Verdantas builds agents for proposal development, contract management, and collaboration using Microsoft Copilot Studio

Verdantas is a leading environmental science, engineering, and consulting firm formed by three companies. It has expanded quickly to include over 36 companies, 2,500 professionals, and nearly 100 offices across the United States.

The company recognized AI's productivity benefits and responded by acquiring Microsoft 365 and developing AI agents with Copilot Studio. “We’re continually amazed at how easy it is to build effective agents with Copilot Studio,” says Mike Boisvert, VP/Director of Enterprise Business Transformation Office at Verdantas.

At the core of the agentic network at Verdantas is a technical resource agent, which enables employees to access information more efficiently from across the company. The agent utilizes advanced large language models in conjunction with Retrieval-Augmented Generation (RAG) techniques.

Employees are able to employ the agent for a broad range of tasks:

• Smart search and insights: Users can instantly locate critical information across shared drives, Microsoft Teams channels, project documents, spreadsheets, ERP systems, and other platforms by leveraging curated search prompts. This improves efficiency and reduces time spent navigating siloed systems.
• Proposal support: The solution enables teams to summarize RFPs, retrieve previous proposals, identify similar submissions, and access marketing materials tailored to specific markets and submarkets. It also allows proposal teams to identify relevant employee skills and licensure, helping craft more compelling, data-driven responses.
• Project and client intelligence: Users can quickly retrieve information on past and current clients, including project history, financials, and qualification requirements. This functionality supports both business development and seamless project execution.
• Team and resource optimization: The system provides access to employee profiles, including skills, certifications, roles, and departmental information. This improves team collaboration and facilitates more effective resource planning and deployment.

Data done right

One of the keys to the success of the agent is the way it was engineered to handle data. Rapid growth through acquisition had propelled Verdantas into a strong leadership position but it also left the company with a complex and often disconnected network of systems and data. “We realized that, to get the most out of AI, we first needed to clean up our data,” says Boisvert.

To solve the problem, the company moved data from various systems across the network into Microsoft Fabric. The agent connects to a database created in Microsoft Fabric, aggregating data from systems like Workday for employee information, Deltek (financials), and Harbor Compliance (licensure).

Fabric acts as an ideal data engineering and analytics layer, unifying data from different sources and eliminating the need to build separate Extract, Transform, Load (ETL) pipelines. Dataverse then serves as the operational data layer.

At the same time, cloud flows in Power Automate pull data from file shares and other systems like ServiceNow. By integrating diverse datasets within a unified intelligent platform, the agent enables efficient sourcing of essential information from across the organization.

Additionally, Azure AI Search provides the agent with advanced search engine capabilities, including vector-based retrieval and semantic reranking. This results in more precise and relevant answers back to users. A custom API and tailored prompts also help ensure the agent aligns with internal terminology and delivers relevant results.

Advanced agents for complex challenges

“Once we started leveraging Copilot Studio and our consolidated data strategy, we never looked back,” says Boisvert. The company has since accelerated its development of AI-driven agents, many of which go beyond information retrieval.

A contract management agent can quickly assess contract risk. Users upload contracts, and the agent reviews them using legal team policies stored in SharePoint. It identifies escalation clauses and highlights sections that may need negotiation. Tasks that once took hours now take minutes.

In addition to developing new agents, Verdantas leverages the capabilities of Copilot Studio to improve existing agents, such as its technical resource agent. For instance, when supporting users with proposal development, the agent aggregates data from numerous sources, drawing on project experience stored across the organization. This process can impact performance and introduce contextual challenges.

Verdantas is using multi-agent orchestration within Copilot Studio to address this challenge effectively. By establishing sub-agents and assigning each to a specific data source, the principal agent employs generative orchestration to determine which agents should be activated. This capability delivers more accurate responses for employees through a singular agentic interface. Multi-agent orchestration has also dramatically improved the speed of the agent, enabling it to respond to users’ questions up to two times faster.

As Boisvert says, “Our agents process highly complex and varied datasets, and multi-agent orchestration in Copilot Studio has provided an effective solution for managing these resources and enhancing overall agent performance.”

Governance at scale

Another important aspect of the AI strategy at Verdantas was the development of a strong AI governance program.

This began with the development of a Center of Excellence using the Power Platform Center of Excellence Starter Kit. The COE allows for the development of agents in separate test environments before moving to production. Managed Environments capabilities enable automation of various stages of the development process including environment routing and quality checks. For enhanced visibility into agentic activity, Verdantas is also using the Copilot Studio Kit.

The program also emphasizes robust data controls and privacy through the integration of Microsoft Purview, ensuring advanced encryption and compliance capabilities. Overall, Verdantas’ AI governance framework reflects the company’s commitment to lawful practices, transparency, and sustainable growth.

”Being a frontier firm requires intentional foundations,” says Sujan Turlapaty Chief Digital & Information Officer at Verdantas. “At Verdantas, we started our AI journey with a strong focus on data quality and governance, and that early investment has accelerated our ability to scale Copilot Studio development with confidence. This approach is modernizing how expertise moves across the firm and enabling sustainable, responsible innovation.”

Into the future

News of the agent's effectiveness on Copilot Studio is rapidly circulating within Verdantas. The organization has received over 150 requests for additional agents from various departments. Among the proposals under review are enhanced support for proposal development, project risk management, and expanded integration with marketing operations.

With proper data, governance policies in place – and a flexible development platform built in Copilot Studio - Verdantas is ready to meet this growing demand and continue its fast-moving, AI journey, delivering intelligent automation across the company.

As Boisvert says, “Verdantas is taking a pragmatic, governed approach to AI. By leveraging Copilot Studio, Microsoft Fabric, and Foundry within our secure environment, we are scaling knowledge access and productivity in ways that are immediately useful to our teams and aligned with how we operate as a firm.”