Every major industry has an infrastructure moment. It is the point where the tools catch up to the scale of the work, and everything changes.
Retail had it with Shopify, which now processes more than $300 billion in annual commerce. Restaurants had it with Toast, which serves roughly 140,000 locations. Workforce management had it with Rippling, built on a single thesis: fragmented software cost businesses time and money, and a unified platform could fix it.
None of these platforms began with intelligence. It came after the infrastructure was in place. Once a business ran on one system instead of ten, that system could finally see enough to act.
Travel is having that moment now. And one of the largest host agencies in the country has decided to bet on it.
Why nobody solved this before
The travel advisor channel, the agents who plan and book trips for clients, moves $130 billion a year in the U.S. alone, according to ASTA, the travel advisors' trade association. Behind that number sits a fragile chain. An advisor logs a booking into their CRM, then submits it by hand to their host agency, the larger organization advisors operate under. The host receives commission from suppliers, the hotels, cruise lines, and tour operators paying it out, imports it into a separate tool, and matches it line by line. If anything fails to reconcile, the loop starts over. Every layer does the same work twice, and none of the systems talk to each other.
Fragmentation did not only create manual work. It made the data useless. No single system saw enough of the chain to reconcile on its own, let alone catch a problem before it happened.
This was not an oversight. People tried to fix it. TPI, one of the largest host agencies in the U.S., built its own back office platform, Suitcase, and ran it across its entire network for 16 months. Tech companies tried too, and almost every attempt solved one slice: a CRM that never touched commissions, a reconciliation tool with no front end, a reporting layer that couldn't see bookings in real time. A team would build one layer, hit a wall, and find the problem ran deeper.
Adding AI to one layer is easy now, and plenty of tools do it. The limit is context. A reconciliation tool that can't see bookings, or a CRM that can't see commissions, hands the AI a sliver of the business. It can speed up one task. It cannot reason about an operation it cannot see.
Infrastructure is hard to build in travel for an architectural reason, not a technical one. The business has many layers: advisors, sub-agencies, host agencies, consortia (the buying groups agencies join for better supplier deals), and suppliers, each with its own data, commission structures, and workflows. A tool built for one creates friction at all the others. The only thing that works is a platform that spans the whole vertical. That is what nobody built, until now.
What TPI's decision signals
TPI has chosen to move its entire network, 5,500 advisors, onto Tern, and to cover the cost for every one of them. This is not a software switch. TPI had already proven it would build its own system. They know better than almost anyone how hard this problem is.
Giving AI the full context of a business is not a feature you bolt on. It is a platform you build underneath it first, across booking, commissions, reconciliation, and reporting, before the intelligence has anything real to work with. That is more engineering than any single host can take on while running its network.
Their decision is a signal from the inside. A host agency that built its own infrastructure does not replace it unless it has found something that solves the problem at a level it could not reach alone. It is a vote of confidence from a team that knew exactly what it was evaluating.
What a full stack changes
The difference between point solutions and a full infrastructure stack is not only efficiency. It changes what is possible.
On Tern, a booking submitted by an advisor flows directly through to the host agency's back office, and reconciliation against supplier commission payments happens automatically. The AI built into the platform does the matching, reads the payment files, and flags the exceptions, work that used to consume hours of back office time every week.
Automation is only the first thing the AI does. As more of the business runs through one system, it turns that operational work into insight the business can act on before it becomes a problem: which suppliers are paying late, which bookings are at risk of failing reconciliation, where an advisor's book is exposed. Work that used to be reactive becomes something the platform sees coming.
When the whole chain runs on one system, from the moment a booking is made to the moment a commission is reconciled and paid, the gains show up at every layer. An advisor can operate at a volume and margin that disconnected tools can't match. An agency can scale without adding back office headcount. A host can see its network's performance in real time rather than at the end of a reconciliation cycle. A consortium can see demand and production across every member, instead of waiting for each agency to report it after the fact.
Where the industry goes from here
Earlier infrastructure platforms made businesses more efficient. AI built into the infrastructure makes them smarter over time, at every layer at once. And because the picture gets more complete the more the business runs on one system, the advantage doesn't stay flat, it compounds.
In three years, running a host agency without a full infrastructure stack will carry the same disadvantage as running a restaurant without a modern POS, or a retail business without a unified commerce platform. The agencies that move first capture that advantage early: their advisors produce more, their back offices cost less to run, and the platform itself becomes a reason top advisors want to join them.
Tern is the infrastructure layer for travel. With TPI bringing its full network onboard, that model is about to be tested at real scale.
The travel industry's infrastructure moment is here. The question is who moves next.