More AI tools will not fix a broken workflow

The logic seems sound: buy a few AI tools, roll them out across departments, and the gains will follow. It is the assumption behind thousands of AI purchasing decisions made by Operations Managers, Chief Executives, and IT Managers across the UK right now. And it is understandable. Vendors promise productivity improvements. Case studies from large enterprises look compelling. The tools themselves are genuinely impressive in isolation. The problem is that isolated is exactly where most of them stay.

A recent global study of AI in customer experience, covered by IT Brief, found that while the majority of organisations now use AI tools, only about 15% combine agentic AI with cross-departmental orchestration to resolve customer needs end-to-end. Just 5% reported being able to quantify AI's impact on business outcomes at all. The tools are there. The results are not.

This article explains why that gap exists, what it costs in practice, and what a more grounded approach looks like.

Why the "just add AI" assumption feels reasonable

The myth has a strong foundation. AI tools genuinely do work well in specific, contained tasks. A chatbot handles common queries. A writing assistant speeds up email drafts. A scheduling tool reduces back-and-forth. Each of these produces a visible, local win.

The problem is that local wins do not automatically aggregate into business impact. A chatbot that cannot pass context to your CRM means a customer explains their issue twice. A writing assistant that sits outside your approval workflow creates a version-control problem. A scheduling tool disconnected from project management produces meetings with no follow-through.

The gain stays trapped inside the tool. Businesses feel busy with AI without being measurably better for it.

The myth this research exposes

The assumption being tested here is straightforward: "If we roll out a few AI tools across the business, we will automatically see big gains, even if they are all separate systems."

The evidence does not support it. The same research that found near-universal AI adoption also found that disconnected systems and legacy infrastructure were among the most commonly cited barriers to realising value. Over half of respondents reported limited trust in AI decisions, which is a predictable outcome when AI outputs cannot be audited or traced across a process.

Buying more licences does not solve a handoff problem. It compounds it.

What this costs UK SMEs in practice

For Operations Managers, the cost is invisible duplication. Teams in separate departments use different AI tools that do not share data. The same information gets re-entered, re-summarised, and re-processed at each stage. The efficiency gain from the tool is consumed by the coordination overhead around it.

For Chief Executives, the cost is strategic opacity. If only 5% of organisations can quantify AI's impact on outcomes, most leadership teams are making AI investment decisions without a feedback loop. They are spending on tools they cannot evaluate.

For IT Managers, the cost is technical debt at speed. Each unconnected tool becomes a separate integration problem later. Security posture fragments. Data flows become harder to audit. What looks like low-risk incremental adoption is actually accumulating architectural complexity.

The Corporate Vision analysis of common AI missteps frames this clearly: the failure mode is not choosing the wrong tools. It is treating tools as destinations rather than components.

What the evidence actually supports

Three findings from the research are worth holding on to.

1. Orchestration is the differentiator. Only 15% of organisations combine agentic AI with cross-departmental orchestration to resolve needs end-to-end. That minority is not using more tools. They are using connected tools. The competitive advantage lies in the handoffs, not the headcount of applications.

2. Measurement is almost entirely absent. Just 5% can quantify AI's impact on business outcomes. This is not a technology problem. It is a planning problem. Organisations that define KPIs before deployment, such as first-contact resolution rate or average handling time, create the conditions for evidence-based iteration. Those that do not are flying without instruments.

3. Trust follows traceability. The finding that over half of respondents report limited trust in AI decisions points to a specific failure: AI outputs that cannot be explained or traced. When a decision emerges from a black box, people reasonably distrust it. When a workflow has quality gates and structured handoffs, the output can be checked. Trust follows from that, not from the AI's underlying capability.

CFOtech UK research on SME AI adoption reinforces this: smaller businesses that take a more deliberate, phased approach to AI report stronger confidence in their outcomes than those that move fast across multiple tools simultaneously.

A realistic approach for SMEs

The intervention the research points toward is orchestration over accumulation. That means a different kind of question at the start of any AI project.

Instead of "which tools should we buy?", the more productive question is "where does a process currently break down between departments, and what would it take to connect those steps?"

In practice, that looks like mapping where AI already touches your workflows. It means choosing tools that integrate with your existing CRM, ERP, or project management setup rather than sitting alongside them. It means defining what success looks like before deployment, not after.

This is harder than purchasing a licence. It requires someone to own the workflow design. It requires upfront configuration rather than out-of-the-box deployment. And it requires a willingness to measure what is actually happening, not just what the tool dashboard reports.

For SMEs that want external support with this, gecco's AI Agents combine specialist AI Assistants and no-code Automations into end-to-end workflows with quality gates and structured handoffs. Rather than isolated tools sitting in separate departments, Agents are designed to own a process from trigger to outcome, sharing context across steps. This directly addresses the core bottleneck the research identifies: cross-departmental orchestration, not tool volume, is what produces measurable business impact.

What to be honest about

None of this is fast. The organisations in the 15% who have achieved genuine cross-departmental orchestration did not get there in a single project cycle. They started with one connected workflow, measured it, and expanded from there.

There is also a meaningful skills dimension. Connecting tools requires someone who understands both the business process and the technical integration. For many SMEs, that person does not currently exist in-house. Outsourcing the build without building internal understanding of what was built creates a different kind of dependency risk.

And the measurement challenge is real. Defining a KPI for AI impact sounds straightforward. Getting clean, comparable data before and after implementation is often not. Organisations that skip this step do not avoid the measurement problem. They just lose the ability to answer it.

Where to start if this reflects your situation

If your team is weighing up whether your current collection of AI tools is actually delivering, or whether you are about to make another purchase that will sit alongside rather than connect with what you already have, the AI Readiness survey is built to surface exactly that question.

Take the AI Readiness survey. You will get access to 65+ free resources and a custom AI Readiness report. We then offer a free 30-minute showcase to walk through your results.

Most organisations treat AI tools as point solutions rather than orchestrated systems. The bottleneck is not the technology itself but the integration and handoff logic that connects assistants and automations into coherent workflows. If you want help scoping and building that connected layer, gecco's AI Agents service is designed for exactly that.

Website · LinkedIn · Case Studies · Newsletter