Enhancing our Human-Centric Workflows

Introduction:

In my LinkedIn post: “AI isn’t replacing jobs, it’s handling tasks“, I stated that as developers and systems integrators, we are used to automating standardised, systematised tasks. With the rapid development of AI capabilities, what has changed is our ability to take on ambiguous, non-standardised ones.

An ex. colleague (from our Capgemini days over ten years ago) and a friend, Rachel Ragan, asked me to write an article with more details on these ambiguous, non-standardised tasks and share how AI addresses them.

So, first of all how do we recognise an ambiguous, non-standardised task? They are all around us but don’t look the same. How I would define them is that they are tasks that:

1. Lack clear rules or steps: There isn’t a straightforward, repeatable process to follow.
2. Depend on human judgment: The task requires interpreting unstructured information, such as free-text notes, emails, or conversations.
3. Vary in context: Each task instance may be different and need human reasoning or creativity to handle it effectively.

These tasks may seem relatively routine to us as humans, but they usually take a lot of time, and we can’t get rid of them or automate them… until now!

The Challenges of Ambiguous Non-Standardised Tasks (The CAN’TS):

1. Rule-based systems need explicit programming for every potential scenario, making them impractical or impossible for dynamic, evolving tasks.
2. Traditional systems EXCEL at processing structured, well formatted, data (e.g., spreadsheets), but they struggle to process partial data or unstructured data, particularly free-text notes, images of documents and audio recordings.
3. Rule-based systems rely on predictability; high variability leads to frequent errors or the need for manual intervention. If there is also subjectivity in decision-making, traditional automation cannot make those subjective evaluations.

Case Study: Salesforce CRM Solution

Abstracting away from a specific project so as not to focus on the specifics of a single customer case study and in turn, provide something generalised enough that you may be able to identify and recognise enough similarity in your situation, I will describe a scenario I have worked on quite recently;

• The Situation: CRM Users spend 70-80% of their time on a single task – manually matching two ‘contact’ types. Many attempts have been made to automate the matching process using the structured data of each contact record, but every attempt has failed to reliably identify suitable matches
• The Problem: Users manually match two ‘contact’ types based on details found in many unstructured case notes, not the structured data. This requires significant time and effort from the CRM user, who must work through a comparative process on a record-by-record basis. They find themselves going “back and forth” between records until they identify a match.
• Solution: Use AI to analyze case notes of all contacts, identify the relevant data to matching that is unique in each pairing, and identify the most suitable matches. Have the AI explain why they are a good match or identify what additional information may be required to confirm with more certainty the suitability of a match based on what is already known but what has also been identified as unknown.
• Outcome: AI eliminates the most laborious and massively time-consuming aspect of the job by addressing the most time-consuming and complex task. This enables CRM users to focus on their work’s strategic and relational outcomes.

The Broader Implications:

The value AI delivers in handling ambiguous, non-standardised tasks extends far beyond CRM systems. Similar challenges exist across industries, and the solutions share a common theme: augmenting human effort rather than replacing it. Here are a few examples:

1. Legal Research: Legal professionals often sift through mountains of case law, contracts, and documentation to identify precedents or relevant clauses.
2. Medical Diagnoses: Healthcare providers must interpret unstructured patient data such as doctor’s notes, imaging results, and test reports to form diagnoses.
3. Content Summarization and Generation: Teams in media, marketing, or research fields often need to distil lengthy reports or discussions into concise, actionable insights.

Each one of these examples illustrates a recurring theme: AI handles the repetitive, time-consuming, or highly nuanced parts of a task, empowering these industry professionals to focus on their core strengths—strategic decision-making, creative problem-solving, and building relationships. Far from being a replacement, AI becomes a collaborator, enhancing productivity, engagement, and job satisfaction.

Rachel, that feels like the starting point for yet another article!

Our role in the Agentive Enterprise:

The future belongs to those who embrace this collaboration, adapting to and thriving within an agentive enterprise—an ecosystem where humans and AI work together seamlessly to achieve superior outcomes. We can do this for ourselves, but we should also build solutions which solve these problems for our colleagues and our clients.

Conclusion:

AI is not here to replace us but to augment what we do best. By taking on the most time-consuming, repetitive, and ambiguous tasks, AI allows us to refocus our energy on the work that truly matters—creative thinking, strategic problem-solving, and building meaningful relationships. It’s a partnership, not a replacement, and one that has the potential to significantly enhance productivity, engagement, and overall quality of work.

I’d love to hear your thoughts and experiences. Have you encountered tasks in your field that are difficult to automate but could benefit from AI? How do you see AI fitting into your work? Rachel, I look forward to hearing from you!