Contents

Hudson Labs Co-Analyst vs. Perplexity AI: Which Delivers Better Results for Investment Research?

Blog Thimble Image
Aly Somani

Many professionals experiment with Perplexity for financial research. But despite its general usefulness, Perplexity’s AI is not built for investment workflows. In contrast, Hudson Labs Co-Analyst consistently delivers accurate, structured, and finance-specific outputs that outperform Perplexity across every dimension.

Looking for alternatives for Perlexity? See our previous blog with 10 alternative tools.

Is Perplexity Good for Financial Research? Here’s Why It Fails

Perplexity Pro can summarize information, but it suffers from several limitations in investment contexts. All answers below are generated using Perplexity’s latest model, Sonar. Here’s what we found:

  1. Frequent Hallucinations: See detailed analysis below. Perplexity did not generate a single complete and correct answer.
  2. Incomplete and Misleading Outputs: Perplexity often shows "N/A", despite data being publicly available.
  3. Untrustworthy Sources: Relies on third-party sites like Stock Titan, pushing official press releases far down the source list. 
  4. No Financial Math Capability: Can't compute percentage changes or interpret ranges (e.g., guidance bands). Provides no numerical interpretation of soft guidance.
  5. Unusable Format for Analysts: Default outputs are text-heavy and often start with actuals before guidance, confusing and poorly structured. Prompt engineering is required to get a table; outputs can't be copied directly into Excel for modelling.

Why Hudson Labs Co-Analyst is the Best AI Tool for Investment Research

The Co-Analyst is designed specifically for investment research. It delivers precision, structure, and relevance by default—no special prompting required.

  1. Accurate, Verifiable Data
    • The Co-Analyst pulls directly from company earnings calls, SEC filings, and conference calls. Earnings press releases are coming soon.
    • Precision is maintained down to the millions (not one decimal point).
    • Hallucinations are virtually eliminated.
  2. Guidance Parsing with Math Integration
    • Extracts recurring, one-time, numeric, and semi-quantitative guidance.
    • Handles ranges, percent changes, and annualized values with built-in mathematical reasoning.
    • Understands soft guidance and contextual references—something Perplexity fails to grasp.
  3. Structured and Tabular Output
    • Delivers tabular formats by default—ready for Excel.
    • Organizes data intuitively: guidance first, actuals second.
    • Verbatim quotes and source attribution are embedded in the table.
    • No prompt engineering is needed.
  4. Comprehensive Coverage
    • Pulls all available data, including overlapping KPIs across peers for true comparability.
    • Finds guidance and operational metrics even in niche segments like AMH/INVH occupancy or COST/WMT comp sales.

Investors also use Hudson Labs to assess forensic risks in governance, related parties, and management integrity, surfacing downside risks beyond traditional keyword searches.

Summary: How Perplexity and Co-Analyst Extract Data for Key Stocks

Query

Perplexity AI

Hudson Labs Co-Analyst

RPO at cybersecurity firms

3 out of 7 figures are wrong. Missing data.

All data is accurate with source verification.

META and AMZN capex

Hallucinated Q1’25, soft guidance missed.

No soft guidance retrieved.

Accurate guidance and actuals, correctly ordered by dates.

Soft guidance and verbatim guidance extracted. 

Comp sales at COST and WMT

Missing 5 quarters; poor labelling.

Complete comp sales history and accurate formatting.

AMH and INVH occupancy 

Wrong quarters and figures. Missing data.

Correct occupancy and period matching.

Detailed comparison of Hudson Labs and Perplexity output

Remaining performance obligation (RPO) for Palo Alto Networks $PANW, Cisco Systems $CSCO, and Zscaler $ZS

Here, 3 out of 7 numbers for CSCO were incorrect. Perplexity simply copied the numbers from another quarter. PANW and CSCO are also missing data from Q2 and Q3 2023.

Using perplexity ai to get remaining performance obligation - incorrect and missing numbers

This is compared to complete and accurate data extraction using the Co-Analyst.

Using Hudson Labs Co-Analyst to get remaining performance obligations - complete and accurate data

CapEx guidance for Meta Platforms $META and Amazon $AMZN

While Perplexity generated visually pleasing graphs, the default answer is misleading and only provides actual figures instead of guidance.

Using perplexity AI to get capex guidance - produces chart with actual numbers instead of guidance

We tried to engineer the prompt by requesting a table. META Q1’25 data was fabricated—it hadn’t been reported at the time of writing. Past quarters were misrepresented or incorrectly forward-projected for both companies (e.g., Q3’24 guidance copied from Q2). AMZN’s soft guidance wasn’t retrieved, despite being publicly available. 

Using perplexity AI to get capex guidance in table - misrepresented and hallucinated data

Perplexity also cited and searched for unofficial sources, lowering accuracy and credibility. The Hudson Labs Co-Analyst only searches from official documents like earnings releases and call transcripts.

Using perplexity Ai to get capex guidance - using unofficial, unreliable sources

The Co-Analyst generated accurate guidance. Verbatim quotations were also provided for source validation. 

Using Hudson Labs Co-Analyst to get capex guidance for Amazon - complete and accurate data
Using Hudson Labs Co-Analyst to get capex guidance for Meta in bullets - complete and accurate data
Using Hudson Labs Co-Analyst to get capex guidance for Meta - complete and accurate data

Actual comp sales for Costco $COST and Walmart $WMT

Perplexity has a misleading presentation, i.e. Costco’s last quarter should be Q2’25. Costco hasn’t yet reported results for Q3 and Q4 2025. The output is also missing 5 more quarters of data.

Perplexity was unable to retrieve existing information and provided this very unhelpful qualifier: “For precise quarterly data, especially for quarters not covered here, reviewing the companies' official financial reports would be necessary.”

Using perplexity AI to get comp sales - misrepresented and incomplete data
Using perplexity AI to get comp sales - misrepresented and incomplete data

The Co-Analyst generated correct and complete output with ease.

Using Hudson Labs Co-Analyst to get Costco and Walmart's comp sales growth - complete and accurate data

Occupancy at $AMH and $INVH 

All highlighted data is wrong, i.e. $AMH Q4’23 number is copied from Q4’24. The other numbers are not traceable. Perplexity output is again missing data for AMH for three quarters. 

Using perplexity AI to get occupancy data for AMH and INVH - incorrect and missing data

The Co-Analyst generated complete and correct numbers, along with management remarks.

Using Hudson Labs Co-Analyst to get occupancy data for AMH and INVH - complete and accurate data

Bottom Line: Perplexity Isn’t Built for Finance

Perplexity may suffice for consumer search or basic research, but for extracting data to be used in financial modelling, it's unreliable. The Co-Analyst is the clear choice when accuracy and consistency matter.

Start your free 14-day trial today. Try Hudson Labs Co-Analyst and see the difference in your equity research workflow. No prompt engineering. No hallucinations. Hudson Labs is an AI platform built for institutional investors. Unlike chatbots, we deliver consistent, reliable insights based only on source material – just the facts, not the fluff.  

Explore more tools for your equity research workflow: