The State of AI: How Organisations Are Rewiring to Capture Value

A summary and commentary on McKinsey’s latest AI survey

Why this matters

AI, especially generative AI (gen-AI), is no longer a niche topic—it has entered mainstream conversation inside organisations. Yet many companies struggle to turn that excitement into meaningful value. According to McKinsey, while adoption is strong, scaling and measurable impact remain the exception rather than the rule.

Major findings

Here are key take-aways from the latest survey:

• Widespread usage: More than three-quarters of organisations say they now use AI in at least one business function. McKinsey & Company
• Gen-AI is growing fast: Use of generative-AI (which includes large-language models, image/video generation, code-generation etc) is rapidly increasing. McKinsey & Company
• Workflow redesign is critical: The survey found that redesigning processes and workflows is strongly correlated with seeing bottom-line impact from gen-AI. In other words: adopting AI is not enough — you have to change how things are done. McKinsey & Company
• Governance and leadership matter: Organisations where the CEO or board take explicit oversight of AI governance tend to report higher impact. For example, 28 % of respondents whose firms use AI say their CEO oversees AI governance. McKinsey & Company
• Risk-mitigation rising, but uneven: More firms are treating risks (inaccuracy, IP & copyright, cybersecurity) seriously — though many still are not doing so comprehensively. McKinsey & Company
• Scaling remains elusive: Despite high usage, a large majority of organisations report only limited enterprise-wide gains. Many use AI in isolated functions, without full integration. McKinsey & Company

What the numbers suggest

• Although usage is high, meaningful ROI is much lower.
• Bigger firms (those with revenue above USD 500 million) are moving faster — giving smaller firms a risk of falling behind.
• The margin between “doing AI” and “getting value from AI” is defined by practices: workflow redesign, governance, KPIs, talent, infrastructure.

Additional context & insights

To enrich the blog audience’s perspective, here are some extra angles and details:

1. The “pilot-to-scale” gap

While many organisations have run pilots or isolated AI projects, few have scaled those into business-wide transformations. One commentary of the McKinsey report describes this as:

“AI adoption is mainstream — but transformation isn’t.” In Parallel
This gap means many firms may have AI tools, but still operate with old workflows and legacy organisational structures.

2. Talent & workforce implications

• Hiring for new AI-related roles is increasing: data scientists, ML engineers, AI compliance specialists, AI-ethics specialists. McKinsey & Company
• Reskilling is key: the time saved by automation is often redeployed into higher-value or different roles. At larger organisations, headcount reduction is more likely—especially in service-operations, supply-chain, inventory roles. McKinsey & Company
• A caution: Without proper workforce planning, automation may create disruption.

3. Risk & governance

AI isn’t risk-free. The survey highlights:

• 27 % of firms reviewing all gen-AI outputs before they go live — but a similar share review very little. McKinsey & Company
• Risks cited include inaccuracy, IP infringement, explainability, cybersecurity. McKinsey & Company
• Governance models vary widely: many organisations lack full-scale frameworks for responsible AI.
  This means firms must think not just “can we use AI?” but “should we use it?”, “how do we govern it?”, “how do we measure impact?”

4. Process & workflow change matters most

The single strongest correlate with bottom-line impact: workflow redesign. That is, embedding AI into core business processes—not simply bolting it on. This aligns with the broader theme: value comes not just from the tool, but from how you use it. McKinsey & Company
For startups, scale-ups, or established firms, this means: map your key processes, find where AI can change the way work happens, not just speed it up.

5. The future: agents, autonomous-AI & next horizon

The report flags “agentic AI” (AI systems that act autonomously, make decisions, coordinate other systems) as a next frontier. Organisations preparing now may have first-mover advantage. McKinsey & Company

What this means for startup-enablement / early-stage businesses

Given your audience (startups, founders, ecosystem enablers), here are tailored take-aways:

• Start with business function uptake: Many startups may not need enterprise-wide AI yet — focus on one function (e.g., marketing automation, customer-service bots, supply-chain analytics) and embed AI into that workflow.
• Define KPIs early: Without clear metrics, AI initiatives lose direction. McKinsey found tracking KPIs for gen-AI solutions is among the leading practices for achieving value. McKinsey & Company
• Build governance from the start: Even at early stage, creating a mini-governance framework (roles, oversight, risk checklist) reduces downstream problems.
• Combine talent + culture + tech: For early-stage teams, hiring a data-scientist alone is not enough — you need people who can integrate AI into products, and a culture that is comfortable experimenting and changing workflows.
• Focus on process redesign: Before deploying AI, ask: “Which workflow will this disrupt?” If you automate a bad process, you just automate inefficiency. Better to improve process then apply AI.
• Be realistic about impact: The hype is real, but the transformation is hard. The survey shows many organisations still struggle to move beyond pilots. Framing the journey as incremental rather than “overnight disruption” is more realistic.

Summary & final thoughts

The State of AI survey from McKinsey paints a picture of a technology in broad use, but still early in its value-creation phase. Organisations that succeed aren’t just adopting AI—they’re rewiring their workflows, building governance, aligning talent, and embedding AI into the very way they operate. For founders, operators and ecosystem builders, the key lesson is: the biggest lever for value is not the algorithm, but the process.