On the 22nd June, the Cloud Combinator team joined AWS at Right Revenue's offices in Northern Ireland to deliver Right Revenue's build day on the AWS AI Accelerator Programme. By the end of the session, the sales team had a working AI agent that prepares prospect research, built on Amazon Bedrock and shaped around the way the team specifically works.
The AI Accelerator Programme, delivered alongside AWS, pairs businesses that have a genuine commercial challenge with a technical team that can solve the issue. Right Revenue, the hotel revenue management platform used by independent hotels across the UK and Ireland, was one of five companies selected. This is the story of how we got from a discovery call to a working solution in a single on-site day, and the decisions that made it possible.
We did not arrive with a pre-built solution, and that was deliberate. The engagement began with a discovery call to understand how Right Revenue's team works and where the real cost of manual effort sits.
The answer was sales prospecting. Before the first conversation with a prospective customer, the sales team would spend hours researching each prospect: identifying suitable hotels, understanding ownership structures, reviewing the technology each property runs, and pulling commercial context together from multiple sources. The research matters, because it makes that first conversation worth having. Yet, it is also relentlessly repetitive, and this is where the challenged surfaced.
That combination, essential but repetitive, is exactly what a strong use case looks like. The task is well defined, it repeats constantly, and the value is immediate. The objective was never to replace the sales team. It was to remove the manual research, so Right Revenue’s time goes into conversations and relationships.
A one-day build only works if the day is spent building, not experimenting. After the discovery call, we documented the use case and shared it with Right Revenue for review. Our engineering team then tested the core approach in advance, running hotel discovery against a real market and sharing the output with Right Revenue's CTO for feedback before anyone travelled.
By the time we arrived on site, the open questions were about refinement, not feasibility. The sales team spent the day reacting to real output, and when they asked for changes, we made them the same day.
During scoping, the sales team asked whether the agent could also detect commercial signals: a new general manager starting, a completed renovation, an award win, a change of ownership. These are the moments a good salesperson acts on. We confirmed the approach could support it and built signal detection into the solution, running targeted, time-filtered searches per property and surfacing any findings in the output.
We were equally clear about the limits. The solution does not include account-level buying-intent data, because that requires specialised third-party platforms. Defining what the agent does not do is as important as defining what it does.
The solution runs on AWS, with Amazon Bedrock at its core. The agent takes a single request, such as a target market or property, and works through the research steps needed to build a complete prospect profile: property discovery, ownership and technology context, and commercial signals.
The most important design decision was grounding. A prospect profile is only useful if it is accurate, and a sales team will stop using a tool the first time it presents incorrect information. The agent therefore retrieves live information through Tavily's search tooling rather than relying on a model's stored knowledge, so every element of the profile is drawn from a current source. For a tool that feeds real commercial conversations, that is a requirement, not an optimisation.
The result is a single, structured prospect view that previously took hours of manual work across multiple sources, now available to the sales team on demand.
‘To have such a down-right clever team sitting around our table, and solving a real-world problem for us was quite literally mind-blowing!’
Adrienne Hanna Founder/CEO
Three things made the single-day build work.
Tight scoping. A clearly defined output, agreed sources, and a firm boundary between what the agent gathers and what the sales team judges.
Preparation. The approach was tested and reviewed before the build day, so the day was spent refining rather than experimenting.
Building with the users in the room. The sales team saw output during the day and shaped it directly, compressing weeks of feedback cycles into hours.
Behind all three sits the principle that runs through every engagement we deliver: start with the business problem, not the technology. Understand the process first, choose the use case where AI creates lasting value, scope it precisely, and build alongside the people who will use it, on an AWS-funded path that lets the business validate value before committing further investment.
The engagement continues, and the next phase is already in discussion.
If your team has a well-defined process that consumes significant time, the same route is available. Speak to us, or to your AWS account manager, about the AI Accelerator Programme.