Bonafide - a #PhocusWire Hot 25 Travel Startups for 2026
Summary
Layton Han, CEO of Bonafide.ai, presents the company's solution to what he terms the "AI alignment problem" in the travel industry — the gap between what large language models say about travel brands and what those brands actually offer. Bonafide was recognized as one of PhocusWire's Hot 25 Travel Startups for 2026.
Han opens by defining the alignment problem as the experience of receiving incorrect or outdated answers from LLMs like ChatGPT. Through two years of working with travel brands and suppliers, Bonafide found that approximately one in four LLM responses does not align with a travel brand's factual information — a 25% inaccuracy rate with direct commercial consequences.
The core business argument is that LLMs have become the new sales channel for travel. Han asserts that LLMs' killer application is shopping and commerce, but commerce cannot occur if there is a data disconnect between what the LLM says and what the brand actually offers. If an LLM cannot accurately describe a product or service, the booking opportunity is lost entirely. He illustrates this with a concrete example: Mandarin Oriental Hong Kong offers a premium limousine service, yet Google Gemini was unaware of it and therefore omitted it from relevant responses — costing the hotel potential bookings.
Han explains why the alignment problem arises: travelers query LLMs very differently than they use search engines, asking granular, specific, and personalized questions that demand instantaneous answers. LLMs often lack this level of detail, have outdated information, or simply cannot retrieve it — making it fundamentally a data and content problem rather than a model capability problem.
Bonafide's platform addresses this through a process analogous to how OpenAI and other model builders use alignment techniques to train models to behave as intended. The platform enables travel brands to: 1. Build a comprehensive answer knowledge base covering every aspect of their products and services in granular detail. 2. Continuously test and compare this knowledge base against what LLMs are actually outputting. 3. Identify gaps where misalignment exists between brand information and LLM responses. 4. Reinforce corrective content back to LLMs through the various endpoints where they ingest data. 5. Measure the resulting improvement in booking and revenue opportunity.
Bonafide reports that its platform improves LLM alignment by up to 40%, which Han equates directly to a 40% improvement in booking opportunity for the brands it works with. The pitch positions Bonafide as the dedicated alignment platform for the travel industry, with its differentiation rooted in adapting model-builder alignment methodology for commercial use by travel brands.
Key Takeaways
- •One in four LLM responses about travel brands contains inaccurate or misaligned information — a 25% error rate that directly threatens booking conversion.
- •LLMs are now a primary sales channel for travel; if a brand's product is absent or wrong in an LLM answer, the booking opportunity is lost before the customer ever visits the brand's site.
- •The alignment problem is fundamentally a data and content problem: LLMs lack granular, up-to-date, brand-specific information because brands have not structured and fed it to them.
- •Travel brands must build comprehensive, granular knowledge bases — covering every product and service detail — specifically formatted for LLM consumption, not just SEO or web content.
- •Brands should continuously test what LLMs are actually saying about them, not assume accuracy, and treat LLM response quality as a measurable business metric.
- •When gaps are found between brand truth and LLM output, brands must actively push corrective content to the endpoints where LLMs ingest data.
- •Bonafide's process mirrors the alignment techniques used by OpenAI and other model builders, repurposed for commercial brand management.
- •Improving LLM alignment by 40% translates directly to a 40% improvement in booking opportunity, making alignment a quantifiable revenue lever.
- •The Mandarin Oriental limousine service example shows that even premium, well-known brands are invisible in LLM responses for specific services they actually offer.
- •Travel brands that are not actively managing their LLM presence will cede booking opportunities to competitors who are.
Transcript
Hello. Uh at we at bonafide are solving a AI alignment problem. So what is an AI alignment uh problem? It's something that we have all experienced. It is when chat GPT and other large language models actually give you a wrong answer. And we've been working with travel brands over past two years. And what we have discovered is that about one in four answers that the large language models provide actually does not align with the factual information of the travels brands or suppliers. So why does t...
