Generative artificial intelligence engines running across financial advisory queries overwhelmingly point French savers toward online brokers Linxea and Lucya when constructing low-risk life insurance portfolios, while completely omitting traditional retail banking networks from their top recommendations.
Online Brokers Lead AI Recommendations with Yield Focus
- Online brokers Linxea and Lucya capture the highest final recommendation shares across major AI engines, with respective savings-weighted choices reaching 30.7% and 28.7%.
- Traditional retail banking networks fail to place a single product in the top ten recommendations at any stage of the simulated advisory journey.
- Net yield considerations drive 43.1% of the decision weight in final AI output recommendations, up sharply from 11.1% during initial project scoping.
The Methodology Behind AI Financial Recommendations
Between September 22 and 25, 2026, marketing services agency Redpill partnered with financial publication ToutSurMesFinances to conduct a comprehensive benchmark study evaluating how generative platforms respond to typical saver queries. The assessment tested four major engines—ChatGPT, Google AI Mode, Gemini, and Mistral—across 28 distinct saver profiles categorized by age, income brackets, and housing status. Each simulated persona walked through a standardized three-stage conversation: defining the initial project parameters, searching for candidate contracts, and reaching a final allocation decision incorporating fee structures. Researchers repeated the parcours three times, yielding 336 individual conversational datasets.
When weighting results strictly by demographic population distribution, Lucya emerges as the most frequently cited operator during the initial project and search phases, appearing in 55.3% and 49.1% of conversations respectively. Linxea follows closely behind at 52.3% and 47.2%. However, the final recommendation stage shifts the dynamic. When engines evaluate fee loads alongside net returns, Linxea captures the top rank with an 18.4% final recommendation share. Lucya secures 16%, followed by independent broker Placement-direct at 14.8%, mutual insurer Garance at 13.5%, and digital player BoursoBank at 12.3%.
Divergence Across AI Engines and Weighting Metrics
Individual engine architectures display distinct preferences for specific distribution models. Linxea owes its overall pole position primarily to strong allocations from Gemini and Mistral, whereas Lucya leads the field on ChatGPT and Google AI Mode. Adjusting the demographic data by investable savings capacity alters the final distribution further. Under savings-weighted metrics, Linxea climbs to 30.7% at the final choice stage, and Lucya reaches 28.7%. Meanwhile, participants like Garance slide to 7.2% and BoursoBank drops to 6.2%, losing more than six points of share.
| Distributor / Contract | Initial Project Share | Search Phase Share | Savings-Weighted Final Choice |
|---|---|---|---|
| Linxea (Linxea Spirit 2) | 52.3% | 47.2% | 30.7% |
| Lucya (Lucya Cardif) | 55.3% | 49.1% | 28.7% |
| Garance | — | — | 7.2% |
| BoursoBank | 44.0% | 29.3% | 12.3% |
At the individual product level, Linxea Spirit 2 commands 29.6% of final consideration when weighted by capital, ahead of Lucya Cardif at 22% and Placement-direct Vie at 9%. The preference remains highly engine-dependent: Linxea Spirit 2 captures 71.5% of Mistral recommendations and 62.9% of Gemini outputs, whereas Lucya Cardif takes 38% on ChatGPT, and Lucya CNP reaches 29.4% inside Google AI Mode.
The Exclusion of Traditional Retail Banks
Traditional branch-based banking networks remain entirely absent from the top ten lists across every tested phase. Marco Tinelli, general director of Redpill, points to operational friction as the primary driver behind this digital shutout. According to Redpill’s findings, generative models routinely penalize legacy institutions for restrictive baseline banking conditions, narrower ranges of underlying investment supports, and a distinct lack of digital agility.
Marco Tinelli, general director of Redpill, noted that traditional network banks emerge rarely and do not secure any final choices in the studied scenarios, with AIs opposing them due to banking access conditions, narrower ranges of investment supports, and lesser digital flexibility.
Other digital pioneers experienced significant algorithmic attrition as conversations progressed toward final selection. Fortuneo opened the project phase with an 18.9% presence, slipped to 7.2% during search queries, and fell entirely to zero across all four engines at the final choice stage. Generative models cited restrictive entry hurdles for guaranteed euro funds, narrower asset rosters, and uncompetitive fee structures. Yomoni followed a similar downward slope, dropping from an 8.5% initial footprint to zero final recommendations.
To construct these comparative outputs, the underlying AI models drew heavily from three primary web content categories. Insurer and distributor web properties dominated the search traffic, supplying 2,842 scraped pages for ChatGPT and Mistral, 1,269 direct citations across ChatGPT, Gemini, and Mistral, and 378 structured links displayed via Google AI Mode. Social media channels proved statistically insignificant, accounting for just 143 crawled pages and 17 direct citations, while independent media and alternative websites contributed 1,139 crawled pages and 682 citations.
Disclaimer: The information provided in this article is for educational and informational purposes only and does not constitute financial advice.
