AI is transforming how software gets written. Claude Code, Cursor, and a new generation of AI coding tools have fundamentally changed development. A growing share of code is now being written by AI, and the volume of software shipped is accelerating fast.
Yet the true bottleneck to accelerating innovation remains unchanged. Before AI-assisted coding, engineers already spent roughly 70% of their time keeping production systems running – not building new features. As AI generates more code faster, this problem isn’t going away. It’s getting dramatically bigger.
We’re thrilled to partner with Resolve AI, the company helping engineering teams put the management of production code on autopilot.
The Problem: Production Is the New Bottleneck
Modern software infrastructure is extraordinarily complex. A single incident can require a team of engineers to navigate dozens of dashboards across Datadog, Splunk, Grafana, Kubernetes, and Slack, manually piecing together what went wrong across fragmented systems.
The process relies heavily on tribal knowledge – expertise that may live in the heads of your best engineers. When they’re not on call, that expertise is unavailable. When they leave, that expertise often leaves with them.
Meanwhile, the AI coding revolution is compounding the challenge. More code means more services, more dependencies, and more potential failure points. The observability vendors focused on collecting and visualizing data. They weren’t designed to reason through incidents the way an expert engineer does.
Resolve AI: AI That Investigates Like An Experienced Engineer
Resolve AI’s multi-agent system operates across code, infrastructure, and telemetry to triage alerts, investigate incidents, and help with production debugging. Rather than just correlating signals or summarizing logs, Resolve AI conducts structured investigations that are designed to mirror how expert production engineers think, turning fragmented data into an evidence-backed explanation of what happened and what to do next.
When an incident begins, Resolve’s planner agent orchestrates specialized sub agents – each trained on different tools and skills – to systematically triage issues, form hypotheses, and then prove or disprove each hypothesis by gathering evidence across integrated systems. The system constructs a knowledge graph mapping how services, pods, and components interact, capturing the tribal knowledge that previously lived only in senior engineers’ heads.
The results are compelling. Customers like DoorDash, MongoDB, Zscaler, MSCI, and others report massive reductions in mean-time-to-resolution (MTTR). Coinbase measured 73% faster time to root cause across tested incidents. Resolve AI delivers a structured explanation with citations – showing the reasoning process, relevant queries, and even the specific PRs that introduced problems.
A Platform Built for Expansion
Resolve AI’s vision extends far beyond incident response. The production context that its system learns, and the ability to reason and act across code, infrastructure, and telemetry, creates a foundation for managing a much broader set of production workflows. Teams are already using Resolve AI for alert triage, incident investigation and remediation planning, and production debugging. From there, the platform naturally expands into cost optimization, surfacing right-sizing opportunities, and into coding with production context, providing architectural guidance grounded in how systems behave in production.
Today, Resolve AI is owning production – delivering real impact for enterprises and commanding high six- and seven-figure ACVs. We believe the opportunity ahead is massive: the company that makes production work scalable, keeping engineers in control while elevating them to a higher level of abstraction, will define the next era of infrastructure software.
An Exceptional Team Taking a Big Swing
Resolve AI has assembled one of the strongest teams we’ve encountered. Founder and CEO Spiros Xanthos previously ran the 400+ person observability business at Splunk as SVP and General Manager – giving him rare depth in both the technical domain and go-to-market execution. Founder and CTO Mayank Agarwal brings deep expertise in the observability space and is widely regarded as one of the strongest technical leaders in the industry. He and Spiros have been building together for over a decade – meeting in their PhD program and founding multiple companies, including Omnition, which was sold to Splunk.
The caliber of talent extends across the organization: 14 former DeepMind engineers, along with leaders from Meta AI, Google, Tesla, SpaceX, and other top companies. When we visited their office on a Sunday, 40 people were working together preparing for AWS re:Invent. The intensity is matched by genuine camaraderie and excitement for the mission.
Our Partnership
At Lightspeed, we believe the biggest opportunities in AI infrastructure will go to companies that solve real production challenges. Resolve AI is building the intelligent layer that modern engineering teams need as AI transforms how software is built and operated.
We’re honored to partner with Spiros, Mayank, and the entire Resolve AI team as they work to make production engineering autonomous – and help every engineering organization operate with the expertise of their very best engineers.
Authors
