KPMG has reported a sharp rise in organisations embedding artificial intelligence into everyday work, while industry figures warn that adoption without measurement leaves serious gaps in value, cost control and security.

The consultancy’s latest Global AI Pulse Q2 2026 report found that 22% of organisations are now at the “driving adoption” stage, embedding AI into daily workflows. That is up nine percentage points from the previous quarter. The report also found many projects are under cost pressure, with almost half of businesses scaling back or pausing AI initiatives after concluding the cost outweighed the value. Organisations with full visibility of AI operating costs reported established returns on investment at five times the rate of those without that insight.

Specialists across finance, payments and cyber security drew a common link between rapid deployment and relatively weak discipline around proving value. They identified cost tracking, auditability and sovereignty as emerging fault lines as AI moves deeper into critical systems.

At payments technology firm RedCompass Labs, the focus is on whether banks and other regulated entities can show that AI delivers measurable benefits while remaining controllable. The firm works with financial institutions on payments modernisation and AI adoption in areas subject to strict regulatory and audit oversight.

“More firms are implementing AI into everyday workflows, but that does not mean they are creating value. Proving its worth is difficult without a clear way to measure it. Organisations need to know what AI costs to run, what it is actually producing, and who checked the output before it is used in a decision.

“This is particularly important in banking and payments, where speed with AI is of little value if you cannot clearly articulate what it generated, who signed off on it, and whether it would withstand an audit. AI maturity will depend on having that measurement ability, so firms can prove it is safe, accountable and worth the money,” said Oliver St Clair-Stannard, VP, Payments AI Strategy and Go-To-Market, RedCompass Labs.

His comments reflect a wider concern that AI is outpacing governance in highly regulated sectors. Banks must be able to show how AI models behave, what they cost and how humans oversee their output, especially in payments flows that affect financial stability and compliance.

Finance leaders see a related problem in how companies account for AI costs. KPMG’s findings on stalled or scaled-back projects underline how weak financial controls can undermine enthusiastic deployment.

Expense management platform Rydoo argues that AI investment belongs as much to the finance function as to technology teams. Its leadership has urged boards to treat AI spending with the same rigour as any other strategic project, not as an experimental budget line.

“Cost visibility is what separates businesses getting real ROI from AI and those still waiting for it. Confidence in AI is high across the board, but confidence doesn’t equal proof, and most businesses still can’t say with any precision what their AI is actually costing them to run. Without that visibility, AI stops being an investment and simply becomes another operational expense.

“The fact that nearly half of businesses have scaled back or paused AI projects after realising the costs outweighed the value shows that AI can’t be treated as a purely technological decision. Finance leaders need to be involved from the outset, with clear oversight of how AI spend is tracked, measured and linked to overarching business outcomes. AI should be held to the same financial discipline as any other strategic investment if companies actually want to achieve sustainable returns,” said Aidana Zhakupbekova, COFO, Rydoo.

AI’s growing role in national security planning adds another layer of complexity around control and dependency. The UK government recently outlined plans for a national “Cyber Shield” using frontier AI systems in cyber defence.

Cyber security consultancy NCC Group has examined national approaches to AI-enabled defence in its Global Cyber Policy Radar research. Its UK government affairs team has mapped how states approach AI in critical infrastructure protection, incident response and digital sovereignty.

“Frontier AI models are demonstrating the ability to identify software vulnerabilities and security weaknesses at unprecedented speed, creating the prospect of both AI-enabled attackers and AI-enabled defenders operating at machine scale. If offensive capability is accelerating, defensive capability must accelerate too. Governments therefore have a legitimate interest in exploring national-scale cyber defence capabilities for the most critical systems and infrastructure.

“Cyber Shield reflects a wider global trend identified in NCC Group’s Global Cyber Policy Radar: Cyber security is no longer viewed solely as an operational issue but increasingly as an instrument of national security, economic security and geopolitical resilience. States are investing more heavily in capabilities that allow them to protect critical infrastructure, reduce strategic vulnerabilities and respond at national scale.

“The strongest case exists for protecting nationally significant systems. The potential value of AI-driven defence is greatest where scale, complexity and consequence intersect: government networks, critical national infrastructure, essential public services and major digital supply chains. These environments generate more telemetry and vulnerabilities than humans alone can realistically process. AI-enabled detection, prioritisation and remediation could therefore become an important force multiplier.

