Generative AI, agents, and automation can unlock countless benefits for an enterprise; it’s moved from an experimental nice-to-have, to a necessity for organizations to not fall behind their competition. And yet, many business leaders remain hesitant about fully embracing these technologies.
Too often, I see companies begin trialing AI and automation, only for projects to be quietly shelved a few months later. Gartner recently predicted that 40% of agentic AI projects will be abandoned by the end of 2027. So, what is standing in the way of AI projects moving beyond the pilot stage?
Vice President of Global Solution Consulting at Workato.
The reality is that while the influx of generative AI tools in boardroom conversations has pushed many to adopt tools quickly, organizations have ended up skipping some critical foundational steps that, ultimately, slow and derail projects entirely.
As with any new technology, there are several fears and concerns blocking enterprises from embracing AI. A survey from Workato asked UK business leaders about their biggest barriers to AI adoption, and the most prevalent answers were governance, privacy, security and cost.
Let’s unpack why each is considered an obstacle and what can be done to address those concerns head-on. With the right approach and insights, business leaders can understand how to overcome the most common challenges and build an AI strategy with confidence.
Obstacle #1: Governance
The dramatic and exponential rise of generative AI has proven just how powerful and unpredictable the technology is. AI agents are simultaneously exciting and terrifying in their capabilities, presenting seemingly uncapped potential that most of us can’t fully comprehend.
Considering the speed at which AI can be integrated, the risk of hallucinations, and the potential for granting control to multiple systems, it is understandable that governance is the biggest barrier to British businesses’ adoption of AI.
Upholding human-in-the-loop workflows is critical for IT leaders to maintain authority and power over agents and other automated systems. AI should be versioned, reviewed and retired like any other software.
Defining and programming boundaries throughout the full development lifecycle, like structured prompts, contextual grounding and restricted output scopes, helps ensure the technology is working for you, in the way you want. Lifecycle governance, from creation to sunsetting, prevents any errors from falling through the gaps.
When it comes to agentic AI, consider who is the agent? Agents should be treated like system users, and allocated a defined identity, scoped access, and a clear owner. Often an agent will run on behalf of a human user, taking actions as that user across connected systems.
Applying least privilege and non-repudiation to that system user ensures every agent is traceable, intentional, and appropriately limited. Without these controls, agents become security liabilities.
Concern #2: Privacy & Security
Unsurprisingly, privacy and security were critical concerns. Consider Fortune 500 companies and those working in sensitive industries, like defense or healthcare – an agent revealing data it shouldn’t, or an AI tool providing an easily hacked gateway for threat actors, may feel like more risk than reward.
But you don’t have to forego security compliance to be innovative and agile. There are a multitude of security measures that enable businesses to experiment and innovate, while keeping company data secure.
When selecting secure AI tools, business leaders should ensure full visibility and control. For example, it’s possible to granularly limit the extent of an AI model or agent’s reach, as well as who in the company can command the program.
Remember, AI agents should only be granted access to the specific data and functionalities they need.
Embracing zero-trust architecture, where you assume everyone and everything is a potential threat, is a fail-safe, too. AI, agents and automation should meet the same standards as any other person or IT system interacting with the company’s data and be consistently audited.
The who, what, and why behind every automated action must be observable and logged so that security teams have the visibility needed to detect anomalies and respond confidently. Finally, customer data should always be kept separate and encrypted with a unique key that is rotated hourly.
By managing strict access controls and clear audit logs, businesses can be confident that enhancing efficiency through AI will not compromise security and privacy. Maintaining human oversight in this way is critical and often required under regulatory frameworks.
Business leaders can also seek guidance from regulation, which are written with security and privacy in mind.
Concern #3: Cost
Decision makers in UK companies acknowledged that the cost of implementing AI was another major barrier to fully embracing the technology. Sam Altman of OpenAI fame has predicted the cost of using AI to drop ten times every year.
Soon enough, AI tools in enterprise will become fully commoditized, but that doesn’t mean you should wait to grab a bargain.
Businesses need to consider the long-term returns on their operations and strategy by investing today. Delaying investing in AI will lead to losses in productivity and efficiency compared to competitors and ultimately, reduced bottom-line revenue and slower ROI when you eventually do invest.
Instead of not spending on AI, companies need to make sure they’re spending on the right AI. No one wants to waste IT budget on ambitious pilots that don’t pay off because they weren’t designed in line with the wider business needs. Here – I need to make a critical distinction…
Removing AI roadblocks with orchestration
This is where orchestration comes in, to ensure AI investments are well spent and guarantee returns. While AI adoption involves adding generative AI or automation tools to some parts of your business – like a ChatBot or a new tool – AI orchestration is a far more holistic approach to AI.
It requires much more consideration, but brings considerably more value. You can think of it as the connective tissue between your AI initiatives and your existing business logic.
It’s what ensures the new technology works alongside and within existing data, systems and people toward a shared outcome that can continue to develop and thrive longer term.
Crucially, AI needs to complement and work with a business’s existing model and setup to achieve long term impact and financial pay-off. That’s where speedy AI adoption fails and orchestration wins.
Governance, security, privacy and cost are legitimate concerns but not impenetrable roadblocks. Combined with an approach that is grounded in orchestration, UK companies can feel confident that they are embarking on their AI journey for long-haul rewards.
We’ve featured the best AI chatbot for business.
This article was produced as part of TechRadarPro’s Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro
