Isotopes AI came out of stealth on Thursday with a $20 million seed round.
Provides AI agents to solve problems that data analytics products have struggled for decades. Those who know how to run big data infrastructure are not those who actually need to use data.
LLMS allows business managers to ask data questions in natural language. Aidnn, an Isotopes agent, can draft answers and complex planning documents, and collect data from stored locations, such as financial apps, ERP, CRM, and cloud storage.
There are countless agent business analytics offerings there, but the co-founders of Isotopes have a unique pedigree. Their products are very sophisticated, and the startup has already filed for 10 patents, co-founder CEO Arun Murthy told TechCrunch.
More than 20 years ago, when Mercy was in her mid-20s, he worked on a team on Yahoo that built an open source project called Hadoop. Hadoop spurred the first big data frenzy of the 2010s.
In 2011, Yahoo spun it into a company called Hortonworks, and Murthy served as co-founder and chief product officer. Just four years after the launch, Hortonworks was released. However, the rise of New Cloud Storage Tech hit Hadoop’s market, with Hortonworks eventually merged with its biggest rival, Cloudera. The merged company was made private in 2021 after the involvement of well-known activist investor Carl Icahn.
Mercy went to Claudella for several years amid the chaos, managing around 200 people. But he says even there he saw an old data access problem. He remembers a quarterly conference call where Wall Street analysts grill executives about details of their operations.
TechCrunch Events
San Francisco
|
October 27th-29th, 2025
We couldn’t always answer because we couldn’t access the data. “That was embarrassing,” he admits now. “We were a big data company selling this.”
In 2021, he left the job and had no real plans for what to do next, and the VC introduced him to expand on AI’s Alexandre Wang. After some chats with Wang and a little consulting work, Murthy joined Scale as Chief Technology Officer.
It was like “Getting a large PhD.” “Understanding what drives these models and how to improve them.”
However, when his old companion of Hortonworks, Prasanth Jayachandran, called out to him, the two decided to do their own AI startup. They convinced the third Kowalvijayaragavan, who founded the isotopes in the second half of 2024, as well as from the time of Horton Works. Their seed round was led by NTTVC’s Vab Goel. (Goel was previously at Norwest Ventures.)
The founder’s background allowed us to build an agent that could find data from a saved location (such as Salesforce or Snowflake), but we also cleaned the data. Agents also maintain a lot of contextual memory that is useful for complex tasks.
“This goes far beyond simple chatbots,” Mercy said. For example, if your team asks AIDNN to draft a report on monthly recurring revenue trends, “The data you actually want to chat with is at least a form that you need to chat with. Multi-stage planning, very complex plans, metadata, extract and clean data, read data, combine data, and aggregate it.”
The agent also points to anomalies in its procedures, inferences, assumptions, and data. It even makes recommendations on how to proceed. Isotopes also promises that enterprise customers can deploy without sharing data to AI model makers that power their agents.
Still, while it may be as sophisticated as isotopes, startups face a lot of competition. Incumbents like Salesforce’s Tableau already offer agents (among Salesforce’s major AI agent push). Also founders of many other startups with impressive pedigrees are in markets such as Widdia.
