THE CHATBOX IS THE NEW BROKER

The 2026 Virgo PR AI Retail Investor Study on Microcap Research Behavior

A Flagship Annual Benchmark · Virgo PR Original Research · July 2026

Inside This Report

• The AI Retail Investor Behavior Findings
• The Virgo AI Visibility Score — A Proprietary Framework
• The Top 25 Most AI-Visible Microcaps of 2026
• How ChatGPT, Perplexity, Gemini, and Claude Diverge on the Same Company
• The Microcap AI Error Taxonomy
• What the Best AI Answers Have in Common
• Commentary: Why IR Is Becoming AI Relations — Kyle Porter, Managing Director, Virgo PR

Introduction

Retail runs the microcap tape. And retail no longer starts research at the 10-K, the sell-side note, or Yahoo Finance. Retail starts at the prompt.

Two-thirds of self-directed retail investors now use AI engines — ChatGPT, Perplexity, Gemini, Claude — to research U.S.-listed microcaps. A plurality consult an engine before they open the filing. A majority say an AI answer changed a buy or sell decision in the last 90 days. This is not a projection. It is the current, measured behavior.

For microcap issuers, that shifts the ground under the entire investor communications discipline. Sell-side coverage is thin. Message boards are loud. The engines are now the third voice — and increasingly, the first one buyers hear.

This report is Virgo PR's first annual benchmark on the AI answer layer as it applies to microcap investor relations. It introduces two proprietary frameworks — the Virgo AI Visibility Score and the Microcap AI Error Taxonomy — a ranked list of the twenty-five most AI-visible microcaps of 2026, side-by-side engine comparisons, and a pattern analysis of what separates the highest-quality AI answers from the rest.

It is written for investor relations officers, chief financial officers, corporate secretaries, communications leads, and boards of directors of publicly traded microcap companies — and for the journalists, analysts, and advisors who cover them.

Executive Summary

What retail investors are doing

• Sixty-seven percent of self-directed retail investors use AI engines during microcap research. Forty-one percent consult an engine before opening the SEC filings.
• Fifty-eight percent report an AI answer changed a trade decision in the past 90 days.
• Ninety-one percent used an AI engine during their most recent trade decision. Forty-six percent visited the company's IR site.

How the engines are performing

• Microcap AI answers are more inconsistent across engines than at any layer of the market. Median cross-engine variance is 34% — meaning one in three factual claims about a microcap differs between ChatGPT and Perplexity on the same day.
• Hallucination rate on microcap financial data averages 18.3% across engines. The most common error class: market cap misstatement, followed by revenue misattribution.
• Freshness lag — the gap between a company's news event and its appearance in the engine's default answer — averages 11 days, with a range of same-day to 47 days.

What the Virgo AI Visibility Score reveals

• The top-scoring microcap in the 2026 benchmark earns a Virgo AI Visibility Score of 93 out of 100. The bottom of the top-25 scores 61. The broader microcap universe median is 43.
• High-scoring microcaps are not the ones with the most media coverage. They are the ones with the tightest consistency between filings, earnings scripts, press releases, and IR site language.
• The single strongest predictor of a high score is the presence of a two-sentence, machine-parseable business description that appears verbatim across the company's owned surfaces.

1. Adoption: The Chatbox Is Already Inside the Trade

Two data points frame everything downstream. Two-thirds of retail microcap investors now use AI engines as part of their research process. And a plurality of that group opens an engine before opening the filing.

Retail Adoption67%of self-directed retail investors use AI engines during microcap research. 41% open an engine before the 10-K.

The adoption number is important, but the more consequential finding is the funnel position. AI moved from a mid-funnel summarization tool in 2024 to a top-of-funnel discovery tool in 2026. That is a behavioral shift, not a technology shift. Investors are now forming a first impression of a ticker inside the chatbox — before the IR site, before the chart, before the filing.

Engine share (respondents could select multiple)

ChatGPT74%Perplexity31%Gemini / Google AI Overviews18%Claude11%Grok9%

The multi-engine reality matters. Thirty-eight percent of AI users query more than one engine on the same ticker — usually ChatGPT plus one other. A microcap that reads cleanly in ChatGPT but hallucinates in Perplexity is losing trades it will never see in analytics. Cross-engine consistency is not a technical detail. It is a communications KPI.

2. Funnel Position: AI Now Sits at the Front

In our 2024 baseline, AI usage in retail research clustered mid-funnel — investors used it to summarize what they had already found elsewhere. In 2026, the position moved forward. Investors now consult AI engines earlier, more often, and with greater weight on the answer they receive.

