The core message of the session is that financial communicators can no longer assume their equity narrative reaches investors in the form the company intended. Bartfer Hulst argued that AI-enabled transcription, summarisation, scoring and investor tools now sit between corporate disclosures and investor decisions, creating a need for more controlled, machine-readable, and consistent communication. Search IR frames this practical response as “synthetic presenters”: digital presenters designed to deliver the approved equity story accurately, consistently, and without improvisation that could create interpretation risk.

The speakers first introduced their perspectives. Martijn Mosse described himself as head of investor relations and treasury at a Dutch listed real estate small cap, emphasising that the company operates in a relatively non-innovative industry with limited resources. He also noted that, as a Gen Z investor relations officer, his media habits and short attention span shaped how he thinks about communication. Bart then positioned his background in corporate communications and digitisation, and explained the concept of “digital share reputation”: maintaining the reputation of a listed company’s share and supporting a level playing field for investors.

Bart’s main argument was that the equity narrative is increasingly “hijacked” once earnings materials enter the AI ecosystem. Earnings calls are captured by uninvited transcription bots using different speech-recognition systems and LLM cleanup layers. Those transcripts are then published, cached, fed into portals and AI analysis layers, and used by professional and retail investors. He cited examples including Seeking Alpha publishing within six hours, live transcription through Quartr, and a Dutch case where “13.7 million EBITDA” became “13,700,000 EBITDA,” creating a €70 million difference. He also said 61% of AI answers cite the earnings call first, four in ten institutional investors trust AI summaries as much as sell-side research, and 46% may skip earnings calls and wait for transcripts or summaries.

The session then moved from transcription risk to language and signal risk. Bart warned that certain words, such as “dismissed,” “inquiry,” “alleged,” “closure,” and “misstatement,” can trigger negative AI scoring, while tone of voice and facial expressions may also be analysed for trading signals. He argued that AI now makes differences between brand narrative, equity narrative, and quarter-to-quarter disclosures measurable. The proposed solution was a presenter that sticks to the approved story, avoids negative trigger wording, uses clearer articulation for better transcription accuracy, and replaces vague qualitative language with more specific phrasing.