In 2018, Xfund portfolio company Kensho Technologies was acquired by international financial intelligence firm S&P Global. It was the largest acquisition of an AI company ever at that time. Harvard University PhD, visual artist, poet, and AI-focused entrepreneur Daniel Nadler founded Kensho in 2013. The company works with natural language data and complex texts to develop machine learning techniques and mutually compatible products. Its goal is to add structural layers and contribute nuanced insights to the data processing and data enrichment processes.
The company continues to operate as an independent entity under its own brand. Kensho offers a major boost to the quality of the information S&P delivers to clients like government agencies, individuals, and corporations. In 2023, this collaboration received a Waters Rankings recognition as a Best Artificial Intelligence Technology Provider. WatersTechnology is among the major providers of news on how tech and AI affect financial markets, and WatersTechnology readers decided on the award.
One of Xfund’s central commitments is working with founders known for independent, lateral thinking and a global perspective on the liberal arts. Identifying and supporting talented early-stage founders is at the core of the venture funder’s DNA.
Xfund, along with investors NEA, Breyer Capital, and Accel, was among the first to recognize Kensho’s capabilities and invest in the emerging company. Pioneering fintech executive and Xfund Limited Partner R. Martin Chavez, then-CFO of Goldman Sachs, was also among Kensho’s major early investors and enthusiastic customers.
“Daniel is Xfund’s ideal archetype of the liberal arts founder,” said Patrick S. Chung, Xfund’s Managing General Partner. “While creating what turned out to be the largest AI exit in history, Daniel wrote a book of 100 imagined ancient love poems that was very favorably reviewed by the New York Times, and which caused Daniel to be inducted as the youngest member of the American Academy of Poets. He is an astounding talent.”
Xfund connected with Daniel Nadler while he was still at Harvard, anticipating becoming an economist at the Federal Reserve. His work with Kensho began when he created a tool to help users conduct high-quality financial and arbitrage analysis by asking questions and plain English, without the need to write code.
The build-out of Nadler’s ideas for this analytical tool became a key investment for Xfund, which provided seed funding and recruiting assistance at Harvard and the Massachusetts Institute of Technology. Xfund’s collaboration with Kensho then extended to its investment in Synapse, founded by a Kensho engineer to apply machine intelligence to defense and airport security. The funder’s long-term connection with Nadler now extends to the development of his newest venture, Open Evidence.
Within its first five years, Kensho produced analytical systems that were successfully implemented by some of the world’s largest financial services organizations, as well as the United States government. At its 2017 summit, the World Economic Forum recognized Kensho among the world’s “most innovative and impactful” tech-focused companies.
Kensho designs AI solutions with the specific, confidential needs of partners like S&P Global in mind, training its AI on its S&P Global’s proprietary datasets. By collaborating with Kensho, S&P Global can organize large amounts of disordered data in a variety of formats, including audio and text.
This is the key value-added service Kensho offers: speeding up data collection and assembly; increasing ease of access while highlighting relevancy; and broadening and deepening the data universe available to decision-makers more than ever before. When assembled into manageable, meaningful, structured systems, this data can be deployed to help corporate decision-makers work with greater awareness and precision.
Kensho’s leading products include Scribe, an HITL (human-in-the-loop) voice-to-text transcription system. Scribe takes transcription to new levels of accuracy, making it an essential tool for professionals in finance and business. Its capabilities can enhance transcription of voicemails, interviews, podcasts, depositions, and numerous other applications.
Capable of handling the complexities of multi-speaker, real-world audio, Scribe fuses the best capabilities of AI and human insight. Its in-house transcriptionists and editors receive comprehensive security and confidentiality standards training. They are also familiar with industry- and region- specific vocabulary.
NERD (Named Entity Recognition and Disambiguation) offers clients the ability to identify actionable information on market trends in real time, facilitating’s research and analytics. It understands real human speech in context and with people, companies, and other reference points clearly identified. Further, it is capable of the same quality of perception that humans apply to analyzing information.
NERD quickly and efficiently disambiguates entities from those with similar names or nicknames, making large amounts of text increasingly meaningful and searchable. It further anchors these entity reference points to the other data layers populating Wikimedia and S&P Capital IQ to help uncover additional insights. The growing list of entities NERD can recognize totals well over 100 million.
Other Groundbreaking Products
Kensho’s Classify represents a major advance over traditional keyword recognition and content assemblage. Classify applies AI advances to the job of tracing identified concepts throughout given texts accurately so users can make and track more connections through massive amounts of verbiage and data. This gives clients significantly faster concept training and annotation speeds and superior prediction quality. They can take customized content creation and tagging to new levels.
Extract applies automated text and table extraction to facilitate document processing while maintaining the accuracy major businesses and government agencies require. This product updates obsolescent optical character recognition tools by applying machine learning to interpret unstructured layouts of text and bring multiple types of analog documents into machine-readable form. Further, Extract can then identify specific data points within these documents. Its strengths include supporting natural language processing, making translation and annotation easier, and reducing the need for manual searching.
As generative AI products become increasingly powerful, Kensho looks toward delivering solutions with increasing power to transform entire industries.