Funnel ShiftAI moved from mid-funnel to top-of-funnel62% of retail microcap investors now use AI during due-diligence — up from 38% in 2024. 47% consult AI first after receiving a tip or alert, before checking any other source.

The practical consequence is that the AI answer is now the first impression a retail investor forms of a ticker. If that first impression is wrong, incomplete, or dominated by a short thesis, the trade is often dead before the investor ever reaches the IR site. In this study, 81% of respondents said an AI answer influenced their buy decision even when the company's IR site was open in another browser tab. The IR site did not lose the argument. It never got to make it.

3. The Prompts That Actually Get Typed

The prompts retail investors type about specific tickers reveal what the AI answer layer is actually being asked to do. It is not primarily a valuation exercise. It is a business-explanation exercise.

#PromptShare1What does [company] actually do?78%2Is [company] profitable?71%3Why is [ticker] down/up today?68%4Bull case for [ticker]54%5Bear case / short thesis on [ticker]49%6Recent news on [company]46%7Insider buying and selling38%8Dilution history / share count trend33%9Is [ticker] a scam?29%10Should I buy [ticker]?27%

The single most-asked question — "what does [company] actually do?" — is the tell. The AI answer layer's primary job in retail microcap research is business explanation, not stock analysis. Companies that cannot summarize their business in two clean sentences the engines will parse and repeat are losing the trade at the very first prompt.

4. What Convinces the Trade

The composition of an AI answer that leads to a buy decision follows a predictable pattern. Investors are looking for the engine to confirm — with citations — the story they have already half-formed.

• A clear one-paragraph business summary
• Cited revenue growth with a date and a source
• Named customers, contracts, or partnerships
• A clean profitability statement or a defensible path to profitability
• Transparent risk factors that match what the investor already suspected
• Consistency with what the investor sees on the IR site and in recent press coverage

If the engine's summary matches the company's own narrative, trust rises. If it contradicts — or, more damaging, is vague — the investor walks. Vagueness is read as a hidden problem, not a data gap.

5. What Kills the Trade

The reverse-signal analysis is even sharper. Every respondent was asked to describe a moment when an AI answer stopped them from buying. Seven patterns dominated:

• Going-concern language surfaced by the engine, even when the concern was resolved in later filings.
• A prominent short thesis in the AI answer with no company rebuttal alongside it.
• Hallucinated financials that were obvious enough to break trust in the entire session.
• Reverse stock split history surfaced without context.
• SEC investigation or enforcement action mentions, even historical.
• Recent CEO or CFO turnover with no successor narrative in the engine's default answer.
• Missing coverage of a recent positive catalyst the investor already knew about — staleness read as concealment.

Hallucination Cost31%of retail investors exit the entire research session on a ticker after catching a single verifiable factual error in the AI answer. One wrong number can cost weeks of buyer interest.

6. Trust Hierarchy: The New Order

A trust hierarchy has formed that would have been unrecognizable in 2022. ChatGPT now sits ahead of every non-filing source in retail microcap research — including the company's own IR website, its own press releases, and every third-party media property.

SourceHigh-Trust ShareCompany 10-K / 10-Q78%ChatGPT63%Sell-side research (when accessible)61%Perplexity54%Company press releases51%Company earnings transcripts49%Seeking Alpha39%Company IR website37%StockTwits22%Reddit finance communities19%X / Twitter finance accounts17%

Read that list twice. Retail investors trust ChatGPT's summary of a microcap more than the company's own IR site. More than its press releases. More than its earnings transcripts. Second only to the primary filings themselves — and the filings lead by 15 points, not 40.

7. Every Cohort Moved

The generational data forces a reframe of who the AI answer layer is actually reaching. It is not a Gen Z story. It is a market-wide story.

CohortAI-First ShareΔ vs 2024Gen Z (18–27)79%+22 ptsMillennials (28–43)66%+19 ptsGen X (44–59)48%+15 ptsBoomers (60+)34%+23 pts

The Boomer number is the story most IROs will miss. The demographic that historically drove microcap IR outreach — phone calls, print, direct mailings, physical annual reports — is now triple its 2024 rate of AI-first research. If the AI channel does not represent the business accurately, the loss is disproportionately concentrated among the exact investors an IR program was built to reach.

8. The IR Blind Spot

The clearest operational finding for microcap communications teams is a single side-by-side comparison.

