BCD just shipped agentic AI to TripSource. 4 questions to add to your next TMC RFP.

The TMC AI race went live last week. Don't go to RFP unarmed.

Egor Travel Code
June 16, 2026

Boarding Call 🚀

BCD Travel just wired AI agents directly into corporate booking. Amex GBT relaunched its entire platform around conversational AI. And 58% of travel managers say AI has had zero impact on their program. That gap is about to close — whether you're ready or not. This week: what's actually shipping, what your travelers are and aren't comfortable with, and the four questions you need in your next RFP before the vendors write them for you.

📌 This week: The TMC AI arms race stopped being a deck. It's shipping product now.

On May 28, BCD Travel — the world's third-largest TMC by sales ($24.4B in 2025) — pushed Model Context Protocol (MCP) live across its TripSource platform. Air, hotel, and car booking are wired into the framework today; rail is "coming soon." In plain English: AI agents can now shop, book, and check policy on TripSource using natural language, without anyone writing custom API code.

It is the most aggressive single-vendor agentic-AI move in managed travel since Amex GBT relaunched Egencia in April with a conversational AI assistant and a first-of-its-kind native Concur Expense integration. Sabre, CWT, Spotnana, Perk, and Navan are all sprinting in the same direction.

And it is shipping into a market where 58% of corporate travel buyers say AI has had little or no impact on their program to date — and only 1% say AI is being used strategically across their travel program. The gap between what TMCs are now selling and what your program actually uses is about to become a procurement question.

🤖 Why It Matters for Corporate Travel

1. MCP and agentic AI re-route who decides what gets booked. When the AI agent has the policy embedded at the data layer, out-of-policy options never even surface to the traveler. That's a different compliance model than after-the-fact reporting. It also means the quality of your policy data — negotiated rates loaded, sustainability tags, preferred-supplier flags — directly determines whether the AI is useful to your travelers, or quietly steering them around your program.

2. Buyer comfort with AI is uneven — and your TMC's roadmap probably ignores that. GBTA found 95% of buyers are comfortable letting AI recommend negotiated-rate flights and hotels — but only 64% are comfortable letting AI read employee calendars, and only 57% are comfortable letting AI change or cancel bookings. The features TMCs are shipping fastest sit squarely in the low-comfort zone. If you don't write the boundaries into your RFP, the vendor will assume the maximum.

3. The investor narrative is forcing the timeline. When Long Lake announced the $6.3B take-private of Amex GBT on May 4, the explicit thesis was AI-driven margin expansion. Skift reported that 46 of 64 potential buyers walked away citing AI risk — concern that AI will compress TMC margins faster than promised. Either way, TMCs now have to show AI productivity to the market every quarter. Your account is one of the data points they will use. Make sure what they ship to you matches what you want.

🛠️ What This Means for Your Travel Program

📝 Add an "agentic AI readiness" section to your next TMC RFP. Four concrete questions:

1. Does your booking platform support MCP or an equivalent standard, and which domains are live today?
2. How is corporate policy enforced at the agent layer — pre-filter, post-filter, or hybrid?
3. What is your roadmap for AI-driven disruption management, and how do we configure the boundary between agent action and human approval?
4. What customer data trains your agent models, and how is our negotiated content protected?

🔍 Audit your policy data quality before you turn anything on. Agentic booking is only as good as the rate sheets, supplier tags, and policy logic the AI can see. Run a sweep with your TMC: how many of your negotiated hotel rates are loaded and tagged correctly? An AI agent with bad inputs will confidently route your travelers to the wrong places.

🧪 Pilot in one cabin or one region, not company-wide. The TMCs are pushing fast deployments. The buyer-comfort data says your travelers aren't there. Run agentic booking on a single corridor for a quarter, measure leakage and policy compliance, then expand. The vendors will pitch big-bang; resist.

🔒 Lock the data-rights language now. Once your bookings, traveler profiles, and trip patterns are training a TMC's models, getting them back is hard. Add a clause defining what data trains shared models vs. your dedicated instance, and what happens to model weights at contract exit.

📊 The Numbers

• May 28, 2026 — BCD Travel MCP for TripSource goes live: air, hotel, car booking; rail coming soon
• $24.4B — BCD Travel sales in 2025; clients in 170+ countries
• 58% — buyers saying AI has had little or no impact on their travel program (GBTA, March 2026)
• 1% — travel buyers using AI strategically across their program (BTN, 2026)
• 95% — buyers comfortable with AI recommending negotiated-rate flights and hotels
• 57% — buyers comfortable with AI changing or canceling a booking
• $6.3B — Long Lake take-private of Amex GBT, announced May 4, 2026
• 46 of 64 — potential buyers who passed on the Amex GBT deal citing AI risk
• <3 minutes — average booking time on relaunched Egencia AI platform

⚡ The Bottom Line

A year from now, every Tier-1 TMC will pitch you a working agentic AI stack. Most corporate travel programs will buy one without knowing what to ask. The buyers who write the right four questions into the next RFP — MCP support, policy enforcement at the agent layer, disruption-management boundaries, and data rights — will get the version of agentic AI that serves their travelers. The buyers who don't will get the version that serves the TMC's margin story. Your next RFP cycle is the moment the difference compounds.

❝

Manage travel. Don’t just book it.

— Egor Karpovich, Co-Founder, Travel Code

Your Partner in Corporate Travel

Save up to 20% on corporate travel with Travel Code’s powerful tech, no legacy systems, and personal service across flights, hotels, and more. ✈️ 💛

P.S.

New here? Check our Self-Booking travel platform and share it with your colleagues.