SignalPress Blog
← All Articles
April 5, 20269 min read

AI Investment Research Tools: What Actually Works for Fund Managers

An honest comparison of AlphaSense, Bloomberg AI, Hebbia, Kensho—and what they all still miss

The AI investment research market has exploded. Every Bloomberg terminal now has a "GPT" button. AlphaSense raised $650M and acquired Tegus. Hebbia secured $130M from Index Ventures. Kensho processes 750,000+ financial data points per second under S&P Global.

Institutional investors have never had more AI research tools available—and have never been more confused about which ones actually matter.

This isn't a vendor pitch. It's an honest breakdown of what each major platform does well, where each falls short, and what fund managers still can't get from any of them.

The Landscape: Four Categories That Matter

The AI investment research market has consolidated around four distinct approaches:

1. Enterprise search & synthesis (AlphaSense): AI-powered search across millions of documents 2. Terminal AI layers (Bloomberg Terminal AI): AI features bolted onto existing data infrastructure 3. Document intelligence (Hebbia): Deep analysis of large document collections 4. Quantitative signal extraction (Kensho): Structured data derived from unstructured text

Understanding what category a tool belongs to is more important than reading their marketing copy.

AlphaSense: The 500M-Document Search Engine

AlphaSense is the market leader for a reason. Its core value proposition is simple: it lets you search across 500M+ documents—broker research, SEC filings, earnings transcripts, news—with semantic AI search that understands financial context.

What it does exceptionally well:

The honest limitations:

AlphaSense is a search-and-discovery platform. It excels at finding information; it doesn't generate investment conviction. You'll surface 14 analyst reports mentioning NVIDIA's data center margins—but you won't get a synthesized view of what those 14 reports collectively imply for your thesis.

The platform also requires a trained user to ask the right questions. Junior analysts using AlphaSense without research methodology still produce mediocre work.

Pricing: Enterprise contracts starting at ~$25,000–40,000/year per seat. No self-serve tier.

Best for: Research teams at hedge funds, asset managers, and sell-side firms that need to search across large document universes quickly.

Bloomberg Terminal AI: The Infrastructure Layer

Bloomberg Terminal's AI features aren't a standalone product—they're enhancements to the world's most embedded financial data platform. BloombergGPT, launched in 2023 as a 50B-parameter model trained on financial data, powers several in-terminal AI capabilities.

What it does well:

The honest limitations:

Bloomberg's AI is additive, not transformative. If you already pay $27,000/year per terminal for the data layer, the AI features are incremental improvements—not a new research capability. The terminal's strength is real-time data and market microstructure, not deep fundamental analysis.

For long-horizon fundamental equity research, analysts consistently report preferring AlphaSense for document search and using Bloomberg for pricing, data, and live alerts. They're complementary, not competitive.

Pricing: ~$27,000/year per terminal (inclusive of AI features). No standalone AI product.

Best for: Traders, macro strategists, and quant teams who live in the terminal and want AI without switching workflows.

Hebbia: The Document Intelligence Platform

Hebbia occupies a specific niche: extracting structured insight from massive, unstructured document collections. Its Matrix product uses AI agents to break complex research tasks into sub-tasks across hundreds of documents simultaneously.

What it does well:

The honest limitations:

Hebbia's strength is breadth across many documents—not depth on a single company. For a focused fundamental analyst covering 20–30 names, Hebbia's multi-document processing power is often overkill. Its setup time and learning curve also exceeds most other platforms.

It's primarily a document retrieval and extraction tool. Like AlphaSense, it doesn't generate narrative investment conviction—it organizes information so you can generate conviction yourself.

Pricing: Enterprise contracts, typically $20,000–$50,000+/year depending on user count and document volume.

Best for: Quant fundamental teams, private equity shops analyzing large deal pools, and research-intensive firms running systematic comparisons across large universes.

Kensho: The Quantitative Signal Extractor

Kensho (acquired by S&P Global for $550M in 2018) is the most differentiated platform on this list. Rather than document search, Kensho transforms text into structured data—turning earnings call language into time-series signals, and news events into quantifiable market impact estimates.

