Dextr AI raises $6.7m to run hotel reservations and back office on five AI agents

Dextr AI has raised a $6.7m seed round led by Elevation Capital, with Foundation Capital participating. The announcement was dated 24 September.

The company deploys a set of connected AI agents into hotels to handle work that currently sits with staff: voice reservations, guest messaging and management, back-office and staffing tasks, and revenue work. There are five named agents, called Doss, Daisy, Alfred, Manny and Cosmo, and they are directed through natural language commands rather than a configuration screen.

Sajid Shariff is chief executive, Scott Arnold chief technology officer.

The traction claims

More than one million interactions a month across hundreds of properties in the United States, Canada, Europe and the United Kingdom, including franchised hotels operating under Hilton, Wyndham, Best Western and IHG flags.

Two specific outcomes are offered. A six-property hotel group is said to have improved EBITDA by 8% through 1,200 saved labour hours. A reservation centre is said to have captured $80,000 of incremental revenue within 60 days.

Shariff's framing: "The technology finally bends to the operation instead of the other way around."

Krishna Mehra of Elevation Capital said: "This team delivers immediate, measurable ROI, the kind an operator can see on the P&L."

The funding goes toward expanding the agent set and developing Doss Gen 1, a command layer for running AI workflows in natural language at enterprise operators.

How to read those numbers

They are the company's own, supplied with a funding announcement, and they are the kind that look precise without being checkable.

An 8% EBITDA improvement attributed to 1,200 labour hours at a six-property group tells you nothing about the baseline. Six small properties with thin margins can show 8% from a modest absolute saving. The $80,000 incremental revenue at a reservation centre has no stated comparison period and no indication of what else changed in those 60 days.

"Over a million monthly interactions" is also a softer figure than it sounds. An interaction can be a single automated message. Across hundreds of properties it averages out to a few thousand per property per month, which is plausible for a messaging and voice layer and says little about how much decision-making the agents actually carry.

None of that makes the product weak. It means the published evidence is marketing evidence, and the only way to know is a reference call with an operator who runs it.

Where this sits

This is the second funded AI operations company in the hotel space inside a fortnight. Sloan Dean's AI Hospitality Group launched on 21 September with $7.5m in seed funding and the claim that it runs its back office on around 60 AI agents.

The pattern is specific and worth naming. The money is not going into better front desk software. It is going into companies arguing that a chunk of hotel labour, in reservations, guest communication and back office administration, can be replaced rather than assisted. Dextr AI sells that as a layer on top of whatever systems a hotel already runs. AI Hospitality Group sells it as an operating company.

For an operator the question each raises is the same. Which tasks at your property are genuinely repetitive and rule-bound, and which only look that way until a guest does something unexpected at 11pm. The honest answer differs by property, and vendors in this category are currently incentivised to assume the generous version of it.