Larridin Raises $30M Series A for AI Measurement

One Larridin customer had 92% of its engineers actively using AI. One engineer still accounted for roughly 65% of the team’s AI spend during a single week.

That gap is the business Larridin wants to measure. The South San Francisco company announced a $30M Series A on November 2, 2026, giving its platform more capital to connect AI spending with the work produced by people and agents across an enterprise.

Pelion Ventures led the round. Existing investors Andreessen Horowitz and Gradient Ventures participated, alongside new investors GFT, W23, Thayer, DG Daiwa Ventures, Alumni Ventures and Cougar Capital. Larridin says the financing brings its total capital raised to $47M. The company did not disclose a valuation, ownership terms or new board rights.

Larridin Is Following the AI Invoice Into the Work

An enterprise can count licenses, tokens and model calls without knowing whether any of them changed an operating result. Larridin is building an intelligence layer around that missing connection.

The platform traces AI costs to the teams, tools and agents generating them. It then connects adoption and usage with engineering output, workflow activity and the capacity a company believes AI returned to the organization. The product is divided across spend and token intelligence, work intelligence, engineering intelligence and workflow intelligence.

That assignment reaches further than traditional software-spend management. A lower model bill is useful, but it does not show whether a team produced better work. A faster engineering cycle is useful, but it can create a new delay when the volume of code reaches human review. Larridin is trying to keep the expense, the output and the next bottleneck inside the same record.

The company says it has measured more than 2B AI sessions and followed more than 82,000 workflows end to end. Across its customer base, Larridin reports a 47% average increase in measurable team output and a median $2.4M in AI spend optimized during the first year. Those are company-reported figures, not audited industry benchmarks.

The Series A Backs a Measurement Team With History

Russ Fradin is Larridin’s co-founder and CEO. Jim Larrison is co-founder and president, and Ameya Kanitkar is co-founder and CTO. The company was founded in 2024.

Larridin’s founders frame the new assignment through work that came before it. Russ Fradin and Jim Larrison helped build digital measurement at Comscore before working together again at Dynamic Signal. Ameya Kanitkar brought enterprise and AI-infrastructure experience from LinkedIn, Coinbase and Groupon.

The history matters because measurement markets usually become valuable after a new technology has already begun spreading faster than the systems used to understand it. The web needed businesses that could translate traffic into commercial evidence. Enterprise AI now has a different version of the same problem: every department can adopt a tool, every model can produce activity and nobody automatically receives a clean answer about the value created.

Larridin says the Series A will accelerate product development and expand its engineering and customer teams. The company is also building a recurring public record through its quarterly Engineering AI Impact Report.

More AI Output Can Move the Bottleneck

Larridin’s Q3 2026 engineering data shows why adoption alone is an incomplete score.

The company reports that AI-assisted pull requests rose from 35% of merged code changes in April to 62% in September. Output per engineer increased 33% from Q2 to Q3, while code cycle time fell from 9.5 hours to 7.3 hours.

The review system did not move at the same speed. Median time to first human review increased 50%, from 1 hour to 1.5 hours. Time to approval rose from 1.8 hours to 2.5 hours.

The tool may have helped produce more code while placing more work in front of the people responsible for approving it. That is not a reason to stop using the tool. It is a reason to measure the entire production path instead of celebrating the step that became faster.

The same problem appears in cost data. Larridin says another customer saw a 66x difference in cost per shipped engineering change between 2 AI tools performing comparable work. A procurement team looking only at license price would miss the operating comparison. An engineering leader looking only at usage would miss it too.

Customers Still Decide Whether Measurement Changes the Budget

Larridin names SurveyMonkey, Vertiv, Klaviyo, Gainsight, Globality, TigerConnect, EcoVadis and Sundt among its customers. Their presence gives the company a base across software, industrial infrastructure, healthcare and sustainability. It does not make every workflow directly comparable.

That is the harder part of the market Larridin is entering. AI work does not arrive in one unit. A support interaction, a code review, a sales proposal and an automated finance process create different kinds of evidence. A measurement layer has to preserve enough context to make the comparison useful without reducing every job to a synthetic productivity score.

Pelion Ventures general partner Jeff Kearl described the opportunity as connecting AI spending to measurable business outcomes. Alex Rampell, general partner at Andreessen Horowitz, pointed to Larridin’s growth and customer relationships. The investors are funding a company that has found activity around the problem. The next proof is whether customers change budgets, workflows and automation priorities because of the answer.

That decision keeps the record moving. An AI invoice can show what a company bought. Larridin is raising capital to stay with the more difficult question: what the organization became capable of doing once the bill arrived.