: AI is not here to optimize alert lists. It is here to take over parts of on-call work.
3. In early commercialization, brand itself is a moat.
The name “Resolve AI” communicates three ideas at once: AI capability, problem-solving, and a clear production operations positioning. Strong naming is the first layer of sales material.
AI founders often obsess over technology. Resolve AI shows that from product definition to brand naming, every layer must answer: who are you, why do you exist, and why should users trust you?
As this article is written, the timestamp is June 2026.
Only six months have passed since Resolve AI completed its Series A at a $1 billion valuation. This market’s story is just beginning.
Which job will AI redefine next?