How Analysts and Funds Are Using AI to Read Your Disclosures

Most IR teams spend weeks refining their earnings release. Every word is reviewed; every number is double-checked. But bear in mind, within seconds or minutes of your call ending, AI-powered platforms may have already analyzed your tone, flagged inconsistencies, and benchmarked your language against every peer in your sector. For big liquid global stocks, this is not a future scenario. It is happening now.

What Tools Are Investors Actually Using?

Institutional investors increasingly rely on platforms like AlphaSense, used by over 6,500 organizations, to process company disclosures at scale. These tools use natural language processing (NLP) to read your filings, earnings transcripts, and press releases the way a human analyst would, but instantly and across hundreds of companies simultaneously.

It is not just about having the tool; investors have built it into their daily workflow in specific ways:

• Screening before reading: Before an analyst opens your earnings release, the platform has already scored it. Sentiment, tone, and information density are ranked against peers. Low-scoring disclosures get less time and attention.
• Tracking management credibility over time: AI platforms keep a record of everything your management team has said across every quarter. If your CEO said demand was “strong” last quarter and calls it “mixed” this quarter without explanation, that inconsistency is surfaced automatically and weighed against prior guidance.
• Comparing you to your peers instantly: Funds running coverage across dozens of companies use AI to benchmark disclosures side by side. If your competitors are providing more granular data, the gap is visible within seconds and influences how investors allocate their research time.
• Generating summaries for investment committees: Senior decision-makers at funds often never read the full transcript. They read an AI-generated summary. What gets highlighted in that summary, and what tone it carries, is shaped entirely by how your disclosure was written.

What is the AI looking for?

• Tone and sentiment shifts: If your CFO uses more hedging language this quarter than last (“we remain cautious,” “subject to conditions”), the platform scores it as a sentiment decline before any analyst has finished reading the document.
• Consistency across documents: If your press release says demand is “robust”, but your earnings script calls it “steady”, the contradiction gets flagged automatically. Investors interpret misaligned language as a signal that management may not be on the same page.
• Vagueness relative to peers: AI benchmarks your language against competitors. If they are giving specific metrics and you are using general language, your disclosure scores are less informative and get less attention from institutional investors.

Three Things to Do

1. Audit your tone quarter over quarter: Before publishing, read your last two releases alongside the current draft, and as long as AI is secure, and subject to compliance approval, put this through your own AI model for comments. If the language is more cautious, name the reason explicitly.
2. Align all documents before they go out: Your press release, earnings script, and presentation slides should tell the same story with consistent language. Subject to compliance, AI can help with this.
3. Replace vague language with specific data where possible. Phrases like “solid growth” or “challenging conditions” are exactly what NLP tools flag as low information. Where you can say “revenue grew 8% driven by X,” do it.

Good IR has always required clear, consistent, data-supported communication. AI does not change that principle; it accelerates the consequences of ignoring it. A sentiment decline flagged by an algorithm at 7 am can shape an analyst’s tone on a call at 10 am.

At Miranda Investor Relations, we help companies craft IR communications that are clear, consistent, compliant, and built for today’s investment landscape.