What it does well:

The honest limitations:

Kensho operates at a level of abstraction that's genuinely useful for quant strategies but less actionable for discretionary fundamental investors. If you're building a systematic strategy around earnings call sentiment, Kensho is relevant. If you're a long/short equity PM deciding whether to add to a position after a quarter, Kensho won't directly help you write the investment memo.

Pricing: Enterprise only, typically bundled with S&P Global data subscriptions. Pricing not publicly disclosed.

Best for: Quantitative hedge funds, systematic macro funds, and data teams building models that ingest financial text as a signal.

The Comparison Table

ToolCore CapabilityWhat It's ForPrice RangeGap
AlphaSenseDocument search & discoveryFinding existing research$25K–40K/seat/yrDoesn't synthesize into thesis
Bloomberg Terminal AIReal-time data + AI overlayTrading & live market intel~$27K/terminal/yrNot built for deep fundamental work
HebbiaMulti-document extractionComparing many companies at once$20K–50K+/yrHigh complexity, no conviction scoring
KenshoText-to-structured-dataQuant signal generationEnterprise onlyNot for discretionary PMs
SignalPressSEC filing → narrative briefCompany-level thesis generationFree trial availableNarrower scope (public equities)

What All Four Are Still Missing

Here's the honest assessment that no vendor will give you: every major AI research tool on this list helps you find and organize information. None of them generates investment conviction.

The missing layer is what professional investors actually need:

Narrative Thesis Generation

"Revenue beat consensus by 12%, driven by data center strength, though gross margin compression suggests pricing pressure in gaming" is not what you get from a search result. It's the synthesis that turns data into a position.

Most AI tools treat thesis generation as the user's job. They surface the inputs; the PM does the synthesis. For a fund manager running 50+ names across multiple geographies, this remains a significant time sink.

Conviction Scoring

Is the current setup positive, negative, or neutral? Is momentum accelerating or decelerating? The market has signal-extraction tools (Kensho), document-search tools (AlphaSense, Hebbia), and data terminals (Bloomberg). It has almost no tools that deliver a confident directional signal with context.

The tools that approximate this—FactSet's earnings analysis, Capital IQ's consensus tracking—are backward-looking aggregations of what other analysts think. That's not independent conviction. That's consensus.

Where SignalPress Fits

SignalPress is not a replacement for any of the above platforms. It's a focused tool for a specific job: turning SEC filings into narrative investment research.

While AlphaSense searches across 500M documents for what analysts have said about a company, SignalPress goes directly to the primary source—SEC EDGAR filings—and generates the analytical narrative itself. No human analyst wrote the brief. The AI read the 10-K, extracted the signals, and wrote the thesis.

The output is a structured investment brief with:

[The full pipeline is explained here](/blog/how-ai-reads-sec-filings): how an AI agent reads 212 pages of EDGAR filings in 14 seconds and generates a brief that tells you the "so what."

For a PM who uses AlphaSense for broker research and Bloomberg for market data, SignalPress adds the missing layer: independent AI-generated fundamental analysis from primary sources, not aggregated from what the sell-side has already said.

How to Think About Your AI Research Stack

No single tool covers the full research workflow. The most effective setups combine platforms by function:

Discovery & monitoring → AlphaSense (alerts, broker research search) Live market data & execution → Bloomberg Terminal Multi-company comparisons → Hebbia (for large universes) Quant signal generation → Kensho (for systematic strategies) Primary filing analysis → SignalPress (direct from EDGAR, narrative output)

The test for any AI research tool is simple: does it produce output you can put in front of a PM without editing? Or does it produce a list of documents the PM still has to read?

Most AI tools are still very good at finding things and limited at making the case. The gap is closing fast—but it hasn't closed yet.

Start With 3 Free Briefs

See what AI-generated filing analysis looks like for a ticker you're actually following. No subscriptions, no sales calls, no pitch decks.

Enter any ticker on SignalPress. You get the thesis, the metrics, the risks, and the conviction score in 14 seconds—pulled directly from the most recent SEC filing.

The analyst hours you save on reading become analyst hours you can spend on judgment.

Want to learn how to read filings yourself? See our guide: [How to Read a 10-K Filing in 2026](/blog/how-to-read-10k-filing)

See this in action — Get 3 free briefs →

Enter any ticker. AI extracts the key data from SEC filings and generates an investment thesis in 14 seconds.

Start Free