Insights from Vector Space
Insights, scenarios and models to help you understand how artificial intelligence
is transforming the way organisations work.
Communicating AI to employees: the problem isn’t resistance, it’s how it’s introduced
65% of employees are enthusiastic about using AI at work. Only 3% actively avoid it. Yet 37% don't use it because their colleagues don't. The problem isn't worker resistance: it's the urgency of executives that leads to rushed implementations without considering workforce implications.
Automating isn’t installing: the hidden cost of an AI system in production
72% of CIOs report breaking even or losing money on AI investments. Gartner predicts that over 40% of agentic AI projects will be canceled by 2027 due to costs that weren't anticipated. The real cost of AI automation isn't the deployment: it's what comes after.
The AI competence center: is it really needed or just a bureaucratic layer?
Only 1 in 5 organizations is getting value from AI tools, and only 1 in 50 is seeing transformative value. Gartner recommends building an AI center of excellence that works like an air traffic controller. But building it wrong is worse than not building it at all.
Which process to automate first: a strategic choice, not a technical one
72% of CIOs report that their organization is breaking even or losing money on AI investments. Over 40% of agentic AI projects will be canceled by 2027. The problem isn't the technology: it's the choice of starting point. Gartner has defined the criteria to get it right.
NPS doesn’t measure automated customer service. And it never did
Gartner had predicted that over 75% of organizations would abandon Net Promoter Score as a customer service metric by 2025. With AI handling most interactions, the NPS problem has become even more acute. Here's what to measure instead.
How to measure the quality of a customer service that talks to robots
Traditional customer service metrics don't work when most interactions are handled by AI. Average handling time, first contact resolution rate, CSAT: all require a rethink when the agent isn't human. Gartner has already redefined what to measure and how.
Chat, email, phone: AI still doesn’t manage them as a unified whole
51% of customers are willing to use a GenAI assistant to handle service interactions. 91% of customer service leaders are under pressure to implement AI across all channels. But orchestrating chat, email, and phone into a single intelligent workflow is harder than it looks.
Handing the customer to a human: the most critical moment in AI customer service
None of the Fortune 500 companies will have completely eliminated human customer service by 2028. 50% of organizations that had planned significant staff cuts will abandon those plans by 2027. The escalation point between AI and human is the most critical moment in modern customer service.
Data, agents, and risks: security in BPA systems with generative AI
By 2027, over 40% of AI-related data breaches will be caused by improper use of GenAI across geographic boundaries. 57% of employees use personal GenAI accounts for work. Global cybersecurity spending will reach $240 billion in 2026. The problem is already inside organizations.
Low-code and AI: who’s actually building tomorrow’s workflows
By 2026 developers outside formal IT will represent at least 80% of low-code tool users. By 2028, 90% of software engineers will use AI assistants to write code. Between these two trends lies the space where tomorrow's workflows are being built.
AI won’t eliminate legacy IT. But it can help you move past it
Vendor claims that GenAI will eliminate technical debt and automatically modernize legacy systems have inflated expectations. Gartner clarifies what AI can actually do for modernization, and what it can't.
How to measure an AI agent that actually works
45% of high AI maturity organizations keep their projects in production for at least three years. Only 20% of low maturity organizations do. The difference isn't the technology: it's the measurement. Here's what to measure and how to make the numbers actually useful.
When the AI agent must stop
The autonomy of AI agents is an advantage up to the point where it stops being one. Defining when an agent must stop and ask for human confirmation is not a technical limitation: it is a design choice that determines whether a system is reliable or dangerous.
