Resolve AI, an AI for running and operating software in production, announced it has raised $40 million in Series A extension at a $1.5 billion valuation, led by DST Global and Salesforce Ventures. Just 18 months after coming out of stealth, Resolve AI has raised over $190 million and serves enterprise customers including Coinbase, DoorDash, MSCI, Salesforce, and Zscaler. The company also announced the launch of Resolve AI Labs, a strategic investment in building domain-specific models and agent systems needed to operate complex production environments.
“We are honored to partner with Spiros, Mayank, and the entire Resolve AI team to support their vision of bringing AI to production. Their exceptional talent density and decades of industry experience make this team best positioned to leverage AI to operate complex systems at scale,” said Rahul Mehta, co-founder and managing partner of DST Global. “What stood out to us about Resolve AI is its focus on the models, data, and system work required to make AI truly effective in production.”
Resolve AI Labs will be led by Dhruv Mahajan, who will join Resolve AI as Principal AI Scientist. Prior to joining Resolve AI, Dhruv was part of Meta, where he led post-training efforts for large-scale Llama foundational models. At Resolve AI, we apply that experience to building the domain-specific models, evaluation systems, and agent architectures needed to ensure AI works in production.
Also read: AiThority Interview with Glenn Jocher, Ultralytics Founder and CEO
“While the underlying models are rapidly improving, they are still not good enough for production operations,” said Spiros Zantos, Founder and CEO of Resolve AI. “Production environments require inference based on fragmented telemetry, long-running workflows, constantly changing systems, and extremely high standards for accuracy. Bridging that gap requires domain-specific models, post-training, and agent systems designed specifically for this domain, which is why we are creating the AI Lab.”
Generic models are not built for actual production operations. AI systems in this space must infer noisy telemetry, complex dependencies, and multi-step workflows where mistakes can have serious consequences, while also meeting stringent requirements for accuracy, latency, reliability, and control.
Resolve AI Labs focuses on building the models and agent foundation needed for AI to manage production systems. This includes:
- Domain-specific model building and post-training for production
- AI inference across operational telemetry including logs, metrics, traces, infrastructure events, and change history
- Evaluation framework for measuring reliability and accuracy of real-world operational workflows
- Synthetic data generation and simulation environment for scalable evaluation, training, and model improvement
- System architecture for scalable operational AI
- Governance and guardrails for AI in production
Zach Cocosa, principal at Salesforce Ventures, said: “However, managing software in complex production environments remains one of the most challenging problems in enterprise engineering. This requires deep domain expertise layered on top of Frontier AI, which is exactly what Resolve AI has pioneered. With a world-class team and proven traction among global enterprises, Resolve AI is uniquely positioned to lead the next phase of agentic AI operations. We are excited to partner with Spiros, Mayank, and the entire team.”
“Running software at enterprise scale means that incidents in production can be costly in engineering time, customer trust, and business continuity. Resolve AI has changed the way our teams address those incidents,” said Meir Amiel, Salesforce President and Chief Reliability and Infrastructure Officer. “What used to take hours of manual investigation and coordination between teams now takes a fraction of the time. Our engineers are not only faster, but focused on work that has real impact.”
Resolve AI Labs does this work in close collaboration with leading companies running some of the world’s most complex and business-critical production environments. These environments generate a wide range of signals, are constantly evolving, and require domain-specific inference and operational precision that off-the-shelf models cannot reliably provide. Resolve AI uses these real-world challenges to shape the models, post-training methods, and evaluation systems needed for AI to run effectively at scale in production.
This effort aims to enable production systems that can increasingly be managed by AI. In this system, models and agents work together to investigate incidents, diagnose root causes, and take action. Human involvement is determined by risk and operational status.
“Production systems are noisy, imperfect, and constantly changing,” said Dhruv Mahajan, chief AI scientist at Resolve AI. “Building AI that works in these environments requires advances in model building, inference, evaluation, and control systems. The opportunity is to take what is enabled by the underlying model and turn it into systems that are actually accurate, reliable, and operationally useful in production.”
This new funding will support Resolve AI’s continued investment in its platform, go-to-market expansion, and long-term research initiatives including AI Labs to build AI systems capable of taking on more of the work needed to run production environments.
Also read: The infrastructure war behind the AI boom
[To share your insights with us, please write to psen@itechseries.com]
