• Portfolio

Autonomous Production With AI

Our investment in Resolve AI.

April 16, 2026

by Paul Drews, Zak Kokosa and Pascha Hao

• Founders: Spiros Xanthos (CEO) and Mayank Agarwal (CTO)
• Sector: Infrastructure
• Location: San Francisco, CA

The Opportunity

Running software in production has become one of the hardest operational challenges in enterprise engineering, and it’s only getting more complex as modern systems span thousands of services, multiple cloud environments, and layers of infrastructure that evolve constantly.

Observability, the set of tools companies rely on to monitor and understand these systems, has been one of the largest line items in this effort, consuming 20-30% of infrastructure software budgets and generating tens of billions of dollars in annual vendor revenue. Despite all that spend, the core problem remains: when something breaks at 2 am, an on-call engineer has to manually stitch together signals across an average of seven disconnected tools, scattered runbooks, and institutional knowledge that lives in Slack threads and engineers’ heads just to understand what happened, let alone fix it.

And that was the state of things before AI started writing code.

The result is the cognitive load of managing production systems has surpassed what humans can handle alone. Today’s monitoring tools can tell you something is wrong. They can’t tell you why, and they can’t fix it. And if every alert, incident, and production issue still depends on a human to investigate, decide, and act, the operational ceiling remains tied to human attention.

What’s needed is a fundamentally different approach: systems that don’t just surface problems, but can reason through them and help run production end to end. This is the problem Resolve AI was built to solve, and it’s why we’re excited to announce our investment in the company.

The Solution

Resolve AI is building AI systems for production that can investigate issues, reason across signals, and help teams navigate complex environments with greater speed and context. When something goes wrong, Resolve AI investigates the way experienced engineers do: by forming hypotheses, gathering evidence across systems, and narrowing to the likely root cause. Resolve AI can then surface clear remediation plans and support the next step within defined workflows.

But incident response is only one part of the broader shift Resolve AI is working towards. The larger opportunity is to reduce the amount of production work that depends on human effort across alerts, investigations, troubleshooting, and ongoing system maintenance and reliability.

Observability remains foundational, but visibility alone does not resolve complexity. The next layer is systems that can interpret that visibility, reason across it, and help teams act in context. Production is not a single query. It is an ongoing operational responsibility, and it requires systems built specifically for that problem.

The Team

Resolve AI was founded by Spiros Xanthos (CEO) and Mayank Agarwal (CTO), repeat founders who have been building together since 2012. They co-founded three companies with two exits to Splunk and VMware, and co-created OpenTelemetry, now an industry standard for how software systems share operational data. After the Splunk acquisition, Spiros ran Splunk’s observability business as SVP & GM while Mayank served as distinguished lead architect. Few teams know this space better.

Further, Spiros and Mayank have built a team with impressive domain expertise — researchers and engineers who have led pre-training and post-training efforts at some of the world’s top labs, including DeepMind, Meta’s Superintelligence team, and Google, where they worked on projects like Gemini, Llama, and Deep Research. On the go-to-market side, Resolve AI has recruited from Splunk, MongoDB, Grafana, Databricks, McKinsey, and HashiCorp. It’s one of the highest talent-density teams we’ve ever seen.

The impression of the Resolve AI team that stuck with us the most didn’t happen in their office. At Resolve AI’s holiday party last December, we connected with a contractor on their office management team. Unprompted, he told us how positive the culture is and how the team works incredibly hard with evident passion for what they’re building. His impression gave us even more confidence that Spiros and his team will do what it takes to make Resolve AI and their customers successful.

Our History with Resolve AI

Our team has spent the last year and a half going deep on the AI-for-production and observability space, conducting extensive customer diligence across enterprises spanning financial services, tech, cybersecurity, and more.

One of the advantages of Salesforce Ventures is the depth of our enterprise network. Before investing, we connected Resolve AI with Fortune 500 executives across our Innovation Advisory Board and customer network community. The interest in Resolve AI’s solution was immediate and strong. Leaders jumped at the chance to meet with the Resolve AI team, which only increased on conviction that demand for an AI system for production is real.

As part of our diligence process, Salesforce Ventures often connects promising portfolio companies with Salesforce leaders to stress-test their thesis and validate their technology. With Resolve AI, that introduction led to something more: Salesforce became a paying customer, and the ROI spoke for itself. After a rigorous eight-month evaluation against competing vendors, a team at Salesforce deployed Resolve AI to enhance production operations. Following hundreds of successful investigations, the platform delivered transformative results: a ~60% reduction in mean time to resolve (MTTR), ~70% faster alert triage, and a ~30% reduction in investigation time. In one instance, Resolve AI autonomously diagnosed a complex production issue in just 10 minutes — a task that typically necessitates hours of intensive, manual coordination across multiple engineering workstreams. Given Resolve AI’s rapid product velocity, we view these initial enterprise-scale results as a floor, not a ceiling, for what will be possible going forward.

What’s Ahead

Running production has long been one of the hardest problems in enterprise software, and AI-generated code is only making it harder. The burden now exceeds what humans can manage alone. The next phase of AI in enterprise software will not be defined solely by how software is built. It will be defined by how it is run.

We’re excited to partner with Spiros, Mayank, and the Resolve AI team as they define what AI for production looks like. Their combination of deep domain expertise, world-class AI talent, and a product that demonstrably works at enterprise scale is rare. The early traction with customers like Coinbase, DoorDash, Millennium, MSCI, Zscaler, and Salesforce speaks for itself.

This is a team that doesn’t just build great technology. They earn the trust of the most demanding engineering organizations in the world, and then deliver. We couldn’t be more excited to be on this journey with them.