Many organizations are seeing **faster individual output** from AI, but not **better business outcomes** at the team or company level.

According to Atlassian’s 2026 State of Teams report:

• 89% of executives say AI is making their teams work faster.
• Yet only 6% can point to clear, organization-wide ROI.

The core issue is what Atlassian calls the **“AI Fragmentation Tax”** – the hidden cost of speeding up individuals while leaving team workflows disconnected and analog. Today:

• 85% of knowledge workers use AI, but only 29% have it embedded into daily workflows.
• 70% of workers say their processes aren’t optimized for AI.
• 87% say there’s no time to coordinate because everyone is stuck in execution mode.
• 84% report unclear or conflicting goals.

The result is a productivity mirage: AI helps people draft emails, slides, or concepts faster, but it also creates more misaligned work, more handoff friction, and more rework. Across the Fortune 500, this fragmentation tax is estimated at **$161 billion annually**.

What to do instead
To move from fragmented speed to measurable ROI, leading teams:

• Use AI to coordinate work, not just accelerate tasks.
• Redesign workflows so human and AI work is connected end-to-end.
• Invest in shared context and data foundations, not just more tools.

When AI is treated as a **team catalyst** rather than a personal productivity add-on, organizations start to see clearer, scalable returns.

The **AI Fragmentation Tax** is the cost you pay when AI speeds up individuals but leaves teams misaligned and workflows disconnected.

In practice, it looks like this:

• A designer uses AI to generate 50 campaign concepts before lunch.
• Those concepts flood the engineering team’s intake queue.
• Most of them don’t match sprint priorities or strategy.
• Engineering slows down, bottlenecks appear, and coordination debt grows.

AI didn’t fix the problem – it amplified misalignment.

This pattern is widespread:

• 84% of workers report unclear or conflicting goals.
• 69% of workers say their data foundations aren’t AI-ready, so they don’t fully trust AI outputs.
• About half of workers spend more time **cleaning up AI-generated work** than they would have spent doing it from scratch.

There’s also a growing capability gap:

• Executives are 84% more likely to invest in new tools than in the people using them.
• Workers are 90% more likely to fear being unprepared for AI than being replaced by it.
• Nearly 7 in 10 say they lack adequate AI training.

How leading teams reduce this tax
The top-performing 14% of teams focus on three pillars to cut fragmentation:

1. Context – A shared, centralized knowledge base so everyone (and every AI agent) works from the same source of truth. For example, software firm Datasite reduced **216 OKRs** down to **3 company-wide objectives and 12 key results**, supported by real-time, self-serve dashboards.
2. Workflow – Redesigning how work flows between humans and AI. Cisco, after building a System of Work with Rovo, saw:
   • 40x faster program management reporting,
   • 24x faster status updates, and
   • 54% reduction in tooling total cost of ownership.
   As Cisco’s VP of engineering Jason Andrews puts it: “Transform, don’t migrate… you really have to rethink the way you want to work and operate.”
3. Culture – Treating AI as a teammate, not just a tool. At payments firm Riverty, among 3,000 Atlassian users, over **50%** are active Rovo users, building their own agents with autonomy. They report faster answers for both employees and customers, and stronger connection across teams.

When you address context, workflow, and culture together, you start to **reimagine** how teams coordinate – and the fragmentation tax shrinks instead of growing.

High-performing teams don’t just add AI on top of existing processes. They **rethink how work gets done** across three pillars: context, workflow, and culture.

1. Build shared context
They start by making sure everyone – and every AI system – is working from the same information.

• Create a centralized, searchable knowledge base.
• Align goals and metrics so AI-generated work maps to real priorities.

Example: **Datasite** rationalized **216 OKRs** down to **3 company-wide objectives and 12 key results**, replacing fragmented, static reporting with real-time, self-serve dashboards. This kind of shared context makes teams **12x less likely** to suffer from conflicting priorities.

2. Redesign workflows for human + AI collaboration
Instead of “lifting and shifting” old processes into AI tools, they redesign workflows around how humans and AI agents should interact.

• Map end-to-end workflows and identify where AI should assist, automate, or orchestrate.
• Connect tools so AI output flows directly into the next step, not into disconnected silos.

Example: **Cisco** built a System of Work with Rovo and saw:

• 40x faster program management reporting,
• 24x faster status updates, and
• 54% reduction in tooling total cost of ownership.

These teams are **13x more likely** to be aligned on strategy because workflows and AI usage are designed together, not bolted on.

3. Shape a culture where AI is a teammate
Top teams encourage people to treat AI as a collaborative partner, not just a shortcut.

• Give employees autonomy to build and refine their own AI agents.
• Invest in training so people feel prepared, not threatened.

Example: At **Riverty**, among 3,000 Atlassian users, over **50%** actively use Rovo and build their own agents. According to product owner Andrei Tuch, this leads to faster answers for teams and customers, and stronger connections across the organization. Teams that adopt this mindset are **20x more likely** to feel connected to colleagues.

What this means for leaders
To follow the path of these top 14% of teams:

• Stop treating AI as a personal productivity perk and start treating it as a **team catalyst**.
• Invest as much in your **Human–AI playbook** (training, workflows, governance) as you do in the underlying models and tools.
• Focus on moving the whole team in the same direction, not just moving individuals faster.

By doing so, you can begin to **reshape** how work happens, reduce your AI Fragmentation Tax, and unlock the value that’s currently being left on the table.