The analysis below compares the Hudson Labs Co-Analyst to Perplexity Pro across investor-specific workflows. Results show that the Co-Analyst consistently outperforms Perplexity Pro in multi-period retrieval, guidance extraction, and numeric precision.

This is not a claim of overall AI superiority—our findings apply to targeted investment-research use cases. Perplexity works well as an AI-powered search engine for general queries, but fails in investor-specific queries, as seen below.

Perplexity Pro’s web-driven architecture allows for broad topical coverage but struggles when data spans multiple reporting periods or requires interpreting nuanced context. In investor workflows, these gaps often translate into missing or fabricated figures.

The Co-Analyst avoids these pitfalls through a multi-model architecture (more than five LLMs), proprietary retrieval, precision source selection, and pre-/post-processing tuned to assist in investment research workflows with precise and accurate results. Also see comparisons of the Co-Analyst to OpenAI's o3 and ChatGPT-4o.

Where the Hudson Labs Co-Analyst Excels:

Multi-period document analysis

The Co-Analyst excels when analyzing multiple long-form documents, such as earnings calls spanning more than four quarters. Its proprietary retrieval system scales better than generalist alternatives.

Guidance identification

Guidance often requires interpreting subtle context—sometimes from previous sentences or adjacent paragraphs. Our multi-model architecture ensures accurate extraction, even across periods.

Reliable results

Unlike other tools, Hudson Labs only generates answers using information from source, not model memory. The Co-Analyst minimizes hallucination through confidence-based decoding and post-processing. When information is unavailable, it returns “N/A” rather than fabricating results.

Numeric focus

Many AI systems provide general commentary but omit key figures. The Co-Analyst prioritizes numeric specificity and precision.

Under-covered names

Unlike models reliant on web search, the Co-Analyst maintains performance across the cap spectrum—from micro caps to mega caps.

The evaluation below tests the investor-specific areas where Hudson Labs specializes. 

The Hudson Labs Co-Analyst compared to Perplexity Pro

Executive Summary

We evaluate Perplexity Pro across two modalities: Web and Uploads. 

• Web: Perplexity Pro uses web search to provide an answer in this case. Nearly every response contains a critical error—missing information, incorrect results, or hallucination. 
• Uploads: In this test, we provide the earnings release and earnings call transcripts in PDF format. Essentially, the test assumes that the AI application chooses the correct source documents (uploads) at inference time*. Perplexity Pro achieved a 33.3% percent success rate when the correct sources were uploaded. Though, we often needed to query Perplexity Pro a second or third time with more detailed prompts.

By contrast, the Co-Analyst delivered 100% complete and accurate results with no retries or prompt tuning required.

* In most production scenarios, accurate source selection is itself a difficult problem.

Methodology

We compare the Co-Analyst to Perplexity Pro using targeted queries that reflect Hudson Labs’ core strengths: multi-period analysis, guidance extraction, numeric precision, and reliability.

Tests

1. Steve Madden China sourcing exposure

Perplexity Pro - Web 

Perplexity Pro correctly identified figures for only 2 of 8 quarters and hallucinated the rest. The China exposure figures for the first three quarters in the table are from an article about Steve Madden’s Q3 2024 earnings call (citation 1 in the image above). Perplexity Pro got the correct figure for Q4 2024, citing one of our blog posts. We tried this query several times. Results varied slightly depending on which sources were selected, but all web-based responses included fabricated figures for multiple quarters. See the full output here. 

Perplexity Pro - Uploads

We provided PDF transcripts and earnings releases for Steve Madden covering 8 quarters. The first response was unrelated, summarizing the “Key Take-aways from Steve Madden’s 2024 results and 2025 YTD developments”. A second, more explicit query, produced similar results. See full results for both attempts here. 

Hudson Labs Co-Analyst

The Co-Analyst returned precise, cited figures for all quarters and “N/A” where no data existed. See the full Co-Analyst result here. 

2. Apple revenue breakdown by reportable segments

Perplexity Pro - Web

Missed the most recent quarter, Q3 2025 and returned breakdowns by product instead of Apple’s reportable segments. (Apple reports revenue for its segments under the heading "Net sales by reportable segment" in the earnings release.) Several figures were approximate (“~”) and/or incorrect.

On retry with a prompt specifying “report actual figures, no approximations,” Perplexity Pro included the missing quarter but produced a table mixing correct and incorrect values.

‍

Here’s Apple’s Q1 2025 revenue breakdown by product category from the earnings press release.

See full Perplexity Pro results for the two queries here. 

