In this episode of Selling Signals, we're joined by Vinesh Jha, CEO of ExtractAlpha and former quant researcher at Morgan Stanley's PDT, to explore how quantitative data products are developed, validated and commercialised.

Drawing on ExtractAlpha's new AI Earnings Transcript Model, Vinesh explains how his team identifies predictive opportunities, builds investment signals and differentiates its research from what hedge funds can develop internally.

We also get into the commercial realities of selling to funds: how they evaluate external signals, why research transparency matters, how ExtractAlpha structures trials and what vendors routinely get wrong when bringing datasets to market.

We finish by discussing how AI is changing the economics of quantitative research and allowing vendors to evaluate more potential sources of alpha.

In this episode of Selling Signals, we’re joined by Arsham Ghahramani, CEO of Ribbon AI. Arsham explains how a product built to interview job candidates found a new market in expert systems and investment research.

He describes a hedge fund that went from consulting a handful of experts to interviewing a hundred on the same question. We discuss what this changes about research and how Ribbon handles the confidentiality of a fund’s questions.

We also explore the value of expertise that has never been written down. Arsham shares his view on why demand for human knowledge could grow as AI improves, including the possibility of models calling experts when they get stuck.

In this episode of Selling Signals, we’re joined by Matthew Bernath, co-founder of Alternata. Matthew works with companies exploring whether their existing data could support a new business.

We discuss how to assess the commercial opportunity before committing resources. Matthew explains how prototypes built with synthetic data can help test a proposed product with buyers. He also shares his approach to pricing around the value a customer expects to receive.

The conversation turns to consumption pricing and Matthew’s preference for long-term relationships in which buyers help improve the product. He explains why one-off AI training deals can still make sense when the revenue justifies the work involved.

Throughout the discussion, Matthew returns to one constraint. A data business should not damage the core operation that produces its data. We explore how that affects the opportunities a company should pursue.