ChatGPT has become the default association people make when they hear "AI." Travel agents across the industry are using it — some experimentally, some daily — for everything from drafting emails to researching destinations to generating itineraries.
But ChatGPT is a general-purpose tool. It wasn't designed for travel selling, it doesn't know your products, and it carries risks that many agents don't fully understand. This article provides an honest assessment: what ChatGPT does well for travel agents, where it falls short, the genuine risks, and the purpose-built alternatives that deliver better results.
Where ChatGPT Helps
Email Drafting
ChatGPT is genuinely useful for drafting customer communications. Give it the context — "Write a follow-up email to a couple who enquired about a two-week trip to Bali, focusing on the cultural experiences they mentioned wanting" — and it generates a professional, engaging draft that you can personalise.
The time saving is real. An agent who writes 10-15 follow-up emails daily might save 30-60 minutes by using ChatGPT for first drafts. That time goes back to customer conversations — where the revenue is generated.
Research Assistance
ChatGPT can help with preliminary destination research — visa requirements, seasonal weather patterns, general travel advice, cultural considerations. It provides a reasonable starting point for destinations you're less familiar with.
Itinerary Structure
For complex, multi-destination trips, ChatGPT can generate a logical itinerary framework — suggesting an efficient route through Southeast Asia, for example, with reasonable time allocations and transition points.
Content Ideas
Need social media post ideas for promoting summer holidays? ChatGPT generates creative angles, caption suggestions, and content themes that a social media manager can refine and personalise.
Where ChatGPT Falls Short
Product Accuracy
This is the most critical limitation. ChatGPT doesn't know your actual products. It doesn't know which hotels are in your portfolio, what prices you offer, which rooms are available, or what your specific selling propositions are.
Ask ChatGPT "What's the best family hotel in Tenerife?" and you'll get a reasonable general answer. But it won't know that your company has an exclusive rate at Hotel Costa Adeje, that Room 312 has the best pool view, or that the transfer from South Airport is 20 minutes shorter than the guide books suggest.
An agent who relies on ChatGPT for product recommendations risks suggesting hotels they don't sell, quoting prices they can't offer, and sharing details that don't match reality. This doesn't just lose bookings — it damages credibility.
Hallucination
ChatGPT generates plausible-sounding information that is sometimes factually wrong. It will confidently state that a hotel has a kids' club when it doesn't, that a destination requires a visa when it's visa-free, or that a resort is beachfront when it's a 15-minute walk.
Research from Stanford University documents that large language models produce factual errors in 15-30% of specific claims — a rate that's improving but still far too high for professional travel advice where customers make financial decisions based on agent recommendations.
In travel, inaccurate information has consequences. A customer who arrives at a hotel expecting an all-inclusive package that doesn't exist has a legitimate grievance. An agent who recommended a destination based on incorrect visa information faces serious professional liability.
No Learning About You
ChatGPT doesn't learn your selling style, your product portfolio, your customer base, or your development areas. Every conversation starts from scratch. It can't say "Last time you struggled with price objections on luxury holidays — here's a technique that might help." It doesn't know what you sold last week or what you need to learn next week.
Purpose-built AI coaching platforms build a model of each agent's strengths and weaknesses over time, providing increasingly relevant and specific feedback. ChatGPT provides generic, context-free responses.
No Performance Tracking
ChatGPT can't tell you whether your selling skills are improving. It can't correlate your training activity with your booking performance. It can't identify that your conversion rate drops on luxury enquiries and suggest targeted practice.
Performance analytics that connect learning to selling outcomes require a platform designed for that purpose — not a general-purpose chat tool.
No Roleplay or Practice
ChatGPT can simulate a customer conversation if prompted, but it's not designed for it. It doesn't evaluate your selling technique, identify missed opportunities, provide structured coaching feedback, or track your improvement over time.
AI roleplay platforms are specifically designed for selling practice — with realistic customer personas, defined selling scenarios, coaching rubrics, and performance tracking. The practice experience is fundamentally different from chatting with ChatGPT.
The Genuine Risks
Confidentiality
When you type customer information into ChatGPT — names, travel dates, budgets, personal preferences — that data is processed by OpenAI's systems. Depending on your settings and your company's data policies, this may violate GDPR, breach customer confidentiality, or conflict with your company's data handling requirements.
