← All insights
Founding Thesis·8 min read

On the Gap Between Bloomberg and Robinhood

The founding thesis, in long form.

The financial services market is, when one looks at it honestly, organised around two poles. At one end sits the Bloomberg Terminal, an instrument of formidable capability whose annual subscription places it beyond the reach of any individual not employed by an institution. At the other sits a class of consumer brokerage applications whose virtues, real enough in their own register, are oriented towards the speed of execution rather than the depth of analysis. Between these two poles lies a substantial and underserved population.

The available tools at this middle altitude are either thin renderings of Terminal-class capability at Terminal-class prices, or marginal extensions of the consumer brokerage model with little real analytical character.

The Underserved Middle

Who occupies this middle territory? In our reading, four populations of professional or quasi-professional users, each with material capital at risk and each ill-served by the present offerings.

  • The serious individual investor, managing a portfolio of meaningful size with the analytical seriousness of a professional, but without an institutional affiliation that would grant access to professional tools.
  • The independent wealth manager or family-office principal, advising on portfolios that justify analytical depth, but whose firm is too small to bear the Bloomberg cost structure across its desks.
  • The financial advisor working in a non-institutional setting, for whom the existing tools are either too expensive or too superficial.
  • The small to mid-sized investment fund, particularly in jurisdictions outside the major financial capitals, for whom every line item in the technology budget is examined.

What This Population Actually Needs

It would be a mistake to suppose that what this population requires is Bloomberg at a discount. The Terminal is a product of its institutional context, designed for users who sit at it for ten hours a day and who derive value from the breadth of its function. Reproducing that breadth at a fraction of the price would be, in commercial terms, a mug's game; the cost structure does not permit it.

A more useful question is this: what analytical work does the underserved population actually require, and could a far narrower instrument perform that work to a standard sufficient to displace ad-hoc methods? Our view is that it can. The work that matters most to portfolio decisions, the work of understanding a session in context, of testing scenarios under stress, of computing risk on positions actually held, is well within the capability of a carefully constructed product. It does not require the full breadth of a Terminal. It requires depth, in the specific places that depth matters.

The Role of Artificial Intelligence

The recent advance of large language models has materially altered the economics of this work. Analytical commentary that, five years ago, would have required either a human analyst or a brittle template-driven system, can now be generated reliably from a properly constrained model with access to verified data. This is not a trivial advance. It collapses the cost structure of producing institutional-quality commentary at a price the underserved middle can pay.

The qualification is important. Generated commentary is reliable only when the model is properly constrained and the data context is properly verified. Without these disciplines, the output is plausible-sounding nonsense, which is in some ways worse than no output at all. The work of building a credible product in this space is therefore largely the work of building the verification layer that ensures every figure stated has a verifiable source and every claim has a defensible basis.

What Follows

Drusus is the instrument that follows from this thesis. It is not a Terminal, nor a brokerage application. It is a focused analytical product, oriented towards the work the underserved middle actually does, priced for the audience it serves, and constructed with the verification disciplines required to make AI-generated analysis trustworthy. Whether the thesis was correct is, of course, a question only the market can settle.