One use case, many paths: Solving RFP Automation across the Microsoft Cloud

Introduction and the RFP Challenge

The webinar begins with Sean from the ESPC community introducing Jake Harvey. Jake explains that the session focuses on how the Microsoft Copilot product family can be used to increase productivity and accuracy when responding to RFPs (Requests for Proposal).

He describes the standard RFP process as receiving documentation, understanding the customer’s challenge, leveraging internal subject matter expertise, delegating tasks, and finally writing and submitting the response in Word or PowerPoint. Common challenges in this process include:

• Information Overload: Reading and digesting paragraph-heavy requirements is extremely time-consuming and increases the risk of missing important details.
• Knowledge Management: Captured knowledge is often difficult to access, especially when key employees leave the organization.
• Tight Deadlines: There is rarely enough time to assemble the ideal team, making SME availability a constant challenge.
• Manual Review: Conducting a holistic review to identify gaps in a final response is an exhausting manual process.

Automation Strategy and the Microsoft Ecosystem

Jake notes that many organizations are moving away from dedicated off-the-shelf RFP software because they prefer to work where their content already resides, such as within Microsoft 365. He advocates for an “ecosystem” approach, viewing automation as a series of “Lego blocks” categorized into three types:

• Personal Automation: Tools that help individual employees work faster and more effectively.
• Defined Automation: Workflows that handle repeatable tasks such as task allocation and delegation.
• Intelligent Automation: Solutions that hand parts of the process over to technology for autonomous deep research and reasoning.

Demonstrations of RFP Automation

Jake presents four distinct approaches to RFP automation within the Microsoft ecosystem:

M365 Copilot

He demonstrates how Copilot can intelligently retrieve company knowledge from SharePoint to draft a response regarding Power Platform governance. During the demo, he clarifies that RFP stands for “Request for Proposal.”

SharePoint Agents

Jake shows how to create focused agents for specific content libraries, such as “social value” commitments. These agents deliver high accuracy because they are restricted to curated, high-quality datasets, helping to reduce AI hallucinations.

Copilot Studio

This demo features a “Technical Architect” agent that ingests licensing guides and past project data to automatically generate a first-draft technical proposal.

Custom Azure Solution

For more complex requirements, Jake showcases a custom Azure-based tool that extracts questions from an RFP document and matches them with “exemplar” responses stored in SharePoint to create a complete response draft.

Conclusion and Key Findings

Jake emphasizes that the goal of these tools is to shift humans from being “document authors” to “document refiners.”

He highlights that real-world RFPs are highly complex; questions are often implied or divided into multiple parts, requiring sophisticated AI reasoning capabilities.

During the closing Q&A session, Jake identifies data quality as the number one reason RFP automation initiatives fail, noting that technology cannot compensate for poor underlying knowledge management. He also defines Microsoft Foundry as a collection of professional-grade AI capabilities within Azure that provide developers with greater control and fewer restrictions than standard Microsoft 365 products.