Why Saying “No” Was the Right Outcome: Lessons from Finding the Right AI Use Case

In the age of AI, it’s tempting to assume automation is always the answer. But the truth is, not every process is a net positive from adding this advanced technology, and knowing when not to invest can be just as valuable as building the solution. Our recent engagement with a chemical manufacturer is a perfect example: we didn’t move forward with the second phase of the project, and that was a good thing – for them and for us.

The Initial Problem

This company’s one-person compliance team was consistently bogged down by manual workflows. Customer requests for quality surveys and regulatory documents required hours of effort, searching across SharePoint, ERP systems, and network drives, then uploading files into a supply chain technology one by one. This created delays, increased audit risk, and consumed valuable analyst time that could have been devoted elsewhere.

Where We Started

Rather than jumping straight into development, we began with a discovery engagement focused on mapping the current state and validating assumptions. This phase included deliverables like:

• High-Level Solution Architecture: A diagram and explanatory brief showing the recommended technology components.
• Requirements & Feature Prioritization Matrix: A ranked list of functional and non-functional requirements, tagging each as Phase 1 (POC), Phase 2 (Pilot), or Phase 3 (Production).
• Executive Readout Deck: A board-level presentation summarizing findings, recommended approach, ROI considerations, and decision checkpoints.

By progressing through these iteratively, we were able to show our client value in the work we were doing and visibility into the solution we were envisioning.

What We Learned

Through detailed sessions with the process owner and supporting teams, we uncovered critical insights that shaped the outcome:

• Document applicability rules were complex: Some certificates apply globally, other certificates require item-location logic, and regulatory profiles often map to multiple document types.
• Quality Surveys are highly variable: We couldn’t guarantee compatibility with every document format, and because this would increase project cost, we opted to keep this process manual.
• API limitations: There was no vendor support for automating connection approvals and no multi-document-type tagging, requiring workarounds for regulatory profiles.

These findings were pivotal. They showed that, while we could definitely add value to the process, there were gaps that would be difficult to bridge.

The Outcome

After completing discovery, this company decided not to move forward with implementation. Why? Because the additional investment for the project didn’t justify the cost given the variability and manual oversight still required.

This was a win in disguise, as our client avoided spending money on a solution that wouldn’t pay off at this point in time, but still walked away with clarity, governance improvements, and a roadmap they can revisit when the timing is right. From our perspective, we gained deep insights into where AI truly adds value and where it doesn’t.

Looking Ahead

AI is powerful, but only when applied thoughtfully. Before investing, ask:

• Does automation solve a real bottleneck?
• Is the process stable and repeatable?
• Will the ROI justify the complexity?

For this project, the answer was “not yet” – and that’s okay. Sometimes the most valuable consulting outcome is helping a client avoid unnecessary spending. At Concurrency, we prioritize a discovery-first approach, focusing on business needs rather than pushing technology. This process is critical because it allows us to step into the client’s world and understand their operations as if we were a new hire. By doing so, we collaborate with organizations to reach a logical, well-informed conclusion together.

Is your team spending too much time on manual, repetitive tasks? Let’s explore whether automation could be the right solution, together.