Cyrus Khajvandi, a Stanford biology graduate and seasoned entrepreneur, faced significant challenges balancing his daily workload with the need to stay current on scientific research. Realizing he wasn't alone in this struggle, especially with the increasing accessibility of AI technology, Khajvandi set out to create a platform capable of summarizing and answering questions about documents, particularly scientific studies.
The result was Humata AI, which launched in February, with former Labelbox founder Dan Rasmuson stepping in as CTO. The platform quickly made waves, processing tens of millions of pages and amassing a user base in the millions. Humata AI secured $3.5 million in funding from notable investors, including Google’s Gradient Ventures, ARK Invest, and M13.
“Our mission at Humata is to enable individuals and organizations to make smarter, faster decisions by allowing them to inquire about their files,” Khajvandi shared in an email interview. “Humata functions like [OpenAI’s] ChatGPT, but for all your documents.”
Humata AI prides itself on its straightforward functionality. Users can easily ask questions related to their files—primarily PDF documents—and receive precise answers. By uploading one or multiple PDFs, users can query the content, catering to a diverse clientele ranging from academics to professionals in fields such as law, oil and gas, and customer support.
While other chatbots, including ChatGPT and Anthropic’s Claude, have begun offering similar document analysis features, Khajvandi argues that Humata’s streamlined focus and specialized functionality make it a more powerful tool. “Users can ask the AI any question and receive instant answers with highlighted references from their data,” he explained. “Recent advancements in AI are enabling every worker to receive quick answers to their inquiries.”
However, AI's summarization capabilities are not without flaws. Fast Company tested ChatGPT's summarization skills and found that the model sometimes misinterpreted information, omitted key details, or even fabricated facts not present in the original documents.
Privacy is a pressing concern as well. Many organizations and individual users may hesitate to upload sensitive documents to Humata's platform for processing.
Khajvandi defends Humata’s summarization quality, assuring that the company’s models have been trained on varied datasets and undergone rigorous bias testing. He emphasizes that Humata only collects essential data and employs “strong safeguards” to thwart unauthorized access.
“We ensure informed consent, helping users understand what they agree to,” Khajvandi added. “As our AI evolves, we take care not to infer sensitive information without explicit permission. We follow legal and ethical standards across various regions, making Humata a trustworthy enterprise solution.”
With thousands of customers now subscribed to its paid plans, Humata intends to utilize its current capital of $3.58 million (which includes pre-seed funding) to enhance its AI capabilities, improve user experience, and expand its market presence.
“We decided to raise funds now due to the increasing demand for efficient, AI-driven solutions to synthesize insights from extensive enterprise documents,” Khajvandi stated. “This funding will enable us to develop new features, refine our existing products, and venture into new markets—ultimately empowering businesses to make informed and rapid decisions with their data using Humata.”
Ranking