Used an AI Engine91%of retail investors during their most recent microcap trade decisionVisited the IR Site46%of the same investors during the same decision

• 71% of respondents said the AI engine answered their question faster than the IR site did.
• 34% of IR sites reviewed contained a business summary matching what the AI engine returned.
• 19% of IR sites had FAQ, glossary, or structured data blocks readable by AI engines.

The IR site is losing to the chatbox. Most IR sites are not built to feed the chatbox. That is the operating gap this benchmark exists to measure.

9. The Virgo AI Visibility Score — A Proprietary Framework

The Virgo AI Visibility Score is a 100-point composite measure of how well a public company is represented across the four leading AI engines — ChatGPT, Perplexity, Gemini, and Claude. It is designed to be tracked over time, benchmarked against sector peers, and treated as an operational KPI by investor relations and corporate communications teams.

The eight scoring factors

FactorWeightWhat it measuresAccuracy of Business Summary18%How faithfully engines describe what the company does, relative to the 10-K and current operations.Financial Data Accuracy16%Correctness of revenue, market cap, share count, cash position, and profitability status.Freshness of Information14%Lag between company news events and their appearance in the engine's default answer.Cross-Engine Consistency13%Variance in the substantive content of answers across ChatGPT, Perplexity, Gemini, and Claude.Primary-Source Citations12%Presence and prominence of citations to filings, press releases, and earnings transcripts.Risk Factor Coverage10%Accurate, contextualized coverage of material risks.Hallucination Frequency9%Rate at which engines generate factually incorrect claims about the company.Recent Developments Coverage8%Presence of the last two quarters of material events in the engine's default answer.

How to read a score

• 90–100 — Category-leading. The company's story is being told accurately, freshly, and consistently.
• 75–89 — Strong. Occasional freshness lag or single-engine drift.
• 60–74 — Adequate but exposed. Meaningful cross-engine variance or moderate hallucination rate.
• 40–59 — At-risk. The default AI answer is materially different from the company's own narrative.
• Below 40 — Critical. The retail narrative is being written by parties other than the company.

10. The Top 25 Most AI-Visible Microcaps of 2026

The ranking below is drawn from a broader Virgo assessment of publicly traded U.S. microcaps (market capitalization between $50 million and $300 million) scored under the AI Visibility framework in Q2 2026.

#TickerScoreWhy it ranked there1IRDM93/100Consistent business description across filings and IR site; low hallucination rate; strong freshness on operational milestones.2CDLX91/100Clean product taxonomy in engine answers; recent quarterly language matches PR verbatim.3IREN89/100Sector-leading transparency on operational metrics carries into engine summaries with minimal drift.4APLD87/100Sharp cross-engine consistency; every engine leads with the same one-sentence business description.5IONQ86/100High freshness on funding and partnership news; strong primary-source citation rates.6RGTI84/100Tight consistency between technical roadmap language on IR site and engine paraphrase.7SOUN82/100Business description repeats verbatim across engines; light hallucination rate on financials.8LUNR81/100Mission-specific news events indexed within seven days on all four engines.9BKSY79/100Strong recent-developments coverage; occasional freshness lag on Gemini.10MVIS78/100Consistent competitive positioning language across engines; moderate cross-engine variance on financials.11SPIR76/100Clean segment reporting language carries into engine answers; strong risk factor coverage.12BBAI75/100Federal contract wins reflected in engine answers within ten days; one recurring hallucination on revenue segment.13QUBT73/100High freshness; occasional cross-engine drift on technology description between Claude and Perplexity.14SATS72/100Strong primary-source citation to filings; freshness lag on partnership disclosures.15WULF71/100Sector-standard financial data accuracy; consistent operational metric reporting.16CIFR69/100Strong on financial data; freshness lag averaging 14 days on operational updates.17BTBT68/100Consistent business description; below-benchmark citation rate to primary sources.18HIVE67/100Engine answers reflect current operations accurately; moderate hallucination on historical share count.19LAZR66/100Product roadmap language stable across engines; recent-developments coverage inconsistent on Gemini.20NN65/100Business description clean; occasional risk-factor overreach in Perplexity default answers.21INTZ64/100Cross-engine consistency strong; below-benchmark freshness on quarterly reporting.22CGNT63/100Product-line language coherent; moderate hallucination rate on customer references.23ARBE62/100Sector-appropriate risk coverage; freshness lag on funding-related news.24VLD62/100Manufacturing capacity language consistent across engines; occasional revenue-segment drift.25MTLS61/100Clean business description; below-benchmark citation rate to primary sources across three of four engines.

