On Thursday, Isotopes emerged from stealth mode and successfully completed a $20 million seed funding round.

The company has launched an AI agent designed to address a longstanding challenge in data analytics products: the people who understand big data infrastructure operations are not the same people who actually need to use the data.

Using large language models (LLMs), business managers can query data through natural language. Isotopes' agent product, Aidnn, not only provides answers and drafts complex planning documents but can also collect data from various storage scenarios including financial applications, enterprise resource planning (ERP) systems, customer relationship management (CRM) systems, and cloud storage.

While the market already has countless business analytics products with agent capabilities, Isotopes co-founder and CEO Arun Murthy told TechCrunch that his team members possess unique industry backgrounds, making their product highly technically complex, with the company having filed 10 patents.

About 20 years ago, Murthy, then around 25, worked at Yahoo and participated in developing the open-source project called Hadoop. It was Hadoop that sparked the first "big data boom" in the early 2010s.

In 2011, Yahoo spun off its Hadoop-related business to form the independent company Hortonworks, where Murthy served as co-founder and Chief Product Officer. The company went public just four years after its founding. However, the rise of new cloud storage technologies impacted Hadoop's market share, and Hortonworks eventually merged with its biggest competitor, Cloudera. In 2021, after intervention by prominent activist investor Carl Icahn, the merged company went private.

During this industry turbulence, Murthy worked at Cloudera for several years, managing a team of about 200 people. But he said that even at this company, he still discovered that "long-standing data access problem." He recalled quarterly conference calls with Wall Street analysts where analysts would repeatedly question executives about operational details, and executives often couldn't provide answers because they couldn't access the required data. "Looking back, it's embarrassing," Murthy admitted. "We were a company selling big data solutions."

When Murthy left in 2021, he didn't have a clear next step planned. Later, a venture capitalist introduced him to Scale AI's Alexandr Wang. After several conversations and completing some consulting work, Murthy joined Scale AI as Chief Technology Officer (CTO).

He described this experience as "getting a PhD at Scale AI": "deeply understanding the core logic driving these models and how to optimize them."

It wasn't until his former Hortonworks colleague Prasanth Jayachandran called that the two decided to co-found an AI startup. They then convinced Gopal Vijayaraghavan, who also had Hortonworks experience, to join, and the three officially established Isotopes at the end of 2024. This seed round was led by NTTVC's Vab Goel (Goel previously worked at NorWest Ventures).

Leveraging the founding team's industry background, they developed an AI agent that can not only locate data from various storage platforms (whether Salesforce.com or Snowflake) but also complete data cleaning tasks. Additionally, the agent has powerful contextual memory capabilities, supporting complex task processing.

"This is far more than just a simple chatbot," Murthy said. For example, if a team asks Aidnn to draft a monthly recurring revenue (MRR) trend report, "the data you want to obtain through conversation doesn't actually exist in a 'directly conversational' format. This requires a multi-step, highly complex process: extracting metadata, reading raw data, cleaning and standardizing, data correlation, revenue apportionment, and data aggregation."

The agent also reveals its processing steps, reasoning logic, core assumptions, and marks outliers in the data, even providing suggestions for subsequent operations. Isotopes also promises that when enterprise clients deploy this agent, they don't need to share any data with the AI model developers providing technical support.

However, despite Isotopes' competitive technology, this startup still faces intense market competition. Industry leaders like Tableau under Salesforce.com have already launched similar agent products (Salesforce.com is heavily promoting AI agent business); additionally, the market has numerous startups founded by founders with "impressive backgrounds," such as WisdomAI.