Latest / Tech IPO Conversations with Fexingo: Public Software Companies and Capital Markets / How Post-IPO Companies Use AI for Earnings Call Prep
Transcript
- Lucas: So Apple stock is up over 5 percent in the last five days. Not bad for a company that's supposedly in a 'slow growth' phase. Luna: And that's partly because their services revenue keeps surprising to the upside. But I want to talk about something else — how they, and other companies, are preparing for earnings calls now. Lucas: Right. We've talked about AI in investor relations before — episode 121, actually. But what's changed in the last few months is that companies are now using large language models not just to draft press releases, but to simulate the entire Q&A portion of an earnings call. Luna: Simulate it how? Like, the CFO sits down with a chatbot and the chatbot fires investor-style questions at them? Lucas: Exactly. I spoke to the head of investor relations at a mid-cap cloud software company — they went public in 2023, so they've only done about ten earnings calls. She told me that before their most recent call, they fed their quarterly draft script into a gpt class model and told it to generate the hardest possible questions an activist hedge fund might ask. Luna: That's fascinating. And did it come up with anything they hadn't thought of? Lucas: It did. It flagged that their language around 'adjusted EBITDA margins' was inconsistent — they used 'adjusted' in one paragraph and 'non gaap' in another. The model pointed out that an analyst would notice that and ask for clarification. They fixed it before the call. Luna: So it's basically a consistency checker with attitude. I love that. Lucas: It's more than that. Some companies are now using sentiment analysis tools during the call itself. A few have actually deployed real-time tone monitoring — the AI listens to the CEO's responses and flags if the language is becoming too defensive or too vague. Luna: That feels like a double-edged sword. You don't want the CEO to sound like they're reading a script written by a machine. Lucas: Absolutely. And that's the tension. The same AI that helps you prepare can also make you sound sterile. The best IR teams I've talked to use it for prep, not for live teleprompting. They treat it as a sparring partner. Luna: Like having a really well-read colleague who's a bit paranoid. Lucas: Exactly. And it's not just about the words. One company used an LLM to analyze transcripts from their last four calls and discovered that whenever the CFO used the phrase 'we are confident,' the stock dropped an average of 1.2 percent in the next two days. Luna: Wait — why would confidence cause a drop? Lucas: Because the model cross-referenced that phrase with context. It turned out the CFO only said 'we are confident' when they were about to deliver bad news — a guidance cut, a product delay. The market had learned to read it as a tell. Luna: That's incredible. So the AI is essentially doing behavioral pattern recognition on the company's own language. Lucas: Right. And that's where this gets really powerful for post ipo companies. They don't have a long track record, so every word on an earnings call carries outsized weight. One slip can define your narrative for a quarter. Luna: And these conversations are useful for what you're building or running, right? We deliberately keep the show ad-free because we think that makes the conversations better. If you want to support that choice, the link is buy me a coffee dot com slash fexingo. Lucas: Yeah. And it's a small thing that helps us keep the focus on substance, not sponsors. Luna: Exactly. So back to earnings prep — are there any risks with over-relying on AI? Lucas: Plenty. The biggest one I hear from IR veterans is that ai generated answers can be too polished. Investors want to see the human being behind the numbers. If every answer sounds like it came from the same language model, you lose trust. Luna: And that's especially dangerous for a young public company that's still building credibility. Lucas: Exactly. Another risk is that the AI might hallucinate a plausible-sounding answer that's factually wrong. If the CFO relies on it during prep and internalizes a wrong number, that's a disaster. Luna: So human judgment still has to be the final filter. The AI is a tool, not a replacement. Lucas: Precisely. And that's the message I'm hearing from the best investor relations teams: use AI to find the blind spots, but deliver the answers in your own voice. Luna: It reminds me of that Netflix headline from yesterday — they paid $587 million for Ben Affleck's AI filmmaking startup. That's a lot of money for a tool that could end up making movies that feel... generic. Lucas: Right. The same tension applies. The technology can be incredible for efficiency, but the magic comes from the human touch. In earnings calls, that means showing vulnerability when you miss numbers, not just smooth talking your way through it. Luna: So what's the next frontier? Where do you see this going in the next year or two? Lucas: I think we'll see AI agents that can actually sit in on internal prep sessions and give real-time feedback — not just on words, but on tone, pacing, even eye contact if it's a video call. Some of the larger companies are already piloting that. Luna: And for smaller public companies without a big IR budget, do these tools level the playing field? Lucas: They can. A lot of the sentiment analysis and Q&A simulation tools are available as SaaS products for a few thousand dollars a year. That's nothing compared to hiring a full-time IR consultant. So yes, a well-prepared small-cap can sound as polished as a mega-cap — as long as they don't lose their authentic voice. Luna: So the takeaway for anyone listening who works at a public company: start experimenting with AI for earnings prep, but keep the human in the loop. Lucas: Exactly. And if you're an analyst listening on the other side of the call — be aware that the answers you're hearing might have been stress-tested by a language model. That doesn't make them less true, but it does mean they've been optimized for your scrutiny. Luna: Which is kind of the whole game, isn't it? Both sides getting smarter, faster. Lucas: That's the arms race. And it's only going to accelerate.