ICO guidance on AI and data protection is clear: organisations must ensure that personal data processed by AI systems complies with data protection principles. Using free consumer AI tools for processing customer data is a compliance risk that many travel businesses haven't assessed.
Professional Liability
If you provide incorrect travel information based on ChatGPT output — wrong visa requirements, inaccurate health advice, incorrect insurance coverage details — the professional liability falls on you, not on ChatGPT.
ABTA members operate under a code of conduct that requires accurate information provision. Sourcing that information from an AI tool known to produce errors doesn't absolve the agent of responsibility.
Brand Consistency
ChatGPT generates content in its own voice, not yours. Customer emails, social media posts, and marketing materials generated by ChatGPT sound like ChatGPT — and they sound like everyone else's ChatGPT-generated content. In a market where differentiation matters, generic AI voice undermines brand identity.
Over-Reliance
The convenience of ChatGPT can lead to over-reliance. An agent who uses ChatGPT to answer every question stops building their own product knowledge. When the technology is unavailable (connectivity issues, API changes, company policy changes), they're less capable than before.
The goal of AI should be to augment human expertise, not to replace the need for it. AI training platforms build the agent's knowledge; ChatGPT bypasses it.
Purpose-Built Alternatives
For Training and Knowledge Building
ChatGPT: Can answer questions about destinations and products (with accuracy risks). Doesn't track what you know, personalise learning, or verify knowledge.
AI training platform: Creates personalised learning pathways based on your assessed knowledge gaps. Tracks your progress. Verifies your knowledge through adaptive assessments. Provides content specific to your company's product portfolio. Completion rates of 85-92% versus ad hoc ChatGPT usage.
For Sales Practice
ChatGPT: Can role-play a customer if prompted. Doesn't evaluate technique, provide coaching feedback, or track improvement.
AI roleplay platform: Provides realistic customer personas with specific objections and buying signals. Evaluates your selling technique across defined criteria. Provides AI coaching feedback with specific improvement recommendations. Tracks your performance over time.
For Sales Coaching
ChatGPT: Can provide generic selling advice. Doesn't know your specific performance data, your customers, or your products.
AI coaching platform: Analyses your selling conversations and provides specific feedback tied to your performance data. Identifies patterns in your approach that correlate with higher or lower conversion. Suggests personalised improvement actions.
For Content Creation
ChatGPT: Generates generic travel content. Doesn't know your products, brand voice, or sales data.
Purpose-built content tools: Generate content from your actual product data, in your brand voice, optimised for your specific channels and customer segments.
The Comparison Table
| Capability | ChatGPT | Purpose-Built AI Platform |
|---|---|---|
| Product-specific knowledge | Generic internet knowledge | Your actual product portfolio |
| Accuracy | 70-85% (with hallucination risk) | 95%+ (trained on verified data) |
| Personalised learning | None | Adaptive pathways based on assessed gaps |
| Sales coaching | Generic advice | Specific feedback from your performance data |
| Roleplay practice | Basic simulation | Structured scenarios with coaching rubrics |
| Performance tracking | None | Analytics connecting training to sales |
| Data privacy | Risk — data processed externally | Designed for compliance |
| Cost | Free-£20/month (consumer) | £5-£25/agent/month (business) |
| Travel specificity | None | Purpose-built for travel |
A Pragmatic Approach
The most effective approach isn't "ChatGPT versus purpose-built AI" — it's understanding what each tool does well and using the right tool for the right job.
Use ChatGPT for:
- Drafting customer emails (always review before sending)
- Brainstorming content ideas
- General research (always verify accuracy)
- Formatting and structuring text
Use purpose-built AI platforms for:
- Product knowledge training (accuracy and personalisation matter)
- Sales practice and roleplay (structured coaching matters)
- Performance coaching (specific, data-driven feedback matters)
- Customer-facing content where accuracy is essential (professional liability)
- Any task involving customer data (compliance requirement)
Never use ChatGPT for:
- Visa, health, or safety information (verify with official sources only)
- Specific product details (availability, pricing, features) — use your booking system
- Financial or insurance advice — use approved sources
- Any communication where inaccuracy creates liability
The agents who get the most value from AI are those who understand these boundaries — using the right tool for each task, verifying outputs where accuracy matters, and building their own expertise rather than outsourcing it entirely.
Explore purpose-built AI for travel agents with TravAI →
This article is part of our AI in Travel & Tourism series. Related reading:
