i-Pharm Consulting is a specialized pharmaceutical staffing firm with offices in London, New York, and Sydney. Part of the IPA Talent Group, the company places candidates across the life sciences industry, running its recruitment operations out of Bullhorn, the CRM that most agencies of its size treat as the system of record.
When Simon Burns, i-Pharm's Executive Director, started exploring what an AI-assisted recruitment workflow could look like, the answer was clear: it had to run through Copilot. The entire team lived in Microsoft Teams, everyone had access to Copilot, and the firm was in the process of rolling paid Copilot licenses to all staff globally. Building agents in Copilot Studio and delivering them through Teams was the natural path. The only problem was that Bullhorn didn't offer a way to connect.
No connection between Bullhorn and Copilot Studio made AI feel manual
The most immediate use case Simon wanted to build was candidate-to-job matching. Recruitment consultants routinely evaluate whether a candidate profile is a good fit for an open role, a task that involves reading a job specification, reviewing a CV, and making a judgment call. AI is well-suited to accelerate exactly this kind of structured comparison, and Copilot Studio made it straightforward to build an agent that could do it.
The problem was access to the data. Without a direct connection between Bullhorn and Copilot, the only way to make it work was manual: download the job spec from Bullhorn, download the candidate's CV, upload both into Copilot, then ask the question. That workflow added friction at every step and defeated much of the purpose of having an agent in the first place. What Simon needed was for the agent to pull directly from Bullhorn, so a consultant could simply say "compare candidate 123 to job 456" and get an answer.
Bullhorn doesn't currently provide an MCP endpoint meaning third-party connectivity a necessity for any team wanting to build on Bullhorn data using an external AI platform.
"Even with AI, the process was quite manual. We had to download two individual files and then re-upload them back into an agent—which defeats the whole point of AI."
—Simon Burns, Executive Director, i-Pharm Consulting
An MCP layer that provides direct access to live Bullhorn data
Simon found Connect AI by querying Copilot and Perplexity directly: he wanted to know what his options were for connecting Bullhorn to Copilot Studio. Connect AI appeared alongside a couple of alternatives; one that was free but technically demanding to configure, and another at a comparable price point. He started with the Connect AI trial, configured the Bullhorn connector out of the box, and had it running before he'd spoken to anyone at CData about pricing.
Getting the AI agent configured and running required no custom code, no API wrapper, no developer on the Connect AI side. The only configuration work was on the Copilot Studio side, shaping the agent's instructions and testing prompts.
Simon deployed the candidate matching agent to a pilot group of four to five recruitment consultants. The agent takes a candidate record and a job record from Bullhorn and produces a structured comparison: a gap analysis highlighting strengths, potential gaps, and relevant experience. The output provides recruiters with a consistent framework to support candidate review, helping ensure the same criteria are considered for each candidate-job comparison. A second agent, built to interrogate i-Pharm's commission system, is also in test.
"CData Connect AI was easy to set up and it functioned as intended right away: real-time access to Bullhorn through Copilot. Now I have the foundation to build agents and keep on running."
—Simon Burns, Executive Director, i-Pharm Consulting
The road ahead: From pilot AI to central agent orchestration
The candidate comparison agent is in active pilot with a small group of consultants. The early signal is positive: the agent applies the same structured comparison framework to every candidate-job review, helping recruiters assess opportunities in a more consistent and repeatable way while keeping decision-making firmly in human hands. Before rolling it out to the wider team, Simon is running the pilot long enough to establish a reliable cost-per-query baseline. Once that number is understood and extrapolated across the wider consultant population, the scale decision becomes straightforward.
Sage 200C is next source on the roadmap. i-Pharm's finance director is already interested in building agents on top of the firm's financial data. CData has a connector ready for Sage 200C, which means the data layer won't be the bottleneck when that project kicks off.
Longer term, Simon's vision is agent orchestration: a central agent coordinating specialist sub-agents for Bullhorn, Sage, and other systems, all surfaced through Microsoft Teams, where the entire i-Pharm team already works. Connect AI is the data layer that makes each of those sub-agents possible, one source at a time.
CData Connect AI: Live enterprise data where your agents need it
For recruitment teams building on Microsoft Copilot Studio or any other AI tool, access to live CRM data is what separates a useful agent from a general-purpose one. CData Connect AI gives you a direct MCP connection from Copilot Studio to Bullhorn and to the other systems your agents will eventually need without custom development or API work on your side.