“Success will ultimately depend on partnership rather than government acting alone. The UK’s cyber capabilities, expertise and innovation are distributed across industry, academia and government. A sovereign cyber defence capability is likely to be most effective if it leverages this wider ecosystem through trusted public-private collaboration rather than attempting to centralise capability entirely within the state.

“Moreover, no country is an island in the digital sphere. While sovereign assets are critical for resilience, interoperability and shared assurance with trusted partners will determine effectiveness.

“The growing debate around frontier AI access has highlighted a genuine strategic risk: dependence on a small number of foreign-controlled providers for technologies that may become critical to national security and cyber resilience. Recent discussions around restricted access to advanced AI capabilities illustrate why governments are increasingly considering sovereign options.

“Sovereignty is fundamentally about control, resilience and assurance rather than nationality alone. NCC Group’s work on digital sovereignty argues that organisations should focus on who controls the technology, data, infrastructure and decision-making processes, rather than reducing complex security questions to simple geographic ownership tests.

“A narrow interpretation of sovereignty could create unintended consequences. Across Europe, NCC Group has observed growing efforts to restrict market access through sovereignty requirements. While these measures are often motivated by legitimate security concerns, overly protectionist approaches can reduce access to expertise, innovation and trusted partners, potentially weakening resilience rather than strengthening it.

“The challenge for Cyber Shield will be balancing sovereign control with access to the best available capabilities. A successful model is likely to require sovereign governance, sovereign operational decision-making and strong assurance over data and infrastructure, while remaining open to collaboration with trusted technology partners and suppliers. Resilience comes from assurance and diversity, not isolation.

“The bigger policy question is not whether Cyber Shield is sovereign, but whether it avoids creating new strategic dependencies. This mirrors concerns emerging in both UK and European policy debates: as governments adopt advanced AI-enabled cyber defence tools, they must ensure those capabilities don’t simply replace one dependency with another. Sovereignty must therefore be viewed as an outcome of resilience, assurance and control, rather than an end in itself,” said Louise Horton, Head of UK Government Affairs, NCC Group.

CMap has launched CMap Chat, an artificial intelligence agent for its professional services automation platform. The tool adds a conversational interface to the company’s existing AI features.

The product is aimed at consulting and architecture, engineering and construction firms that use CMap’s software to manage sales, operations, project delivery, finance, reporting and administration. Users can ask questions in plain language and receive answers drawn from their own operational data, rather than building reports or navigating dashboards.

The launch expands CMap Intelligence, the company’s earlier AI offering, which is built around six specialist agents covering sales, operations, delivery, finance, reporting and admin. CMap Chat sits above those functions, giving staff access to insights through a single conversational prompt.

Early users are applying the tool to routine commercial and operational queries, including checking which invoices are overdue, forecasting quarterly revenue, identifying sales opportunities most likely to close, finding outstanding timesheets and seeing which consultants are available to start new projects.

The system can also generate reports from natural language requests, reducing the time spent searching across multiple systems for information. That puts the new product in a growing market for AI assistants designed to sit inside business software and answer questions using company data rather than general web results.

Operational focus

CMap argues that generic AI models can produce plausible answers but often lack the business context needed for operational decisions. Its pitch is that a system trained on a firm’s own quotes, margins, resourcing choices, wins and write-offs can provide advice that is more relevant to day-to-day management.

CMap’s customer base spans consulting and AEC businesses in the UK and US, with more than 45,000 users on its platform. By placing a chat interface over those records, the company aims to make operational data easier to access for managers and project teams that may not regularly use reporting tools.

Jon Stead, Chief Executive Officer at CMap, outlined the company’s view of the problem facing professional services firms.

“The firms that are pulling ahead now aren’t working harder, they’re working with better data. Too many businesses which have been built on expertise are now being run on guesswork. Accessing the data and insights that could power your growth is being held back by information being stored in spreadsheets, emails and locked in the heads of a few senior people.

CMap Chat is the elite advisor every firm needs, but few can afford. It’s always on, knows your entire operational history and gets smarter with every project you deliver. Natural language insights means it is the front door to all of your internal data. It is how your MD checks margins across the firm, how your PM spots a project going over budget before it’s too late and how your ops lead checks utilisation rates across different offices,” said Stead.

Broader trend

The introduction of chat-based tools inside sector-specific software reflects a wider shift in enterprise technology. Software providers are trying to make analytics and workflow systems easier to use by replacing fixed menus and report builders with natural language queries, while keeping access tied to internal permissions and business records.