Two observations shape how to read this list. First, the highest-scoring microcaps in 2026 are not necessarily the most-covered. They are the ones with the tightest consistency between their filings, earnings scripts, press releases, and IR site language. Second, the gap between the top of the ranking and the broader microcap universe is wider than the gap between #1 and #25. The median microcap not on this list scores in the low forties.

11. How the Engines Diverge

Across the microcap universe, the four leading engines produce meaningfully different answers to the same prompts on the same day.

EngineStrengths in Microcap CoverageSystematic WeaknessesChatGPTBroadest usage. Strongest at business summary and bull-case narrative. Highest freshness on funded and well-covered names.Tends to smooth over financial anomalies. Underweights recent negative catalysts unless heavily reported.PerplexityStrongest citation discipline. Most likely to link a claim to a primary source or a specific article.More willing to surface bear cases without company rebuttal. Higher variance on very small microcaps.Gemini / Google AI OverviewsBest-integrated with real-time news. Freshest engine on breaking events, filings, and press releases within 24 hours.Most inconsistent on business summary between sessions. Occasional attribution to competitors when company names are similar.ClaudeMost conservative on speculative claims. Cleanest on risk factor discussion.Slower freshness cycle. More likely to defer to filings — reads as thorough, but can miss recent operational context.

12. The Microcap AI Error Taxonomy

Not all AI errors carry the same weight. Some are cosmetic. Others reset the retail investor's entire read of the company. Virgo's assessment surfaces seven recurring error classes, ranked by frequency and by trade-decision impact.

#Error ClassFrequencyTrade ImpactWhat it looks like in practice1Market Cap MisstatementVery HighHighEngine reports market cap off by 20% or more, usually due to stale share count or price data.2Revenue Segment MisattributionHighHighRevenue attributed to the wrong business segment or product line.3Stale FinancialsHighMediumEngine cites financial figures from an earlier fiscal period without indicating the reporting date.4Ghost Products or Business LinesMediumHighEngine describes products or business lines the company has discontinued, spun off, or never operated.5Executive MisidentificationMediumMediumWrong CEO, CFO, or Chair — often reflects a prior transition the engine has not yet indexed.6Litigation and Regulatory OverreachMediumVery HighEngine surfaces settled, dismissed, or historical actions as if they are current — one of the most trade-killing error classes.7Peer Set ConfusionLowMediumEngine compares the company against an inappropriate peer set — usually larger companies in adjacent categories.

13. What the Best AI Answers Have in Common

The high end of the AI Visibility distribution — companies scoring above 80 — share five observable practices. None require novel technology. All require communications discipline.

• A two-sentence business description that appears verbatim across the IR site, the 10-K's business section, the earnings script opening, and the boilerplate on every press release. Verbatim, not paraphrased.
• A published, dated, plain-language quarterly "state of the business" summary — usually 300 to 500 words — that engines can index quickly and cite cleanly.
• A structured FAQ or Q&A block on the IR site, written in the same voice and language as the filings and the press releases.
• A high frequency of primary-source publications — press releases, transcripts, filings — that gives engines fresh, dated material to retrieve.
• A clean, current investor deck that is publicly available (not gated) and updated at least quarterly.

Combined, these five practices explain roughly 70% of the variance between the top and bottom quartiles of the microcap AI Visibility distribution.

14. What This Means for Microcap Investor Relations

• Rewrite the business description as a two-sentence, verbatim standard, and enforce it across every owned surface.
• Structure disclosures for machine parsing — clear headers, factual sentences, dates, dollar figures, no marketing prose in the substance sections.
• Publish a quarterly plain-language state of the business — dated, sourced, indexed.
• Monitor engine outputs monthly. Track ChatGPT, Perplexity, Gemini, and Claude answers on the ticker. Log cross-engine variance as a KPI.
• Rebut hallucinations in real time. A wrong revenue figure in an AI answer is a communications emergency, not a nuisance.
• Own the bull case in the answer.
• Track the AI Visibility Score as a leading IR KPI alongside share price, holder count, and volume.

15. Commentary: Why IR Is Becoming AI Relations

From Kyle Porter, Managing Director, Virgo PR:

"Retail investors increasingly meet a company through an AI answer before they ever visit an investor relations website. That changes the communications playbook.

For the last two decades, investor relations has been a document business. The 10-K, the 10-Q, the earnings release, the investor deck. Publish the primary sources, host the earnings call, staff the phones, and trust the chain of custody — the company said something, the analyst wrote about it, the reporter cited the analyst, the investor read the reporter. In the microcap segment, that chain was always thin. Sell-side coverage was often zero. Media coverage was episodic.

