As usage shifts from seat-based software to API-based workflows, a big question emerges: What happens to professional information / research terminal companies?

Old model: A hedge fund or IB team might have 20 analysts, each with a Bloomberg Terminal ($30K+ per year) or AlphaSense seat ($15K+ per year). Total spend can easily reach $300K-$600K+ annually.

New model: Buy API access, pipe the data into an internal Claude / LLM workflow, and let the whole team query it through one internal interface. In many cases, cost could fall by 50% or more.

The likely progression:
First, customers add API spend while cutting marginal seats - even without reducing headcount.
Next, vendors probably try to reprice API access higher, or move toward consumption-based pricing. A lot of these products were not built for heavy usage-based monetization.

AI centralizes information retrieval.
If the end interface becomes one chat box, why keep buying multiple seats for every employee?
Power users may still need the native product, but the value of the UI + search / query layer is clearly under pressure.

AI may not reduce data consumption.
But it could reduce the number of people directly logging into the original terminal.

And this goes well beyond finance.

Any category with these traits could face similar pressure:

• historically sold by seat

• a lot of the value is search / retrieval / summarization / basic analysis / light workflow

• the data or functionality can be exposed via API

• end users do not truly need the native UI

Examples:

• legal research / legal information

• medical information / clinical decision support / pharma knowledge tools

• customer support knowledge systems

• market research / expert network / transcript / survey data platforms

The key questions:

• Are users paying for the interface, or for the answer / data?

  • If it is the answer / data, that is riskier

• Do users spend hours a day living inside the product?

  • If yes, more defensible

• Is the data proprietary, or just a strong aggregation layer on top of public / semi-public info?

  • Aggregators look more exposed

• If the product were plugged into Claude, would 80% of the value still remain?

Bloomberg is more than just data lookup - and that distinction matters.
The pure information retrieval layer is exposed to AI.
But Bloomberg also has messaging, workflow, execution, trading, and a real two-sided network between buy side and sell side. That part is much harder to displace.

Finance has always been strangely fragmented:
Bloomberg has real-time data and some consensus.
Visible Alpha has the deepest consensus detail, but not all management guidance.
AlphaSense is a strong aggregator for sell-side research and expert content.
TradingView has great charting, but gaps elsewhere.

Maybe AI is finally the moment that breaks the walled gardens - and makes the whole stack more efficient.