Business · 10 February 2026
AI in Business
For business leaders and professionals
At askKira, we provide secure, personalised enterprise AI for UK businesses. Our platform prevents data leaks, protects IP and client data, and gives your teams AI-powered tools at a fraction of the cost of other enterprise solutions.
This week's roundup brings you the latest AI developments that matter for UK organisations.
Enterprise Strategy
Deloitte report reveals AI adoption stall
Despite a 50% increase in access to AI tools, daily usage remains flat according to Deloitte's State of AI in the Enterprise 2026 report. The analysis suggests the problem is not a lack of training but a failure to redesign workflows. Leaders are attempting to bolt AI onto existing human processes rather than empowering staff to redesign the work itself from the bottom up.
Why it matters: Buying licences without restructuring daily schedules and workflows will result in zero ROI; staff must be given time to reinvent their own processes.
Read the source on LinkedIn →The risk of the 'Hollow Expert'
As organisations push for AI transformation to boost speed by 25%, a new skills gap is emerging. Analysis suggests that when teams use AI to solve problems, they bypass the cognitive struggle required to build deep intuition. This creates 'hollow experts' who can deliver results but lack the expertise to fix systems when the AI fails or strategy shifts.
Why it matters: You must balance short-term efficiency with long-term talent development to ensure your future leadership pipeline remains competent.
Read the source on LinkedIn →Shift from knowledge to judgment
With AI making knowledge retrieval instant, the value of corporate training is shifting. The focus is no longer on content accumulation but on developing judgment. Employees do not need to memorise facts but must learn to discern what matters, what to trust and how to act in consequential contexts.
Why it matters: L&D budgets should move away from information transfer and toward critical thinking and decision-making scenarios.
Read the source on LinkedIn →Governance & Risk
EU AI Act ends 'we did not know' defence
Under the new EU AI Act provisions, claiming ignorance of a system's behaviour is no longer a valid legal defence. Regulators are shifting the test to 'foreseeability' meaning if a risk was documented in literature or visible in similar use cases, organisations are liable. Silence in documentation will now be interpreted as an absence of control rather than an unfortunate accident.
Why it matters: UK companies doing business in the EU must proactively document that they have anticipated risks rather than just reacting to harms.
Read the source on LinkedIn →PayPal incident highlights model drift risk
A recent fraud detection failure involving PayPal and German banks blocked over €10 billion in transactions, causing significant disruption. The root cause appears to be model drift where an AI system degrades over time unnoticed. This incident underscores that governance is not just bureaucracy but critical infrastructure for financial stability.
Why it matters: Automated drift detection and continuous monitoring are now mandatory for any AI system making financial or critical operational decisions.
Read the source on LinkedIn →Agents & Tech
OpenAI announces 'Frontier' platform
OpenAI has introduced Frontier, a new platform designed to help enterprises build and manage AI agents that perform actual work rather than just chat. Moving beyond fragmented point solutions, Frontier offers a shared context layer for enterprise data and an agent runtime to deploy at scale. It aims to allow agents to collaborate across the organisation with proper permissions.
Why it matters: This signals the shift from 'chatting with AI' to 'assigning work to AI' within secure enterprise environments.
Read the source on LinkedIn →Distinguishing Agents from Automation
A critical distinction is emerging between Robotic Process Automation (RPA) and Agentic AI. While RPA automates clicks and follows strict rules suitable for static environments, Agentic AI perceives context and makes decisions. IT teams are advised to stop labeling everything an 'agent' to avoid confusion and ensure projects deliver the expected ROI.
Why it matters: Understanding this difference prevents costly procurement errors; use RPA for repetition and Agents for complex problem-solving.
Read the source on LinkedIn →