Shan

Teams frequently assume that the bottleneck lies in model development speed or data availability, because these are the most visible pain points. However, in several large-scale MathCo engagements, the real constraint has proven far more subtle. A concrete example is MathCo’s work on demand forecasting for a global CPG client. Multiple teams were involved: data engineering, data science, business analysts, and the client’s supply planning team. Early retrospectives pointed to “slow model iterations” as the bottleneck. Human judgment, based on anecdotal delays and team feedback, led to adding more data scientists — which did not improve cycle time or forecast adoption. When we instrumented the end-to-end process and applied an AI-driven analysis using Nuclios-style monitoring and simulation, a different constraint emerged. By analyzing timestamped data across stages — data refresh, model runs, approvals, planner overrides, and decision execution — the AI detected that planner override variability and approval batching were creating downstream congestion. Although the model ran daily, planners approved changes only twice a week, causing inventory decisions to lag. The true constraint was not technical capacity, but policy-driven human workflow variability. AI outperformed humans here in three key ways: Pattern detection across long time horizons The AI identified recurring lag patterns that were invisible in weekly reviews — such as how forecast accuracy dropped after extended approval gaps. Simulation of counterfactual scenarios By simulating “what-if” scenarios (e.g., daily auto-approval for low-risk SKUs), the AI quantified throughput gains and service-level improvements before any policy change was made. Separating symptoms from root causes What appeared to be slow model performance was actually a downstream constraint amplified by workload batching — something human intuition consistently misattributed. However, AI still struggled in areas where context and organizational nuance mattered. It could not independently assess why planners resisted automation — concerns around accountability, incentive structures, and trust in AI recommendations required human judgment and stakeholder conversations. AI surfaced the constraint, but humans had to decide how to resolve it sustainably.

In my view, AI demonstrates pattern-based creativity, which feels original in output, but is still recombining what it has learned rather than inventing from nothing. A relevant example comes from my experience at MathCo, where we integrate GenAI into analytics operations. We recently experimented with using GenAI agents to craft response for a supply chain analytics proposal. Traditionally, our consulting team writes these proposals manually — identifying business pain points, mapping them to MathCo’s solutions (digital twin, forecasting, root-cause analysis), and designing a storyline that resonates with the client’s industry context. When we prompted the AI to draft the first version, it generated a structured narrative including: A compelling opening about volatility in global logistics A recommended solution flow using our capabilities A future roadmap with AI-driven scenario planning Custom messaging tailored to C-suite priorities The output was polished, logically ordered, and context-aware. However, upon reviewing it, we realized the AI didn’t invent a new solution or narrative — it recombined common language patterns from prior proposals, case studies, and public knowledge. The creativity emerged in how it blended existing components quickly into fresh-seeming messaging, but it did not show imagination that a consultant brings — for example, linking the client’s SKU variability issue with a novel optimization approach we had tried in an unrelated CPG project. So, while the AI accelerated structuring and ideation, the final storyline and value articulation still required human creativity to introduce nuance, analogy, and risk-reward framing. Conclusion: AI can appear creative, but in our domain, it mainly remixes learned patterns to generate useful starting points. True creativity — connecting ideas across contexts still comes from humans. Instead of replacing creativity, AI augments it.

At MathCo, our culture is built on three fundamentals — ownership, transparency, and continuous learning. As we embed AI deeper into our work through platforms like Nuclios, it’s already influencing how teams think, collaborate, and make decisions. On the positive side, Nuclios’ auto-documentation, lineage tracking, and explain-ability features give everyone — from new analysts to senior consultants — clear visibility into how models behave and why certain decisions were taken. Knowledge that earlier sat with a few experts is now across employees, helping teams learn faster and make more confident decisions. With automated prototyping, synthetic data generation, and quick scenario testing, teams have cut down on their experimentation cycle time. This encourages a “test, learn & refine” mindset. But there are risks. As AI handles more data prep and QC, teams may begin to trust outputs without questioning them — weakening accountability. There’s also the danger of insights becoming a “black box” if explain-ability isn’t consistently enforced. To ensure AI strengthens our culture, leaders need to be intentional: Mandate explain-ability over speed in Nuclios workflows. Reward validation & questioning over blind acceptance of AI recommendations. Keep humans in the loop for validation and client insights. Build AI-literacy pathways so everyone understands limitations and biases. Implement Responsible AI checkpoints into daily work.