Perplexity Pro - Uploads

Perplexity Pro ignored the query and returned an unrelated answer about the key takeaways from Apple’s earnings calls. Our second attempt also resulted in a similar, unrelated answer. See the full Perplexity outputs here. 

Hudson Labs Co-Analyst

The Co-Analyst provided complete, accurate and precise results. See the Co-Analyst results for revenue by reportable segments for all 8 quarters here. See the results for revenue by product here. 

3. Accenture bookings trends

Perplexity Pro - Web

Did not return data for the most recent quarter (Q3-2025) and only covered 4 quarters in total. On the second attempt, when asked for a breakdown by type of work, returned correct data for 7 of 8 quarters. See full results here. 

Perplexity Pro - Uploads 

First run ignored the query, summarizing “Financial Performance Highlights.” The second run produced a complete and correct bookings table. See full results here. 

Hudson Labs Co-Analyst

The Co-Analyst provides detailed and precise bookings data for all 8 quarters, as seen here.

4. Apple revenue guidance over time

Perplexity Pro - Web 

Perplexity Pro reported actual revenue for the past 7 quarters and an “estimated” revenue figure for Q4 2023. 

With a slightly different prompt, Perplexity Pro provided correct guidance from 2 out of 8 quarters. For most other quarters, Perplexity incorrectly stated that Apple gave no guidance when it did. See the results here. 

Perplexity Pro - Uploads 

Perplexity ignored the query, summarizing key takeaways from Apple’s earnings calls. We tried to extract revenue guidance multiple times, but Perplexity kept providing a summary of the earnings results. See the full results here.

Hudson Labs Co-Analyst

The Co-Analyst results illustrate the platform’s multi-period guidance capabilities. See the results here. 

5.  All soft guidance from the Delta Airline’s earnings call

Perplexity Pro - Web

Perplexity returned quotes from press releases instead of the earnings call and included statements from the prior-quarter call. It also missed multiple significant guidance statements, including:

• "We remain on track for full year remuneration of approximately $8 billion, providing durability to both earnings and cash flow.”
• “So potentially a low single-digit cash taxpayer next year, and then that would march up over a longer-term period, multiple year period to kind of mid-teens is how we see it today.”

See full results here. 

Perplexity Pro - Uploads

Perplexity captured relevant guidance from the transcript but slightly reworded some statements instead of quoting verbatim, e.g., “We expect full year remuneration from American Express of approximately $8 billion”. Once again, Perplexity missed some important forward-looking statements:

• “Our outlook reflects stable demand across both consumer and corporate segments with the midpoint similar to second quarter performance, excluding the impact from lapping the CrowdStrike caused outage.”
• “You're going to see growth for the first time this year for the total year [in MRO revenue]. And this is one that you're going to continue to grow well above the rate of the core airline.”

See full results here.

Co-Analyst

Co-Analyst results were complete and accurate. See the results here.

6. Medtronic revenue breakdown by subsegment

Perplexity Pro - Web

Perplexity Pro provided a complete and correct revenue breakdown for 2 out of 8 quarters. It often omitted the Diabetes subsegment and duplicated or fabricated numbers in other subsegments. See full output here. 

Uploads

With uploads, our first run ignored the query, summarizing key takeaways from Medtronic’s earnings calls. On the second attempt, Perplexity produced a table with multiple duplicated numbers. See full results here.

Co-Analyst

This example demonstrates the Co-Analyst’s numeric focus. In this instance, the Co-Analyst unnecessarily includes segment details co-mingled with sub-segment data. However, all figures are precise and accurate. See the results here.

Conclusion

In investor-focused workflows, Hudson Labs Co-Analyst delivers greater accuracy, completeness, and consistency than Perplexity Pro. While generalist models remain competitive for creative or non-financial tasks, they show persistent weaknesses in numeric fidelity, guidance retrieval, and multi-period analysis. The Co-Analyst is purpose-built to close those gaps—requiring no prompt engineering and delivering reliable results out-of-the-box. Also see comparisons of the Co-Analyst to OpenAI's o3 and ChatGPT-4o.

Why not use FinanceBench?

A full comparison using FinanceBench queries is coming soon. However, FinanceBench does not test the areas that matter most to our users. It emphasizes mathematical accuracy, whereas the Co-Analyst avoids doing math to ensure traceability to source. FinanceBench lacks robust tests for guidance and multi-period retrieval. Like most benchmarks, it is easy to optimize for and rarely reflects real-world workflows.

Our analysis is designed to show how the tools perform in actual use. We encourage you to run your own queries via demo.

‍