Josh Green and the trouble with a good idea

In 2005, Josh Green was given an assignment that sounded like an afternoon’s work: find a supplier in China. He was working at E Ink, the display company whose technology would become familiar to Kindle readers. There was an internet. There were factories. Surely the two could be introduced without too much ceremony.

The search produced plenty of candidates and very little confidence. A factory could tell you what it made; deciding whether to believe it was another matter. Reliability, working practices, the experience of previous customers: useful information seemed to disappear precisely when somebody needed to make a decision. Green’s sister, working in apparel, recognized the difficulty. Different products, same fog.

The predicament had the makings of a business. Green and James Psota, an MIT computer scientist, founded Panjiva in 2006. Their ambition involved bringing the convenience of online discovery to global commerce. A buyer looking across an ocean ought to have something more useful than a promising description and a telephone number.

Green had studied government at Dartmouth and earned graduate degrees at Harvard’s business and government schools. He graduated from Harvard Business School as a Baker Scholar and had worked at Boston Consulting Group. That education supplied several ways to analyze a problem. An actual supplier search supplied the problem itself. The distinction would become useful more than once.

Nobody wanted to review their supplier

An early Panjiva concept borrowed from the consumer internet: let suppliers build reputations through ratings. It sounded reasonable until prospective participants explained their incentives. Suppliers wanted an audience before investing in a new site. Buyers wanted useful information before visiting. And a buyer with a wonderful supplier had an excellent reason to keep quiet. Why provide a shopping list to the competition?

Public customs records offered another route. They documented transactions that had already happened. Panjiva gathered shipment information and connected it to companies, making the activity of global trade searchable. The company became a subscription service for examining commercial shipments and the relationships between buyers and sellers. A repeat order could reveal something a polished sales pitch could not.

Making those records useful involved a less photogenic task: cleaning them. Names were inconsistent. Formats were awkward. The same factory could appear under different spellings. Green and Psota worked on the algorithms needed to organize that disorder. By July 2009, Panjiva had attracted customers including Home Depot. The raw material was public; the work of turning it into a usable product was substantial.

There was a pleasingly physical detail to this information business. Some government data arrived on CDs delivered by UPS. While the product promised a clearer view of international commerce, part of its supply chain still involved somebody carrying a disc to the office. The future, apparently, required a delivery signature.

The investor who removed an option

Panjiva raised venture capital in 2008, shortly before the financial crisis. An investor then gave Green an uncomfortable assessment: another financing would be difficult to obtain, so the company needed to become a business that could support itself. The room for an elaborate marketplace experiment had narrowed.

Green later described the discipline this imposed. The team had to work out where its strengths met a problem customers would pay to solve. Data was that meeting point. The original marketplace ambition gave way to selling information. A company could use shipment records to understand trading relationships and competitors’ activity without waiting for a new trading marketplace to materialize.

That is an unceremonious way to improve an idea. The customer declines the grand proposal and buys the apparently smaller thing. For a founder, accepting the transaction means giving up some authorship. It also supplies a useful piece of evidence: someone has decided the work is worth money.

In 2018, S&P Global acquired Panjiva. Green subsequently joined the leadership of S&P Global Market Intelligence. The business that began with uncertainty about a supplier now belonged inside a financial information company. His career had moved from a question about whom to buy from to an organization serving people whose decisions depended on understanding companies.

A year off, interrupted

After Panjiva, Green intended to spend a year exploring what to do next. He wanted work his children could be proud of. Six weeks into that exploration, S&P contacts introduced him to an emerging project involving private markets and sustainability. He had said he would never build another startup. Apparently, the sentence had a short shelf life.

Alex Friedman and the other people around the project had identified a reporting problem among privately held businesses and their investors. Green was initially skeptical about the combination. Conversations with people in the market changed his view: expectations were passing from capital providers to fund managers and onward to the companies they owned. There was demand for help.

