Business owners consuming the media storm around AI could be excused for buying into the day-by-day reboot of industry understanding and being paralysed into inaction. The fear is that anything we do today will be made obsolete tomorrow. The breathless coverage of AI agents running riot on systems, or appearing sentient when set up to converse with each other about their ‘owners’, mixed with doubt about whether this is really happening or just click-bait, all makes for a very low signal-to-noise ratio. Businesses trying to find out what is really happening and possible with AI have already fuelled a profusion of consultants telling them what they want to hear. In enterprises, this has led to expensive and unproductive AI experiments that took the hype and tried to apply pre-trained generative transformer technology to complex, multi-year processes and business intelligence, expecting overnight transformation. They mostly failed.
As organisations go back to the drawing board and embrace the mantra that “2026 is the year we get meaningful ROI from AI”, they’re realising the lessons from the experimentation phase are important. One of the most important is that AI implementation, as a core part of business transformation, is like many things: best approached carefully, stepwise, through maturity, and by engaging partners with long experience of technology-led transformation. When businesses move away from the hype and refocus on their core, they rediscover the value of the data, customers, processes and people they had previously overlooked. And working with partners who understand their business and their sector becomes worth far more than working with new consultancies being paid to learn on the job.
One organisation of interest at the higher end of the enterprise market is Helikai.ai, whose recent presentation on the IT Press Tour goes into some detail on how they engage and operate – most importantly, with referenceable, real-world outcomes of value. As you would expect, this is not an overnight approach. It uses maturity models such as MITRE’s to determine where to begin in an organisation; uses architectural frameworks that keep LLMs interchangeable to prevent lock-in, and even allow competing LLM outcomes; keeps people in the loop at all times for oversight and certainty; and, most importantly, uses a Secure Private Retrieval-Augmented Generation (SPRAG) platform, an on-premise or VPC-hosted private capability that is explicitly relevant to the business where it is deployed. Because the business problems being solved remain deterministic, allowing probabilistic systems to approximate determinism remains too risky for the enterprise. Applying AI to individual business problems resembles the way many large projects are achieved: as they say, “How do you eat an elephant? One bite at a time.”
For existing partners and technology service providers, this means changing if they are to remain relevant. While continuing to focus on data, security, infrastructure maturity and the other things their not-so-sexy offerings deliver, they also need an intimate knowledge of how to work with AI-focused partners who operate at the business level. And they need to deliver a range of vendor-agnostic, technology- and partner-led and, most importantly, explicitly business-value-oriented outcomes. Enterprises are already demanding different pricing and value models from their service providers, and the mid-market is sure to follow soon. Gone are multi-year contracts focused on tickets resolved or devices babysat; observability-driven, highly automated and AI-directed services are now expected and demanded. Service providers who recognise and capitalise on these changes in the market will survive. The others, not so much.
Taking a mid-market MSP to the place where it remains a relevant, business-focused technology service provider, centred on the business outcomes an AI-embedded transformation strategy demands, is no mean feat. But it is what is needed now to meet the market quickly.
And it is the leadership of these organisations that will decide whether they are still around – re-formed, rebuilt, but above all relevant partners to business in the years ahead.
This article first appeared on LinkedIn on 9 February 2026.