Businesses are preparing to hand a growing share of their business travel purchasing to AI agents.
By 2028, respondents estimate that AI could manage or authorise an average of 37% of their spend in this area, according to virtual payments fintech Conferma.
Based on a GBTA estimate, that would equate to $700 billion.
The study was conducted in September 2026 by Sapio Research for Conferma among 600 decision-makers in business travel and finance in the UK, the US, Australia and New Zealand.
It also finds that 28% of organisations expect AI to take on more than half of their travel budget.
That share rises sharply among large companies.
Among those with annual business travel spend of at least $6.8 million, the proportion likely to be managed by AI reaches 51%, compared with 27% among organisations spending less than $136,000 a year.
Senior leaders are particularly confident about these new uses.
Some 86% of executives and heads of department say they trust AI to manage travel bookings and spend.
That proportion falls to 62% among travel, procurement and travel management professionals.
“Business travel could be one of the first testbeds for agentic payments.”
That confidence is nevertheless tempered by some concerns.
97% of respondents say they are worried about AI-driven autonomous purchasing.
Data security is the top concern, cited by 59%, ahead of the risk of error, mentioned by 43%.
While 81% of organisations say they have already put in place a policy, control or guardrail to govern AI-made purchases, only 36% regard those measures as comprehensive.
The question is therefore no longer simply whether an AI can book a trip, but what it is authorised to buy, from which suppliers, within what limits and with what level of human oversight.
Mark Ledsham, chief executive at Conferma, said: “Agentic spending authority is developing faster than the controls around it. Corporate travel could be one of the first proving grounds for agentic payments because it combines high transaction volumes, clear policies and significant spend.
Giving an AI agent authority over company money is very different from using AI to recommend a flight or hotel. Businesses need clear limits on what agents can spend, where they can spend it and when human intervention is required.”
Ledsham also said businesses will need to be able to pinpoint the origin of a decision, keep a record of every step and assign responsibility clearly in the event of an error.
“The controls around those decisions must remain fixed, traceable and accountable.”
