Theia Insights raises $8 million to replace outdated industry classification systems

Theia Insights raised $8M in a Series A round to develop an AI system that describes the economy and business models of companies in a new way. Instead of a single industry label, the startup offers a dynamic map that shows exactly what a company earns from and how this changes over time.

Why old schemes don't work

The problem that Theia is targeting is well illustrated by the example of Amazon. Formally, the company can be classified in the retail category, but this immediately loses half the picture: it has cloud business, logistics, an advertising platform, media services, and hardware. For an investor, bank, or index provider, such rough classification distorts the real structure of the business.

And yet many indices, portfolios, and risk assessment models are still built on such schemes. Old industry standards like GICS and ICB were created in an era when companies were much less multi-layered. Now large public players operate in multiple markets simultaneously, quickly change product lines, and redistribute revenue between segments.

If the system sees only one label, it is poorly suited not only for analysts, but also for new AI tools in finance, which need more accurate, structured, and machine-readable data.

How the platform works

Theia says it builds a self-learning map of the world economy. To do this, the platform uploads regulatory filings, investor call transcripts, press releases, and financial data from public companies, and then applies its own NLP and quantitative modeling methods. The output is not a static label, but a multidimensional model of the company that tracks what everyone does and what share of the business each direction occupies as revenue changes. On this basis, the company has already assembled several products for the institutional market:

• TIIC — dynamic industry classification system
• Concept2Universe — a tool that translates investment themes into a justified list of companies
• Thematic Factor Model — a model that links stock movement to structural trends
• Theme Watch Indices — indices for tracking global industry themes in real time

The startup was founded in 2022 by Dr. Ye Tian, a former Amazon Alexa researcher with a PhD background in NLP and AI. The team is assembled at the intersection of technology and finance: it includes people with experience at Nasdaq, Morgan Stanley, Meta, UC Berkeley, and Cambridge University. This is important for Theia's product: it must simultaneously be understandable to the capital market, accurate enough for quant models, and suitable as infrastructure for new AI workflows in investment analysis.

Where the money goes

The round was led by MiddleGame Ventures, and also participated in by Further Ventures and Unusual Ventures, which had previously invested in the company. After the new deal, Theia's total raised capital reached $14.5M. The money will go not only to commercial growth, but also to expanding the engineering and research side of the product, meaning the company intends to deepen the technology base itself, rather than simply scaling sales.

"Financial markets still rely on static classification systems that have barely changed in decades," is how

MiddleGame Ventures describes the problem it's investing in.

A separate goal of the new round is to enter the private assets market. By Theia's logic, that's where the gap is especially noticeable: for private companies, there is almost no comparable dynamic classification system, although institutional money is increasingly flowing into closed assets. If the startup can transfer its approach from the public market to this segment, it will get not just another class of customers, but a chance to become the basic data layer for a wider range of financial solutions.

What this means

Theia is betting that the next generation of financial software needs not PDF reports and not outdated industry labels, but a living map of the economy that both people and models can work with. If this approach takes hold, AI in investment analysis, portfolio construction, and capital allocation will get a more accurate foundation than the familiar classifiers of the previous generation.