RAG in AI: How It Can Help a Technical Delivery Manager

Imagine This…

You are a Technical Delivery Manager managing a large software project.

Over the last year, your team has created hundreds of documents—requirements, user stories, architecture documents, Jira tickets, meeting minutes, change requests, risk registers, project plans, test cases, client emails and status reports.

One day, your client asks: “Why was this feature implemented differently from what we originally agreed?”

You know the answer exists somewhere in your project documentation. But where?

You start searching through emails, Jira, SharePoint and meeting minutes. After almost an hour, you finally find the answer.

Now imagine asking an AI assistant: “What did we originally agree about this feature, when was it changed, and who approved the change?”

Within seconds, the AI finds the relevant information and provides a summarized answer with references to the source documents.

This is where RAG can become extremely useful.

What Is RAG?

RAG stands for Retrieval-Augmented Generation. 

The name sounds technical, but the concept is simple. Think of RAG as giving an AI assistant access to your organization’s knowledge before asking it to answer a question.

A traditional AI model answers primarily using its existing knowledge and the information provided in the conversation.

A RAG-based system follows a different process:

Question → Search relevant information → Retrieve the right content → Generate an answer

For example:

“What was the decision regarding the reporting module?”

The RAG system searches approved project information, finds the relevant meeting minutes, change requests or project documents, and then generates an answer based on those sources.

In simple words:

RAG is like an intelligent librarian who can search your project library and explain what it finds.

A Simple Project Story

Let’s take Raj, an experienced Technical Delivery Manager. Raj manages a large insurance application implementation involving multiple development teams, QA teams, vendors and business stakeholders.

During a client meeting, the client asks: “Why isn’t the reporting module included in the current release? We discussed this months ago.”. Raj remembers the discussion but cannot recall the exact decision. He searches Jira. Then Teams. Then emails. Then project documents. Finally, he finds an old Change Request. The decision was made six months earlier to move the reporting module to the next release. It took Raj almost an hour to find the information.

Now imagine Raj has a RAG-enabled project assistant.

He simply asks: “Why was the reporting module moved to the next release?”

The assistant searches the project’s approved information and responds: “The reporting module was moved to Release 2 following a Change Control Board decision. The change was documented in CR-247 and approved during the March steering committee meeting.”

Raj can then verify the information from the source documents.

One hour of searching has potentially become a few seconds of retrieval.

Why Should a Delivery Manager Care About RAG?

As Delivery Managers, we deal with huge amounts of information every day.

The challenge is often not a lack of information. The challenge is: Finding the right information at the right time.

RAG can help address this problem in several areas.

1. Faster Requirement Analysis

Imagine your project has hundreds of requirements.

Instead of manually searching through documents, you could ask: “Show me all requirements related to customer notifications.”

RAG can retrieve relevant requirements and summarize them. This can help Delivery Managers and Business Analysts understand scope faster and reduce time spent searching.

2. Better Change Management

Change requests can affect scope, cost, schedule, resources, architecture, testing and release plans.

Suppose the client asks: “Can we add OTP authentication?”

You could ask: “Which existing requirements, APIs, security controls and test cases could be impacted by this change?”

RAG can retrieve relevant project information and help prepare the impact analysis.

The important point is that AI supports the analysis; the Delivery Manager makes the decision. 

3. Faster Risk and Issue Analysis

Project risks are often hidden across different sources.

For example:

• A dependency is mentioned in a meeting.
• A delay appears in Jira.
• A resource problem appears in a status report.
• A vendor issue appears in an email.

Individually, these may look like small issues. Together, they may indicate a delivery risk. You could ask: “What unresolved dependencies and issues could impact the next release?”

RAG can retrieve related information and help you see the bigger picture.

4. Easier Status Reporting

Every Delivery Manager knows the weekly reporting challenge.

Information needs to be collected from development, QA, product owners, Jira, risk registers and project plans. A RAG-enabled assistant could create a draft project status summary using available project information. You review it, validate the facts, add your management perspective and publish it.

The objective is not to let AI blindly create reports. It is to reduce the time spent collecting and consolidating information.

5. Better Client Meeting Preparation

Before an important client meeting, you could ask: “Summarize the last three meetings with the client and list open actions, commitments, risks and agreed dates.”

Instead of reading multiple meeting minutes, the RAG assistant can retrieve and summarize the relevant information. You walk into the meeting better prepared.

RAG as Your Project’s Digital Memory

One of the biggest challenges in IT projects is knowledge loss.

• People leave.
• Teams change.
• Architects move to other projects.
• New managers join.
• Valuable project knowledge can disappear with them.

A RAG-based knowledge assistant can make project information easier to access.

A new Delivery Manager could ask: “Explain the history of this project.”

Then: “What were the major architecture decisions?”

Then: “What are the unresolved client issues?”

Instead of searching hundreds of documents, they can interact with the project’s digital memory.

RAG vs Traditional AI

The difference is simple.

Traditional AI

Question → AI → Answer

RAG

Question → Search your knowledge → Retrieve information → AI → Answer

RAG can connect AI with sources such as:

• SharePoint
• Confluence
• Jira
• Requirements
• Project documents
• Meeting minutes
• Knowledge bases

The exact sources depend on the organization’s implementation.

RAG and the Delivery Manager

RAG does not mean: “AI will manage my project.”

Think of it as: “AI can help me find and understand project information faster.”

This can allow Delivery Managers to spend less time gathering information and more time on:

• Decision-making
• Risk management
• Stakeholder communication
• Client relationships
• Team leadership
• Problem solving
• Delivery strategy

The traditional approach is:

Search → Collect → Consolidate → Analyse → Decide

With RAG:

Ask → Retrieve → Summarize → Validate → Decide

The human remains responsible for the decision.

What Should a Delivery Manager Consider?

RAG is powerful, but it is not magic. If the underlying information is outdated or incorrect, the answer may also be unreliable.

Before implementing RAG, consider:

• Data Quality: Is the information accurate and current?
• Security: Can AI access confidential project information?
• Access Control: Who should be allowed to see the information?
• Source of Truth: Which document should be trusted when information conflicts?
• Traceability: Can the AI answer be linked back to the original source?

These are not only technical questions. They are also project governance questions.

Start Small

You don’t need to start with a massive AI transformation.

Begin with one practical use case:

Project Knowledge Assistant

Give it controlled access to:

• Requirements
• Architecture documents
• Meeting minutes
• Change requests
• Project plans
• FAQs

Then allow your team to ask questions.

Measure:

• Time saved
• Search effort reduced
• Answer accuracy
• User adoption
• Faster knowledge transfer

If the results are useful, expand gradually.

Final Thoughts

After more than two decades in IT delivery, you have probably experienced this many times: 

The information already exists.

• A requirement is somewhere.
• A decision is somewhere.
• A risk is somewhere.
• A client commitment is somewhere.

The challenge is finding it when you need it.

RAG can help bring that information to the surface. Think of RAG as an AI-powered project librarian or your project’s digital memory. It can search, retrieve and summarize information. But your experience as a Delivery Manager remains critical.

AI can help answer: “What do we know?”

You still need to answer: “What should we do?”

And that is where the real value lies.

RAG doesn’t replace your experience. It can help you access the information needed to apply that experience faster.