Angela Wick talking about Redefining the #BadAssBA in the Age of AI.
She’s just explained something that every BA needs to understand about AI – and it’s a game-changer for how we design systems. 🎯
With a little help from otter and claude, here’s my takeaway on this slide:
🍩Deterministic vs Probabilistic
Historically, our systems were deterministic (binary). If-then logic. You put in THIS input, you get exactly THIS output. Every single time. Predictable. Prescriptive.
With AI-based processes, things become probabilistic (non-deterministic). Same input can give you different outputs.
You see this with ChatGPT – ask it the same question twice, get different responses. That’s why people sometimes say “AI isn’t accurate” – but it’s not trying to be binary. It’s looking for patterns, predicting the most common response.
Google search? Binary Boolean strings looking for matches. Google with AI? Interpreting and predicting.
🍩The BA Sweet Spot 🎯
Angela nailed it: We need to determine which steps benefit from being probabilistic versus deterministic.
Some examples:
Deterministic: Looking up an account number. I need the EXACT account number from the database, not a guess at what the probable account number might be.
Probabilistic: Chatbot responding to “How are you doing today?” – Do we want the same exact response every single time? Or something more human-like that varies? No one wants to feel like they’re talking to a robot.
But here’s where it gets interesting: “Give me accounts with these characteristics” – is that deterministic or probabilistic? It depends on whether the data is perfectly mapped or if we’re looking for patterns.
Customer asks “What’s my outstanding balance?” – Deterministic. Customer asks for recommendations or has a conversational query? – Probabilistic.
🍩Our Role with Agentforce
With agent orchestration, we’re going to need to analyze this for our organizations and clients.
Which steps and interactions are a fit for probabilistic vs deterministic?
Where does the human stay in the loop?
How do we guide AI? Measure it? Monitor it? Supervise it?
What are the risks, and how can we leverage AI to mitigate them?
Good analysis strikes the right balance between the two. That’s the elevated BA role.
With my lovely co-lead Vanessa Grant running our monthly
#BadAssBA meetup.