Novata launched publicly in October 2021 as a public benefit corporation. Its founding consortium included the Ford Foundation, S&P Global, Hamilton Lane, and Omidyar Network. Philanthropy, investing, and financial information had brought different interests to a shared reporting task. Green joined as co-founder and chief operating officer; he became co-CEO alongside Friedman in October 2025.

The early product thinking needed adjustment, too. Benchmarking would help investors understand how portfolio companies compared, but prospective customers had a more immediate difficulty: collecting the information in the first place. A comparison is of limited comfort when the relevant cells are empty. Novata’s work had to begin earlier in the process.

The number has to travel

By 2026, Novata listed more than 400 clients and more than 13,000 contracted companies, with over 300 benchmarked metrics. It had become a B Corp-certified business as well as a public benefit corporation. These are company milestones, assembled by a team; they give a sense of the organization Green now helps lead.

His explanation of sustainability data starts with its movement. Information crosses organizational boundaries: a business answers an investor, a supplier responds to a customer, and the same underlying facts may be needed again in another reporting process. That movement makes definitions, permissions, and consistency part of the product. A number needs to survive the trip.

In January 2025, Green described a practical approach he called “of course, sustainability.” He argued for incorporating a long view into ordinary business decisions: choosing suppliers, understanding customers, and preparing for changing conditions. He wanted the work connected to business strategy. The phrase is deliberately undramatic. It sounds like something that could fit into a meeting without requiring a new department to explain it.

His view of technology follows from the workload. Reporting demands can grow faster than the teams handling them. Software can help map requests to information already held, track changes, and reduce repetitive work. That leaves more room to decide what to improve. For Green, the destination is a company using its information to act, with less time spent simply arranging it for somebody else.

“Of course, sustainability.”

Josh Green · January 2025

A roomful of hands, then none

At Climate Week NYC in 2026, Green saw a small audience experiment expose a larger disconnect. Asked whether AI was a sustainability issue, people raised their hands. Asked whether it was being managed as one, nobody did. In his October reflection on the week, he returned to the gap between recognizing a problem and organizing a response.

He connected AI with energy demand, governance, and commercial risk. He also emphasized human oversight when AI is used in business processes, especially where outputs feed regulatory reporting. The concern is practical: an answer can read smoothly and still be wrong. Faster production creates another reason to check what is being produced.

Novata’s integration with Briink makes that question concrete. It converts documents, spreadsheets, and other unstructured material into sustainability responses, with references and an audit trail. Users can review outputs, edit assessments, and add context. The process still needs judgment. A machine that drafts the answer has also created something a person must be able to inspect.

The thread back to Panjiva is easy to follow. A factory’s description, a shipment record, a portfolio company’s submission, and an AI-generated response all ask the reader to place some confidence in information. Each arrives through a different route. The business opportunity lies partly in making that route understandable enough for somebody else to use the result.

“You can’t have regulatory outputs that are hallucinated.”

Josh Green · October 2026

The people who keep him at it

Green’s account of building companies makes room for the people doing the work. In his 2023 conversation with Propel’s Kathryn Lockhart, he emphasized the role colleagues play in sustaining a founder through difficult stretches. Listening, honest feedback, sales, and teams were central subjects. The founder’s determination had company.

There is another, quieter connection to learning. In 2023, Green became board chair of Reach Out and Read of Greater New York, a childhood literacy nonprofit whose board he had served on for years. The organization’s announcement described a New York household with his wife Lindsey and two children who loved books. His interest in what the next generation might think of his work has a life beyond the startup.

In April 2026, he prepared to take over as interim host of Sustainable Intelligence. He wanted guests to share reasons for optimism and recommend an article that had stayed with them. Those are useful questions for someone who has repeatedly had to revise an attractive plan. They leave space for another person to supply the unexpected detail.

A good idea has taken Green quite a long way. So have the customers who declined it, the partner who helped rebuild it, and the colleagues who stayed to do the work. At Novata, he is again asking what information people need before they make a decision. Experience has given him a further question to keep nearby: will the answer change the plan?