IT Brief Ireland - Technology news for CIOs & IT decision-makers
Ireland
Klarus sees AI, cyber & data trends for mid-market

Klarus sees AI, cyber & data trends for mid-market

Wed, 18th Nov 2026 (Today)
Sofiah Nichole Salivio
SOFIAH NICHOLE SALIVIO News Editor

Klarus has outlined four technology trends it expects to shape mid-market businesses in 2027, centred on AI adoption, cyber security, data and automation.

The London-based technology consultancy said companies are moving beyond AI trials and grappling with more practical questions about how to generate business value while managing weaknesses in legacy systems and data.

Its assessment comes as businesses continue to increase spending on AI-related systems and infrastructure. For the mid-market, the next phase will depend less on experimentation and more on whether organisations can fix the operational problems that limit wider use.

Alper Gunaydin, Chief Technology Officer at Klarus, said defensive AI would become more important as cyber threats evolve.

"In 2027, rising AI-enabled cyber attacks will shift defensive AI from an optional upgrade to a critical investment for mid-market companies. They will face enterprise-level threats, but limited budgets, outdated legacy systems and scarce security-specific skills mean they will favour AI-assisted detection and response that can be embedded into existing workflows.

"Adoption, however, will be pragmatic, with companies demanding evidence that tools improve protection before committing any spend. Concerns about rogue agents will also accelerate investment in governance, infrastructure and safeguards alongside defensive AI," said Gunaydin.

Klarus said medium-sized organisations face the same threat environment as larger peers without the same internal resources. That is likely to make embedded detection and response tools more attractive, provided they can be introduced into existing processes.

Legacy systems

A second theme is the role of older enterprise software in slowing AI rollouts. Many mid-market companies still depend on long-established systems for functions such as finance, customer management and human resources, and those platforms often make integration harder.

Gunaydin said these constraints would become clearer as companies seek broader use of AI in business processes.

"Legacy enterprise systems, such as ERP, CRM and HRIS, will remain a major bottleneck to scaling AI in the mid-market. In our research this year, 40% of respondents at mid-market companies in the UK and Ireland said enterprise applications were preventing or delaying their piloting or deployment of AI.

"In 2027, companies will focus on moving to autonomous workflows, which means they will need to modernise their data architectures and API layers. In the short term, AI applications and modern middleware solutions can act as a thin layer over legacy enterprise systems to overcome some of the inherent difficulties of ageing, fragmented infrastructure. However, scaling AI's benefits will depend on a holistic approach that balances quick wins with longer-term solutions. This will mean modernising the data and application landscape with a prioritised roadmap," said Gunaydin.

Some businesses will try to bridge the gap by adding software layers over older systems rather than replacing them all at once. That approach could help companies move faster, although longer-term progress will still require broader changes to applications and data structures.

Data quality

Data governance was identified as a third priority. Poor data quality continues to undermine AI projects and is shifting from a technical issue handled by IT teams to a matter for senior management.

Gunaydin said organisations would need clearer accountability and sharper investment decisions if they want to prepare data for wider AI use.

"With poor data quality hindering many AI deployments, data governance and cleanup will shift from a back-office IT task to a board-level strategic priority. As a result, mid-market companies will look to increase investment in data readiness to strengthen the foundations for future cross-company AI rollouts.

"Usually, poor data quality is a symptom of a broader data trust problem, and improving data trust depends on answering five key questions: Which data matters, to determine critical domains and requirements? Who is accountable, to determine business ownership, common definitions and quality standards? What should be built, to determine architecture, integration and lifecycle principles? How will it be run, to determine leadership, skills, business and IT decision rights? And what gets funded first, to determine sequenced investment, dependencies and benefit owners?" said Gunaydin.

The focus on data reflects a wider concern among businesses adopting AI: tools are only as useful as the information they draw on. For mid-market groups, that may mean deciding which datasets have the greatest business value before expanding investment.

Robot access

Klarus also said physical robots could become more accessible to medium-sized businesses as hardware costs fall and financing models change. It pointed to leasing arrangements and robots-as-a-service offerings as possible ways to reduce the upfront spending that has traditionally limited adoption.

Gunaydin said new AI models could also lower the effort needed to train machines for workplace tasks.

"Physical robots will be within reach of mid-market companies in five years, transforming the work environment by blending human and machine labour. In 2027, we will begin to see falling hardware costs, along with leasing and robots-as-a-service models, which are likely to lower the upfront investment that has historically kept physical automation within reach only of large enterprises.

"New AI world models and multimodal LLMs that understand physical space, objects, and cause and effect could let robots learn tasks from demonstration or plain-language instruction rather than months of custom programming. At the same time, rising wage costs in sectors such as hospitality, logistics and care may push mid-market operators to automate repetitive or physically demanding roles sooner than they otherwise would," said Gunaydin.

Klarus argues that the next stage of AI adoption in the mid-market will depend less on headline spending and more on whether companies can strengthen cyber defences, update older systems, improve data quality and assess where automation makes economic sense.