Isotopes AI came out of stealth on Thursday with a healthy $20 million seed round.
It offers an AI agent to solve a problem that data analytics products have struggled with for decades: The people who know how to run the big data infrastructure are not the ones who actually need to use the data.
With LLMs, business managers can ask questions of their data in natural language. Isotopes’ agent, Aidnn, can provide answers and draft complex planning documents, gathering data from wherever it’s stored, like finance apps, ERP, CRM, and cloud storage.
There are countless agentic business analytics offerings out there, but Isotopes’ co-founders have a unique pedigree. Their product is so sophisticated, the startup has already applied for 10 patents, co-founder CEO Arun Murthy told TechCrunch.
Just over 20 years ago, when Murthy was in his mid-20s, he worked at Yahoo on the team that built an open source project called Hadoop. Hadoop spurred the initial Big Data frenzy of the 2010s.
In 2011, Yahoo spun it out into a company called Hortonworks, with Murthy as co-founder and chief product officer. Just four years after launch, Hortonworks went public. But the rise of new cloud storage tech took its toll on Hadoop’s market, and Hortonworks eventually merged with its biggest rival, Cloudera. The merged company was taken private in 2021 after famed activist investor Carl Icahn got involved.
Murthy went to Cloudera for a few years during that turmoil, managing about 200 people. Yet, he says, even there he saw the age-old data access problem. He remembers quarterly conference calls with Wall Street analysts grilling execs on operating details.
They couldn’t always answer because they didn’t have access to the data. “It was embarrassing,” he admits now. “We were a big data company selling this.”
In 2021, he left that job with no real plan on what to do next, and a VC introduced him to Scale AI’s Alexandr Wang. After a few chats with Wang and a bit of consulting work, Murthy joined Scale as chief technology officer.
It was like “getting a PhD at Scale,” he describes. “Understanding what drives these models and how to improve them.”
But when his old buddy from Hortonworks, Prasanth Jayachandran, called him, the two decided to do their own AI startup. They persuaded their third-cofounder, Gopal Vijayaraghavan, also from the Hortonworks days, to join them and in late 2024 founded Isotopes. Their seed round was led by Vab Goel at NTTVC. (Goel was previously at NorWest Ventures.)
The founders’ backgrounds has allowed them to build an agent that can find data from wherever it is stored (be it Salesforce or Snowflake), but then also clean the data. The agent also maintains plenty of context memory to be useful for complex tasks.
“This is far beyond a simple chatbot,” Murthy said. For example, if the team is asking Aidnn to draft a report on monthly recurring revenue trends, “the data that you want to chat with actually doesn’t exist, at least the form that you need to chat with. It’s a multistep plan, a very complex plan: extract metadata, read the data, clean and normalize, join the data, prorate revenue, aggregate.”
The agent also shows its steps, reasoning, assumptions, and points to anomalies in the data. It will even make recommendations on how to proceed. Isotopes also promises that enterprise customers can deploy without sharing any of their data to the AI model makers powering the agent.
Still, as sophisticated as Isotopes may be, the startup faces plenty of competition. Incumbents like Salesforce’s Tableau already offer agents (amid Salesforce’s major AI agent push), and plenty of other startup founders with impressive pedigrees are in the market, too, such as WisdomAI.