For professional services firms, that matters because financial and delivery performance often depends on a mix of utilisation, pricing, scope control, billing discipline and project staffing. Information on those areas is often spread across separate systems or held by a small number of senior employees, which can slow decision-making.

CMap designed CMap Intelligence around simplicity, use within existing workflows and a focus on specific tasks rather than broad experimentation. In practice, that means helping teams move from asking a question to analysing performance and acting on the answer within the same environment.

The chat interface is intended as a single route into intelligence already embedded in the platform, rather than a separate application to manage. That approach may appeal to firms that want AI tools without adding another standalone product to an already crowded software stack.

Next steps

Later product updates will add deeper sector expertise, more pre-built reporting options, enhanced permissions and more actions that can be completed directly in the chat interface, including time-off and expenses processes, while retaining final user control. CMap also plans to add Model Context Protocol support so users can connect CMap data to large language models such as Claude, Gemini, ChatGPT and Perplexity.

For now, CMap is positioning CMap Chat as a practical layer over operational data for firms that want faster answers to everyday questions without relying on manual reporting. More than 45,000 users in consulting and AEC businesses already use the wider CMap platform across the UK and US.

ZYMIX held its first brand event at Henley Royal Regatta, using the gathering to present its social app to Generation Z users.

The event also hosted a small survey of 20 Gen Z attendees and UK content creators on social media use, digital wellbeing and creator economics.

ZYMIX describes itself as the UK’s first social SuperApp and is building a platform that combines messaging, short video, communities and payments in one app. It is targeting university students and younger social media users ahead of a wider UK rollout.

The survey findings pointed to a gap between time spent on social platforms and users’ sense of real-world connection. According to ZYMIX, 90% of Gen Z respondents spend more than two hours a day on social media, while 20% said it helps them feel connected in real life to a great extent. Another 70% described that impact as limited.

Among content creators surveyed at the event, every respondent said they wanted a platform that better links online audiences with real-world experiences. The findings also suggested concern over how creators earn money online. According to ZYMIX, 80% of Gen Z content creators preferred traditional brand partnerships over algorithm-driven pay-per-view models, while 60% said better monetisation would be the main reason to move to another platform. Half said a stronger sense of authentic community would also encourage them to switch.

Survey focus

The sample was small, and the results were described as indicative rather than nationally representative. Even so, the findings reflect a wider debate in the social media sector over whether heavy use of digital platforms is strengthening or weakening social ties among younger users.

That debate has widened to include online safety, age restrictions and digital wellbeing. Platforms aimed at younger audiences are under pressure to show not only how users engage with content, but also whether those services help or harm broader social behaviour.

ZYMIX is positioning itself within that discussion, arguing that social media should do more to connect digital activity with offline relationships and events. It says its app is being designed to help users discover events, meet people with shared interests, support creators and build communities away from the screen.

“ZYMIX was created to be one place for your life online and beyond,” said Nikita McKnight, Marketing Director at ZYMIX.

“Young people don’t need another platform that simply encourages more scrolling. They want an environment that helps them discover opportunities, build communities and turn online connections into real-life experiences. That’s exactly what we’re building – a platform designed around life beyond the screen, not just screen time,” McKnight said.

Creator economy

The creator-focused findings are likely to be watched closely by platforms competing for younger influencers and their audiences. Social media companies have spent heavily on revenue-sharing models, subscriptions and creator funds, yet many creators remain dependent on direct brand deals for predictable income.

ZYMIX’s survey suggests those traditional commercial arrangements still matter more to many younger creators than models tied directly to platform algorithms or per-view payments. That points to an unresolved issue for emerging platforms: attracting creators may require not only audience growth, but also a clear path to stable earnings and stronger communities.

ZYMIX also used the Henley gathering to announce support for London Youth Rowing, including a donation to the organisation. The funding, it said, would support the charity’s work with young people through sport, including confidence-building, resilience and life skills.

For ZYMIX, the link with a youth sports initiative fits its broader attempt to tie its brand to offline community activity rather than online engagement alone. It framed the move as part of a wider commitment to initiatives that create opportunities for young people beyond the digital sphere.

The company also set out broader context at the event.

“Our ZYMIX at Henley event celebrates everything social platforms should ultimately encourage – friendship, community and shared experiences,” said McKnight.

“Technology should help create those moments, not replace them. Our mission is to build a platform that inspires young people to turn digital connections into real-world relationships and lasting communities,” McKnight said.