Nigel Dews
Managing Director, Restore Information Management
With nearly nine in ten organisations now using AI in at least one business function, the question for 2026 is no longer whether to adopt AI. The real question is how deeply it should be embedded into everyday operations.
The conversation around AI is changing. The focus is shifting away from generic tools and broad experimentation toward practical applications that quietly improve governance, data quality, retention and optimisation in the background. Organisations are increasingly looking for AI that delivers measurable business value, improving productivity, reducing inefficiency and helping employees make better decisions.
Yet at the same time, in conversations with customers and industry leaders, I am seeing growing caution. CIOs are becoming more sceptical of overhyped AI promises, particularly where accuracy, security and accountability remain unclear. The excitement around AI is still there, but so are the questions.
- Can we trust the outputs?
- Will it improve productivity?
- How do we maintain compliance?
- And perhaps most importantly: are we actually ready for it?
Because while organisations are moving quickly to deploy AI, many are overlooking a fundamental truth. AI is only as intelligent as the information it can access.
The Impact of AI on Document Management
For years, poor document management created inefficiencies that organisations simply learned to tolerate. Employees spent time searching for information, recreated documents they could not find, relied on colleagues to locate files, and worked around fragmented systems holding multiple versions of the truth.
AI changes the equation. Employees may be able to work around poor information management. AI cannot.
In the age of AI, an old technology principle suddenly feels more relevant than ever: Garbage in, garbage out.
The New Business Risk
If organisations feed AI fragmented, duplicated or poorly structured information, they should not be surprised when the outputs become inconsistent, unreliable or difficult to trust. And this is the point many organisations are only now beginning to realise. AI is not creating document management problems. It is exposing problems businesses have tolerated for years. Turning tolerated inefficiency to business risk
Historically, poor document management was viewed as an operational inconvenience. Employees wasted time searching for files. Information sat across shared drives, email inboxes and disconnected platforms. Search functionality was unreliable. Metadata was inconsistent. Naming conventions varied wildly, and multiple versions of “final” documents somehow coexisted.
Most businesses adapted.
If the search failed, employees asked a colleague.
If they could not find a document, they recreated it.
If several versions existed, they made their best judgment.
People are remarkably good at navigating organisational complexity.
But AI introduces a new expectation, immediate access to trusted information. Employees increasingly expect systems to provide clear answers, surface relevant knowledge instantly and help them work more efficiently. The problem is that many organisations are trying to layer AI over information environments that were never designed to support it.
The Rising Expectation for Trusted, Accessible Information
Talking to customers, I still regularly hear frustrations around basic findability and trust in information. In some cases, employees do not trust search results today, let alone AI-generated outputs. And that is an important warning sign. Because if employees struggle to find reliable information now, why would AI suddenly succeed?
Why structured information matters
One of the biggest misconceptions I encounter is the assumption that AI somehow fixes poor information. It does not. AI amplifies whatever already exists. Good information leads to useful outputs. Poor information leads to unreliable outputs.
When I talk about structured information, I am not referring to something overly technical or unrealistic. In practical business terms, structured information means information that is organised, trusted and accessible.
It means:
- A clear source of truth
- Consistent metadata and taxonomy
- Proper version control
- Governed permissions
- Searchable repositories
- Clear ownership of information
These may sound like document management fundamentals because they are. But they are also becoming the foundations of successful AI.
Imagine an employee asking an AI assistant for the latest policy document, customer contract template or compliance procedure.
What happens if multiple versions exist across different systems?
What happens if an outdated document is incorrectly labelled?
What happens if permissions are inconsistent or key information sits inside personal folders?
The issue is not necessarily that AI gets things wrong. The issue is that AI retrieves information based on what already exists. And if what exists is fragmented, duplicated or poorly organised, confusion quickly follows. AI does not eliminate information chaos, it operationalises it.
The hidden costs AI is exposing
As organisations move from experimentation to implementation, the hidden cost of poor document management is becoming harder to ignore.
Poor trust in AI outputs
Trust is essential for adoption. If employees receive inconsistent answers or outdated information, confidence disappears quickly. Many failed AI initiatives are blamed on the technology itself when, in reality, the problem often lies with the quality of the information behind it. Once trust is lost, adoption stalls and so does return on investment.
Productivity gains that never materialise
AI is supposed to improve efficiency. Yet poor information management can create an entirely new layer of work. Employees find themselves validating AI responses, checking whether information is current and manually confirming sources. Rather than removing inefficiency, poor information foundations simply shift it elsewhere.
Greater governance and compliance risk
AI increases visibility into organisational information including information that may not be properly governed. Weak permissions, outdated policies, unmanaged repositories and inconsistent retention practices become more problematic when surfaced through intelligent systems. For regulated organisations especially, this creates real concern. The message I hear consistently from customers is clear: AI must strengthen compliance, not weaken it.
Missed opportunities to retain knowledge
Many organisations see AI as a way to preserve institutional knowledge. But AI cannot surface knowledge that has never been captured. If expertise sits in inboxes, desktop folders or the minds of experienced employees, AI has little to work with. The organisations gaining the greatest value from AI are often those that have already invested in making knowledge structured, searchable and accessible.
Before scaling AI, fix the foundations
None of this means organisations should slow down AI adoption. Far from it. AI has enormous potential to improve document management itself from intelligent classification and summarisation to improved search and better knowledge discovery. But organisations need to approach AI practically.
Before investing further, leaders should ask some honest questions:
- Can employees find trusted information today?
- Do we have a clear source of truth?
- Is our information structured and governed?
- Are permissions and compliance controls clear?
- Have we addressed outdated or redundant content?
If the answer to several of these is “not yet”, the challenge may not be AI readiness. It may be information readiness.
Because while AI technology will continue to evolve rapidly, one thing is becoming increasingly clear from customer conversations and market research alike:
- The organisations that succeed with AI will not necessarily be the ones that move fastest.
- They will be the ones that first build trusted information foundations.
- In the rush to adopt AI, organisations risk treating it as a technology deployment exercise.
In reality, it is an information challenge. And in AI, as in document management, the same principle still applies. Garbage in, garbage out.
More insights
