This is the Trace Id: fe37cab6b0193e4a543bfdb744e1d9fb
3/24/2026

Unifi manages contracts more efficiently with AI using Power Platform and Copilot Studio

As Unifi quickly grew to become North America’s largest provider of aviation ground handling services, the company was spending more and more time manually processing and reviewing documents, including contracts.

Copilot Studio and Power Platform were selected to solve a problem that the company had not been able to address with other AI platforms: process and analyze large files full of specialized industry and legal terms and enable precise, natural language search by Unifi’s legal team.

The system has reduced contract processing from days to minutes and delivers the same level of performance as much more expensive, off-the-shelf products built specifically for the legal industry.

Unifi

Unifi is North America’s largest provider of aviation ground handling services. Operating in over 240 locations and employing more than 45,000 people, Unifi delivers comprehensive services including ground handling, passenger transport, security, cabin cleaning and ground support equipment maintenance. The company operates across the United States, Canada, the U.K., Ireland and the Netherlands.

Since its rebrand in 2020, the company has scaled significantly—from around $450M in revenue and 10,000 employees to more than $2 billion in revenue and more than 45,000 employees, all while expanding its geographic footprint and becoming a full-service aviation services leader.

As the company has grown, Unifi’s legal team was spending more and more time on manual document processing tasks. One particularly time-intensive task was related to the collective bargaining agreements (CBAs) that Unifi has with its union employees.

Multiple full-time resources were required to extract and structure the information in these contracts for use in new agreements and responses to business queries. This process was consuming hundreds of hours a year. Even more time was spent searching the data, which further limited the team’s ability to respond quickly as the business scaled.

Power Platform and Copilot Studio, better together

Unifi tried several AI-driven solutions to streamline the contract management process, but these early projects yielded poor results. The problem was that company’s CBAs are often more than 100 pages long and full of specialized industry and legal terms—which often vary widely across contracts. This lack of structure and consistency across large files made it difficult for the agentic systems to correctly parse and understand the data.

“With other AI products, we never got past the POC phase because of the accuracy of the information coming back,” explains TK Dunlap, Director, Asset Management & Infrastructure. “We’d ask an agent the same question twice about a contract and get back two different answers. The consistency and accuracy of the model just wasn’t there.”

Unifi ultimately solved the problem using Copilot Studio and Power Platform. Instead of relying on one agent to do all the work, Unifi developed an agent supported by multiple AI models and automations, each tackling a specific part of the contract data extraction and AI modelling process. “With Copilot Studio and Power Platform, we didn’t just build one agent but a powerful, multi-stage, AI-driven system. That made all the difference and enabled us to address a complex contract management challenge,” says Dunlap.

Now, the legal team can simply upload a contract into SharePoint and key terms are immediately extracted and made searchable via an AI-driven agent in Copilot Studio. In test after test, the agent was able to do what other agents could not: process and analyze huge contracts—up to 200 pages long and 50MB in size—enabling accurate natural language search by Unifi’s legal team.

“Our AI-driven, contract management solution built on Power Platform and Copilot Studio delivers the same performance as off-the-shelf products costing 30 times more.”

TK Dunlap, Director, Asset Management & Infrastructure, Unifi

The system has not only vastly outperformed POCs built on other platforms but enabled Unifi to bypass expensive, off-the-shelf products specifically built for the legal industry. As Dunlap says, “Our AI-driven, contract management solution built on Power Platform and Copilot Studio delivers the same performance as off-the-shelf products costing 30 times more.”

As one employee says, “This system has literally changed the way I work. Instead of spending hours extracting data from contracts, I now have more time to focus on strategic legal work that’s more interesting to me and much more valuable to Unifi.”

Here’s how the system works in the background: When a contract is loaded into SharePoint, a series of AI models are triggered via Power Automate. These models, developed in AI Builder, part of Power Automate, automatically extract full contract text, identify key clauses (such as contract expirations, health insurance details, paid vacations), generate document summaries, and convert information into structured metadata.

