71% of organizations already provide AI tools to their teams. Yet only 15% of leaders believe their innovation keeps pace with the advance of artificial intelligence.
This gap reveals a fundamental problem: we're applying traditional change strategies to a context that demands something completely different. The pace at which AI evolves doesn't allow for 18-24 month transformation cycles. We need organizations that don't adapt once, but are continuously adaptive.
The problem: an outdated playbook for a new world
Classic change-management methodologies —with their phases of diagnosis, planning, implementation and consolidation— were designed for one-off transformations with a defined end state. But AI doesn't work that way.
"Over the last 12 months, we've seen teams go from using basic Copilot to deploying autonomous agents that review code, generate documentation and propose architectures. The state of the art is redefined every quarter."
To respond to this pace, organizations need three fundamental elements:
- Long-term transformative visions that guide without rigidity
- Redesigned systems for human-AI collaboration
- Empowered teams that can experiment and "hack" their own jobs
The three-horizons framework: from productivity to disruption
At onext, we work with a three-horizons framework that helps teams implement AI progressively but ambitiously. Each horizon requires a different mindset and different actions:
Horizon 1 - Immediate productivity
Goal: Operational efficiency and reduction of repetitive tasks.
6-month actions:
- Roll out GitHub Copilot or equivalent across the whole development team
- Automate unit test generation with AI
- Create technical documentation assistants
- Measure: % of autocompleted code, onboarding time, test coverage
Horizon 2 - Competitive differentiation
Goal: New value propositions and service models.
6-month actions:
- Identify an AI use case that directly impacts customers
- Build an MVP of an AI-enabled product/feature (e.g. recommendations, assistants)
- Experiment with autonomous agents for critical internal processes
- Measure: NPS of the AI feature, user adoption, ROI of the experiment
Horizon 3 - Business reinvention
Goal: Create new categories and industries.
Required mindset: "If we don't build what kills our current product, someone else will."
This horizon requires investment in R&D, teams dedicated to exploration, and tolerance for failure. Few organizations are here, but it's where disruptions are born.
Building an adaptive workforce
Even though 61% of companies offered AI training in 2024, 63% of executives cited "inadequate skills or resistance to change" as the main barrier to adoption.
The problem isn't one-off training. It's creating a culture where experimenting with AI is part of the daily job, not a special project.
Three levers to build adaptability
- Skills-based architecture: Identify key capabilities (prompt engineering, model evaluation, API integration) and create internal communities of practice.
- Redesigned leadership: Technical leaders must be visible experimenters. If the CTO doesn't use AI day to day, no one else will do it consistently.
- Low-value automation: Free up time by eliminating repetitive work. The time gained must be reinvested in exploration, not in more tasks.
A philosophy of continuous reinvention
Adapting to AI isn't a project with a closing date. It's a shift in organizational mindset: from "implement and consolidate" to "explore and evolve".
This means:
- Annual budgets with a % reserved for experimentation without guaranteed ROI
- Team OKRs that include "AI hypotheses tested" as a metric
- Rituals for sharing learnings (not just successes, failures too)
- Explicit permission to "destroy" processes that work if there's an AI alternative that's 10x better
Is your organization adapting, or standing still?
If, after reading this article, you realize your team has AI tools but no one is really transforming the way they work, you're not alone. And it's not too late.
At onext, we've guided 12 transformations of technical teams that went from "using Copilot occasionally" to multiplying their velocity by x5-10 with AI integrated into their workflow. Without stopping deliveries. Without months of planning.
The first step is always the same: an honest conversation about where you are and where you need to be.
Written by the technical team at onext, a Spanish applied-AI consultancy. It reflects the team's practice in transforming development teams, cloud, DevSecOps and quality: 12 teams transformed and 0 sprints lost.