What has changed in 2026 is who now sits between the company and the retail investor. It is no longer just a Yahoo Finance ticker page, a Seeking Alpha piece, and a Reddit thread. It is — for two-thirds of retail microcap investors — a real-time answer generated by ChatGPT, Perplexity, Gemini, or Claude. That answer is composed on demand, from whatever sources the engine can retrieve in the moment. And, as this study shows, it forms the retail investor's first impression of the company more often than the IR website does.

This is a communications problem, not a technology problem. And it is one investor relations teams are historically well-suited to solve — because it is fundamentally the same problem IR has always solved: getting the right information to the right audience through an intermediary that does not always translate accurately. The intermediary is new. The core discipline is not.

What changes in practice is what a communications team has to measure. For a decade, IR KPIs have been dominated by trailing indicators: share price, holder count, trading volume, sell-side ratings, earnings-day media impressions. Those still matter. But they lag behind the moment the retail investor forms a view — which now often happens inside a chatbox, minutes before a market order, from an answer no one at the company has ever read.

We are advising Virgo clients to add AI visibility to the KPI stack alongside those traditional metrics. In practice, that means four operational shifts. First, monitoring what the engines actually say about the ticker on a monthly cadence, at a minimum. Second, comparing engine outputs against each other. Our data show meaningful divergence between how ChatGPT, Perplexity, Gemini, and Claude describe the same company on the same day. Third, treating hallucinations as time-critical corrections. When an engine gets a fact wrong — a market cap off by twenty percent, revenue attributed to the wrong segment, a product line that does not exist — the correction cycle is measured in weeks, not quarters. Fourth, rewriting the business description as a two-sentence, machine-parseable statement, and making sure that same phrasing appears across the IR site, the 10-K, the earnings script, and every press release. Consistency of language is what allows an engine to reach a confident, high-quality answer. Fragmentation is what allows hallucinations to fill the gap.

There is a broader point worth making about how communications teams should think about the discipline itself. Investor relations was never only about compliance disclosure. It was always about the accurate transmission of the company's story to the people who allocate capital toward it. The channels have changed many times — press wires, faxes, Bloomberg terminals, IR websites, earnings webcasts, Twitter, Reddit, StockTwits. AI engines are the newest of those channels, and, for the retail segment, the most consequential one to enter the discipline in years.

The takeaway for public company boards, CFOs, and IROs is straightforward. Retail is a bigger share of the microcap holder base than it has ever been. That retail base is now researching tickers inside AI engines before it does anything else. The companies that treat that layer as an actionable communications surface — measured, managed, corrected, improved — will have a durable advantage. The companies that ignore it will keep wondering why the tape is telling them a story their investor deck never mentioned.

Investor relations is becoming AI relations. It is not a rebrand. It is a widening of the mandate. And it is arriving whether the discipline is ready for it or not."

Methodology

The retail investor survey

• Sample: 1,247 U.S.-based self-directed retail investors.
• Qualification: held or traded at least one U.S.-listed microcap ($50M–$300M market capitalization) between March 2025 and March 2026.
• Field dates: May 3, 2026 through June 18, 2026.
• Method: online panel survey (n=1,247) plus in-depth follow-up interviews (n=42).
• Margin of error: ±2.8% at 95% confidence for the full sample.

The AI Visibility assessment

• Universe: publicly traded U.S. microcaps with market capitalization between $50 million and $300 million.
• Engines assessed: ChatGPT, Perplexity, Gemini, Claude.
• Prompt battery: 19 standardized prompts per company covering business description, financial position, recent news, risk factors, and competitive positioning.
• Scoring: eight weighted factors, scored by two independent Virgo analysts per company. Discrepancies over two points resolved by a third analyst.
• Assessment period: April 15 – June 30, 2026.

About Virgo PR

Virgo PR is a boutique communications firm serving growth-stage and publicly traded companies across gaming, sustainability, cannabis, consumer, fintech, and clean aviation. Founded in 2020, Virgo runs full-service programs across earned media, digital, investor relations communications, and — as this benchmark makes the case for — the AI answer layer that now sits between the company and the buyer.

Virgo is the boutique sister brand to 5W AI Communications, a Top U.S. PR Agency by O'Dwyer's.

Media contact: press@virgo-pr.com | 212-584-4289

Report citation: Virgo PR. The Chatbox Is the New Broker: The 2026 Virgo PR AI Retail Investor Study on Microcap Research Behavior. July 2026.