Most companies are doing AI transformation backwards. They’re starting with the technology and hoping to figure out where it fits later. But Shruthi Shetty from SAP, who’s been deploying AI solutions across enterprise organizations for years, has a different approach that’s producing dramatically better results.

Her framework is deceptively simple: spend 30-60 days mapping your workflows before you buy or build anything. Then use that map to make smart decisions about where AI can actually drive value.

I know what you’re thinking. Another process before we can get to the good stuff? But here’s the thing: the organizations that skip this step end up with expensive AI tools that nobody uses effectively, while the ones that invest in mapping first see transformational AI ROI within months.

Let me show you exactly how to run your own workflow mapping sprint and why it’s the difference between AI success and expensive disappointment.

Why Most AI Implementations Fail Before They Start

Here’s a scenario playing out in organizations everywhere: leadership gets excited about AI, evaluates vendors for months, negotiates contracts, announces the rollout, and then watches adoption stall because the tools don’t fit how people actually work.

The problem isn’t the technology. The problem is that most organizations don’t actually understand their own workflows well enough to make smart decisions about where AI fits.

Shruthi put it perfectly: “Every day, a finance person knows what it is that they’re doing in an Excel sheet, or what it is that they’re doing in their analytics dashboard. So it’s important to understand from the ground up, from a manager. This is what my team does every day, this is what the day in their life looks like.”

But here’s the disconnect: while the finance person knows their daily routine, leadership often has only a high-level view of what “finance work” involves. They might know that the team produces monthly reports, but they don’t know about the fifteen different systems that need to be accessed, the manual data manipulation that happens in between, or the bottlenecks that eat up hours every week.

That detailed understanding is exactly what you need to make smart AI decisions.

The 30-60 Day Framework That Drives AI ROI

Shruthi’s workflow mapping approach is designed to be comprehensive but not overwhelming. Here’s how it breaks down:

Weeks 1-2: Choose Your Focus and Gather Your Team

Don’t try to map everything at once. Pick one function or process that’s strategically important and where you suspect AI could have significant impact. For most organizations, this might be something like customer service, content creation, data analysis, or procurement.

Assemble a small team that includes both managers and practitioners. You need people who understand the strategic goals and people who know the daily reality of getting work done.

Weeks 3-4: Document Current State

This is where you get into the weeds. Map out exactly how work flows through your chosen process. Not the official process documented in your procedures manual, but how things actually happen day-to-day.

Shruthi emphasizes the importance of understanding the full picture: “We never get our data from one dashboard, right? We have 15 dashboards, and then we all download it into an Excel or Sheets, and then, you know, massage the data and get into what we want to tell.”

Track:

• where people spend their time
• what tools they use
• where delays typically occur
• what manual work could potentially be automated

Weeks 5-6: Identify Pain Points and Opportunities

Once you have your current state mapped, look for patterns.

• Where do people spend time on repetitive tasks?
• Where do bottlenecks consistently occur?
• What work requires expertise that’s in short supply?

But also look for opportunities that might not be obvious. Sometimes the biggest AI wins come from enabling people to do things they couldn’t do before, not just doing existing work faster.

Weeks 7-8: Evaluate Existing Solutions

Before you start building custom solutions, audit what’s already available. Your existing software vendors are rapidly adding AI capabilities. You might be able to get significant functionality as part of subscriptions you’re already paying for.

This is where organizations often leave money on the table. They rush to build custom solutions when their existing tools could handle much of what they need.

The Bottom-Up Approach That Reveals Hidden Opportunities

The most important insight from Shruthi’s approach is that workflow mapping needs to be driven by the people doing the actual work, not by management assumptions about how work should be done.

When you involve practitioners in the mapping process, you discover things that don’t show up in org charts or process documents. You learn about informal workarounds people have developed. You understand which “official” processes are actually followed and which ones are routinely bypassed for good reasons.

This bottom-up perspective is crucial because it reveals the difference between optimizing existing processes and reimagining how work could be done better.

