NinethirtyAI lets you screen US stocks and chat with the data in

Managing a dozen open tabs for charts, financials, and screeners while copy-pasting numbers into an LLM for analysis is a productivity killer. I've been using NinethirtyAI to consolidate this entire AI workflow, and it actually solves the fragmentation problem by layering a chat interface directly on top of a curated data engine.

The core strength here is the screener. Most tools force you to use three different modules for fundamental, technical, and news-based criteria. NinethirtyAI lets you combine these into a single query. More importantly, it has a built-in backtesting feature that shows you which stocks would have hit that specific screen historically and how they actually performed. It turns a "hunch" into a data-backed strategy pretty quickly.

The platform includes the standard toolkit—heatmaps, sector views, and seasonality—but the real interest is "Dr. Market," the AI assistant. Unlike a generic LLM that might hallucinate a company's P/E ratio or make up a price target, Dr. Market operates on a processed data layer. When you ask it to validate a technical setup, it's reading real-time indicators. It doesn't completely kill hallucinations (no LLM is perfect), but the errors are usually based on bad interpretation of real data rather than the AI just inventing numbers out of thin air.

Practical breakdown of the features

• Unified Screener: Combines technicals, fundamentals, and news events in one go.
• Instant Backtesting: Validates screen performance against historical data.
• Data-Driven AI: An LLM agent that accesses real-time equity data to minimize "plausible-sounding" lies.
• Market Intelligence: Sector-wide views and event tracking integrated into the dashboard.

Since it's still in the early stages, it only covers US equities and definitely has some rough edges. It's designed for analysis, not as a financial advisor—it won't tell you "Buy this stock," but it will tell you why a stock fits your specific criteria.

For anyone looking for a real-world application of prompt engineering in finance, this is a great example of how a RAG-like (Retrieval-Augmented Generation) setup can make an LLM actually useful for equity research. Instead of trusting the model's internal weights for financial figures, the system feeds the model the actual numbers first.

If you're into quantitative analysis or just tired of jumping between five different websites to vet a single ticker, this is worth a look. I'm particularly interested in seeing if the screener can handle complex, multi-factor queries without breaking.