Alteryx research suggests enterprise AI investment is accelerating faster than organisations’ ability to embed the business rules, data access, and operational knowledge needed to make systems reliable at scale.
Its 2026 IT Leader Research surveyed 1,400 technology leaders using AI across North America, Europe, the Middle East, and Asia-Pacific. Eighty per cent expect AI spending to increase over the next two years, while 69% report moderate or significant returns from investment already made.
The stronger constraint is execution. Some 77% of respondents said business context is critical to producing accurate and relevant AI outputs, but 53% said their organisation struggles to translate that context into AI systems and workflows.
Business context can include the rules, definitions, exceptions, thresholds, and operating assumptions employees apply when making decisions. Much of that knowledge sits in spreadsheets, policy documents, emails, legacy systems, or experienced employees rather than in a structured form that software can apply consistently.
The issue becomes more demanding when AI systems are connected directly to business processes. An assistant producing a draft document can be reviewed before use. An automated or agentic system making decisions across finance, customer service, procurement, compliance, or operations needs stronger access to the rules that determine an acceptable outcome.
The research also points to continuing constraints around data. Only 18% of organisations said business users had fully self-service access to cloud data, while many respondents continued to rely on IT or data teams for routine access and analytics.
This creates an organisational bottleneck. Employees closest to a commercial or operational process may understand the meaning of the data and the exceptions governing decisions, but lack the technical access needed to build or validate AI workflows. Technical teams may control infrastructure while lacking detailed knowledge of how individual business processes operate.
Alteryx found that 71% of respondents believed AI initiatives were most successful when IT and business teams worked closely together. Two-thirds said AI and agent-based systems were most productive when managed within the line of business, yet AI strategy and delivery remained concentrated primarily inside technical functions.
The survey was carried out by Coleman Parkes in April and May 2026. Respondents comprised 1,000 IT leaders and 400 automation specialists working across banking, manufacturing, retail and consumer goods, insurance, and public-sector or education organisations. The average organisation represented had global revenue of $3.4bn and a workforce of 4,890.
The methodology makes the findings a stronger indicator of large-enterprise technology practice than of the economy as a whole. It also means the results should not be presented as UK-specific evidence.
The constraints identified align with a wider change in enterprise AI procurement. Initial experimentation often centred on access to large language models and employee productivity tools. Larger deployments increasingly depend on integration, governance, identity, permissions, data quality, auditability, and measurable return on investment.
Those capabilities are less visible than model performance but are harder to solve through technology procurement alone.
Richard Bovey, Chief for Data at AND Digital, said in supplied commentary: “The organizations leading in AI are the ones investing in high-quality data foundations. Without governance and reliable data platforms, AI workloads become brittle, costly, and difficult to audit.”
Governance requirements increase as systems gain autonomy. If AI is limited to recommending an action, employees can intervene before a decision is made. Agentic systems that trigger transactions, update records, contact customers, or alter workflows extend risk into operational processes.
Reliable automation therefore depends on more than whether a model can generate a plausible answer. Organisations need clear ownership of source data, consistent definitions, authorised access, rules for exceptions, monitoring, and a record of how outcomes were produced.
Technology teams are also under greater pressure to demonstrate measurable gains as AI budgets expand. Alteryx found productivity improvement was the most common measure of AI success, cited by 53% of respondents, followed by cost reduction at 45% and revenue growth or broader business impact at 39%.
Those metrics increase the incentive to connect AI with core processes rather than isolate it in experimentation. They also make weak data foundations more expensive because implementation costs rise without a corresponding improvement in output.
The next enterprise AI divide may therefore be less about access to models than organisational readiness. Most large businesses can procure AI capability. Fewer have converted years of accumulated processes, definitions, permissions, and institutional knowledge into systems that automated tools can use safely.
Investment is continuing regardless. With 80% of surveyed technology leaders expecting higher spending, pressure will grow to turn experimental deployments into repeatable operational returns. That transition depends as much on data, governance, process design, and business ownership as on the underlying AI technology.
You must be logged in to post a comment.