Not long ago, when organizations spoke of digital transformation, they meant grand, sweeping overhauls: enterprise resource planning systems rolled out over years; massive infrastructure migrations; culture change programs that consumed multiple budgets and headcounts. If you weren’t going big-bang, you were behind.
But in 2025, the rules are changing. The winners aren’t those making moonshot bets. They’re stacking small, strategic wins — “small-t” transformations — that compound into lasting advantage. And Generative AI (GenAI) is the catalyst accelerating this shift, not as a flashy experiment, but as a lever for incremental, high-impact change.
From Big-Bang Projects to Continuous Transformation
In the earlier transformation paradigm, organizations committed to multi-year plans: launching new CRMs, redoing entire supply chain systems, moving to cloud at enterprise scale. Success was measured in massive milestones and big spreadsheets. Risks were equally massive — cost overruns, misalignments, user resistance.
But the latest research (MIT Sloan SMR, “Generate Value from GenAI With ‘Small t’ Transformations,” January 2025) shows many organizations are now turning to targeted, iterative change.
Some findings indicate that:
- Out of studies of 21 large companies, those that sequence use-cases with immediate value and build incrementally along a risk slope are gaining traction.
- Rather than designing for radical re-designs first, they begin with automating tasks across roles (e.g., summarizing, drafting, information retrieval), then move to more complex problems.
These “small-t” moves are helping organizations build resilience, agility, and foundations (data pipelines, security, cultural readiness) that make bigger shifts both feasible and less risky.
The Power of “Small-t” Wins with some interesting data insights:
- According to the State of AI in Business 2025 – GenAI Divide (Project NANDA, MIT, etc.), while organizations have invested US$30-40 billion into GenAI initiatives, only about 5% of custom or enterprise-grade AI tools reach production with measurable business value. mlq.ai
- Generic tools like ChatGPT and Copilot are widely adopted; over 80% of organizations have at least explored or piloted them. But actual deployment beyond pilot for custom tools is rare. mlq.ai
- Pilot-to-production drop-off is steep: many organizations have dozens of demo or proof-of-concept projects, but only a tiny fraction yield sustained ROI. mlq.ai
- Investment bias: much of spending goes to visible front-office and customer/marketing functions (where metrics are easier to see), but often the highest ROI shows up in back office, operations, and finance where efficiency gains, cost savings, and workflow automation can be very large. mlq.ai
Imagine a customer-support org that starts by using GenAI to auto-draft standard responses or summarize past tickets. The first use case doesn’t change jobs or org structure; it just saves 10-15% of time for agents. Then, they layer in sentiment analysis to flag urgent cases, integrate with CRM so summaries are auto-filed, and eventually build a memory-augmented assistant agent that learns from prior tickets. Over time, this approach yields measurable improvements in response times, customer satisfaction, and reduces error rates — all without rolling out a new monolithic system.
Or a manufacturing firm: first using GenAI to analyse maintenance logs and detect patterns of common failures (task level). Then building dashboards where team leads can query this automatically. Then using predictive tasks for scheduling maintenance (team level). Each step small, controlled, but over 12-18 months, transforming how maintenance is managed, reducing downtime, and saving costs.
The Mindset Shift: Transformation as a Pipeline
Instead of treating transformation as a project (with a clear start and end), leading organizations are treating it as a pipeline: a sequence of use cases, each building on the previous, gradually increasing in complexity and impact.
Some key dimensions of this mindset:
- Risk Slope: start with low-risk, high-reward use cases (easy tasks, well understood, minimal dependencies), then move up towards higher stakes (role- or team-level transformations, customer-facing or product transformations). MIT’s framework suggests three categories:
Tasks across roles → Role/team-specific tasks → Products & Customer experience. small-t-transformations
- Immediate Value + Scaffolding: each use case should deliver something tangible (efficiency, cost savings, quality, etc.), while also building infrastructure (data architecture, security, monitoring) that enables the next use case.
- Governance, Culture, Metrics from Day One: small pilots are often orphaned. The successful ones have clarity on metrics, accountability, decision paths for scaling, and a culture that embraces experimentation and learning.
From Pilots to Platforms — Avoiding the Pitfalls : Major trap many companies fall into is having lots of glittering pilots that never scale. The shift to “small-t” is helpful, but you still need to design each pilot as a building block, not just a showpiece.
Where Catallyst Comes In
At Catallyst, we help organizations think through the opportunities and challenges of GenAI adoption and identify where “small-t” transformations can deliver meaningful impact.
We support organizations by:
- Framing the Transformation Mindset: Highlighting how incremental, strategic AI use cases can compound into lasting advantage.
- Providing Insights & Approach Notes: Offering a structured perspective on potential pilots, risk slopes, and areas where AI can create measurable value.
- Guiding Strategic Conversations: Helping leadership teams explore where GenAI initiatives can align with business priorities and capabilities.
Our role is to spark ideas, guide planning, and help design initiatives that set the stage for scalable impact — enabling organizations to move from small bets to big shifts without committing to grand, high-risk projects upfront.
Crossing the “GenAI Divide”
A recent report, The GenAI Divide: State of AI in Business 2025 (Project NANDA, MIT) provides further insight into how many companies are on one side or the other — stuck in adoption, but not transformation. mlq.ai
- Only ~5% of enterprise-grade custom GenAI tools are achieving measurable, scaled deployment.
- Generic tools like ChatGPT/Copilot get much higher usage, but often only for individual productivity, not embedded in core workflows.
- Leading organizations are the ones that prioritize learning systems — tools that retain memory, adapt to feedback, integrate with existing workflows.
So the “small-t” approach helps bridge that gap: you start with what works, then build up capacity to cross from pilot to production, from department-level use to enterprise-scale.
Big transformations used to be defended in boardrooms with giant budgets and Gantt charts. Now, they’re built in increments — one use case, one capability, one habit at a time.
So ask yourself:
- Are you still betting everything on one big leap — hoping the “transformation” will happen all at once?
- Or are you building momentum through a sequence of wins that compound into unstoppable change?
Because in this era, the future of digital transformation isn’t big-bang. It’s small-t — and it’s already reshaping the enterprise.
References :
MIT Sloan Management Review (2025). Generate Value From GenAI With “Small-t” Transformations. https://sloanreview.mit.edu/article/generate-value-from-gen-ai-with-small-t-transformations/