Maybe you’ll discover through this process that your “marketing content creation workflow” is actually three different workflows that have evolved organically as the team grew. Maybe some people were using completely different tools and processes to accomplish similar goals. The AI opportunity isn’t just about making any single workflow more efficient, it about creating consistency and eliminating redundant effort.

The Build vs Buy Decision Framework

Once you have your workflow map, you can make much smarter decisions about where to invest in AI solutions. Shruthi’s framework here is elegantly simple: understand what your existing vendors can provide before you decide to build custom solutions.

This approach prevents the “AI cost pileup” that many organizations experience when they build custom solutions for problems that existing tools could solve.

The build vs buy decision becomes much clearer when you can map specific workflow steps to available solutions. Maybe your CRM system can handle AI-powered lead scoring, your content management system can handle AI-assisted writing, and you only need to build custom solutions for the unique integration points between them.

The Value vs Complexity Matrix

Here’s where Shruthi’s framework gets really practical. Not every AI opportunity is worth pursuing immediately.

You want to prioritize based on two factors:

1. the value you’ll get
2. the complexity of implementation

Plot your potential AI applications on a simple matrix:

• High value, low complexity: Do these first
• High value, high complexity: Plan these for later phases
• Low value, low complexity: Consider these as quick wins if you have capacity
• Low value, high complexity: Skip these entirely

This prioritization becomes much easier when you have detailed workflow maps because you can accurately assess both the potential impact and the implementation challenges.

What This Looks Like in Practice

Imagine a mid-sized professional services firm going through this process for their project management workflows.

The initial assumption could be that they need AI for better project scheduling and resource allocation. But when mapping the actual workflows, they might discover that scheduling isn’t the bottleneck, instead it’s the manual effort required to extract insights from project data for client reporting.

So its employees could be spending hours every week pulling data from multiple systems, formatting it in Excel, and creating client presentations. The AI opportunity there is then in automating this reporting process, not in optimizing schedules.

Now, let’s say the existing project management tool already has AI reporting capabilities they aren’t using, and the presentation software has AI-assisted slide creation features. Instead of building a custom solution, they would actually be able to solve 80% of the problem by better leveraging tools they already have. See what I mean?

The Three-Week Validation Rule

Once you’ve identified your AI opportunities and chosen your initial implementations, Shruthi has another insight that can save you months of wasted effort: validate quickly.

“You would be surprised, one of the biggest companies in the world has a validation period of 3 weeks. So if they can do it in 3 weeks, smaller companies can do it in one week.”

This means putting your AI solutions in front of real users doing real work within weeks, not months. If it’s not delivering value quickly, you adjust or move on to the next opportunity.

This rapid validation is only possible when you have detailed workflow maps because you know exactly what success looks like and can measure it quickly.

The ROI Transformation

Organizations that follow this approach see dramatically different ROI profiles than those that start with technology selection. Instead of spending months evaluating AI platforms and then struggling with adoption, they spend weeks understanding their actual needs and then deploy solutions that people immediately find valuable.

The workflow mapping investment pays for itself many times over by preventing expensive mistakes and identifying opportunities that might otherwise be missed.

More importantly, the process creates organizational capabilities that extend beyond any single AI implementation. Teams that understand their workflows at this level become much better at identifying future opportunities and adapting quickly as AI capabilities evolve.

Getting Started This Week

If you want to run your own workflow mapping sprint, here’s how to begin:

• Pick one process that’s both important and frustrating. Don’t start with your most complex workflow. Start with something that’s strategically valuable but small enough to map thoroughly.
• Involve the right people. You need both the strategic perspective and the daily operational reality. Don’t try to map workflows from conference rooms.
• Focus on documenting reality, not ideals. You’re not trying to design the perfect process. You’re trying to understand how work actually gets done today.
• Look for vendor solutions before building custom ones. You might be surprised by what’s already available in tools you’re already paying for.

The goal isn’t to create the perfect process map. The goal is to understand your workflows well enough to make smart decisions about where AI can drive genuine value.

Ready to run your own workflow mapping sprint and identify where AI can drive the biggest impact in your organization? I’d love to help you design an approach that fits your specific context and constraints. Connect with me on LinkedIn to explore how to turn workflow insights into AI transformation wins.