The data is then stored in a searchable contract knowledge repository in Dataverse. This enables fast retrieval and query-based insights for legal teams using an AI-driven agent which runs as a custom agent in Microsoft 365 Copilot. Also, a Power BI dashboard provides managers with a high-level view of documentation and activity. For example, managers can view active contracts by industry, value, and expiration.

Innovative, low code design

There are several innovative parts of the system worth noting that make Unifi’s contract management solution perform so well.

One is the text recognition pre-built model in AI Builder. Most of the PDF files containing Unifi’s CBA data are scanned images which are often inconsistent and low quality. Some images show off-centered documents, other images are blurry and contain hard-to-read text. Despite this, the model performed with 100% accuracy compared to manual extractions. “We were blown away by the accuracy of the OCR capabilities that came with AI Builder in Power Automate. We really weren’t sure if it could handle the low-quality images that we needed to process but it worked perfectly,” says Dunlap.

Once raw text was extracted, the next important step was making sure the model knew exactly which text and attributes should be added to Dataverse. This is not always straightforward with CBAs. For example, when Unifi lawyers ask an agent for geographic location for services provided in a CBA, they don’t just need a city but also the airport code. A generative AI model might miss the nuance. Run a Prompt within Power Automate made it easy for the team to add key deterministic details to course correct the extraction model and even customize the format of the attributes. In fact, using low code tools, the entire solution took less than two months to complete and go live.

“We were blown away by the accuracy of the OCR capabilities that came with AI Builder in Power Automate.”

TK Dunlap, Director, Asset Management & Infrastructure., Unifi

Along with tools to support the accurate extraction of data, Power Automate was used to help the agent in Copilot Studio better understand the data extracted. Again, with the specialized terms found in CBAs, this can be complicated, especially for a general purpose GPT model not specifically trained on legal documentation.

With this system, the legal team simply enters specialized terms into an excel spreadsheet and a Power Automate flow automatically updates the knowledge sources used by the LLM working behind the agent. This ability to quickly inform the LLM with specialized knowledge is especially valuable to Unifi as contract terms are often changing and being updated. “With Power Automate, an automated knowledge flow essentially turns our general-purpose GPT model behind Copilot Studio into a specialized AI model for contracts,” explains Rima Reyes, Principal Product Manager at Microsoft who helped Unifi develop the initial build of the solution.

The system’s innovative use of Dataverse also contributes to agentic performance. Here’s the trick: Extracted data is moved into not one but two tables.  One table contains the raw text and page numbers, and another table contains headers, metadata and other elements of the text. “Metadata like headers, sub-headers—even page breaks and special characters—are noise that can impede agentic search, especially with large documents,” explains Reyes. “Isolate those elements from raw text across separate tables in Dataverse and you get more accurate results from your agent.”

Building on a flexible platform

Based on the success of its contract management solution for CBAs, Unifi has plans to complete a similar system for managing contracts with customers. By building on the existing CBA solution, Unifi expects the new system to be completed in a matter of weeks.

The company is currently exploring opportunities to leverage the solution and contract data for other business processes, including answering manager queries and linking to financial workflows. Similar extended capabilities are planned for the CBA solution.

“We’ve been able to take what was once a cumbersome, hard to manage set of contract files and turned them into a powerful database that can be easily leveraged for all sort of business uses,” says Dunlap. “With Power Platform and Copilot Studio, the possibilities are endless - and extremely exciting.”

Learn more about Unifi on LinkedInFacebookTwitter, and Instagram.

Take the next step

Fuel innovation with Microsoft

Explore more customer stories

Find out how customers are achieving more with Microsoft products and solutions.
A man wearing headphones and smiling.

Talk to an expert about custom solutions

Let us help you create customized solutions and achieve your unique business goals.
Three people in a meeting room.

Transform work with Microsoft AI

Bring intelligence into the flow of work and help your organization achieve its goals with secure, scalable AI solutions.

Follow Microsoft