Resolve AI secures $125m Series A to revolutionise AI-driven software production management

Resolve AI has raised $125 million in a Series A led by Lightspeed Venture Partners, aiming to accelerate automation and operational efficiency in live production environments through AI-powered multi-agent systems.

Resolve AI said it has closed a $125m Series A at a $1bn post‑money valuation in a round led by Lightspeed Venture Partners, bringing the startup’s disclosed funding to more than $150m just 16 months after it emerged from stealth. The company, which desc...

ribes itself as building “AI for prod” to operate software in live production, said the capital will be used to speed product development, hire engineering and go‑to‑market teams, and support enterprise adoption.

The founders, Spiros Xanthos and Mayank Agarwal, are former observability executives who helped create OpenTelemetry and previously led parts of Splunk’s observability business. “The next frontier for software engineering is applying AI to the problem of running software in production,” Xanthos said in the announcement, arguing that production operations present a distinct challenge from development and require models that learn the unique behaviour of each customer’s systems.

Resolve AI positions its product as a multi‑agent system that lives inside customers’ production environments to triage alerts, investigate incidents autonomously, and surface latent reliability problems across code, infrastructure and telemetry. The firm said this approach is aimed at reducing operational toil, shortening mean time to resolution and freeing engineers to build new features rather than firefight outages.

Independent coverage confirms the headline figures but offers additional detail and some divergent accounts. Reporting by technology outlets notes Lightspeed’s lead and existing backers’ participation, and describes the product as an automation layer for site reliability engineering tasked with addressing rising system complexity and a scarcity of SRE talent. One report suggested the Series A may have been structured with multiple tranches at different prices, a detail not included in the company’s statement.

Market summaries and investor materials highlight rapid customer traction: the company and external reports list enterprise customers including Coinbase, DoorDash, MongoDB, MSCI, Salesforce and Zscaler, while another industry data provider indicated Resolve AI had more than 20 customers after emerging from stealth. Financial database entries show discrepancies with the company’s announcement on timing and totals: one commercial data service records a Series A closed in December 2025 and lists a smaller cumulative funding figure and an annual recurring revenue figure notably lower than implied by rapid enterprise adoption. These differences suggest variations in how rounds, tranches and disclosed totals are being reported across sources.

Lightspeed described Resolve AI as building a full‑stack AI solution, combining foundation models and custom agents tuned to production contexts. Investors and the company argue that general‑purpose models cannot safely operate live systems because operational knowledge is proprietary, constantly shifting and embedded in tribal workflows; Resolve AI says its models learn organisations’ specific stacks and run inside those environments.

The product approach raises questions industry observers have flagged elsewhere about risk, control and governance when AI is authorised to act inside production systems. Resolve AI emphasises human oversight, stating its system keeps engineers “in control of decisions and execution,” but it remains to be seen how customers balance autonomy with safety, compliance and incident‑response accountability as such tools are adopted more widely.

Resolve AI’s founders and backers frame the investment as confirmation that applying AI to run software in production is an emerging enterprise software category. As the company scales, observers will watch customer deployments, independent performance evidence, and the degree to which competing vendors or large cloud providers seek to replicate or counter the multi‑agent, production‑embedded approach.

Source: Noah Wire Services