Are we getting a first real taste of what artificial general intelligence (AGI) could look like with the roll out of agentic AI solutions from OpenAI, Anthropic and Google DeepMind? The future of AI’s impossible to predict, but we do believe that AI agents are a significant advancement towards that future. While a sceptic might say AI agents are just traditional robotic process automation (RPA) with a large language model (LLM) plugged in, we’re more excited about the impact agents may have on people’s thinking:
Which use cases can we dream up to make our lives better if we allow a reasoning language model to act on our behalf?
If you compare the capabilities of an AI agent to those of an intern or even a junior co-worker, what could happen if you roll out 1,000 of them simultaneously?
Which business processes can we re-engineer and which business models can we reinvent?
We’re convinced that agents will be the key for organisations to earn the return on their AI investments. Companies that embrace agentic AI early will gain a competitive edge, while those who hesitate risk falling behind. However, the key to successful integration lies not in merely jumping on the bandwagon, but in the strategic, thoughtful adoption of this technology. Therefore, business leaders are faced with an imperative: Focus on practical execution rather than falling victim to hype.
So, how can your organisation make sure it maximises its return on investment (ROI) from its AI ventures? Here are ten essential dos and don’ts to guide you on this journey.
The dos of smart AI agent adoption
1. Prioritise user experience
At its core, AI should serve to enhance and streamline user interactions, not create additional complexities. It’s crucial for businesses to design AI systems with the end user in mind. This means being transparent about where AI’s used, what its limitations are and thinking about how a human will stay in the loop.
2. Start small, scale smart
Before rolling out AI at full scale, begin with pilot projects. When done right, with aspects like scalability and security in mind, these initial tests provide invaluable insights into what works and what doesn’t. This approach helps make sure that when you scale, you do so with confidence and clarity.
3. Ensure high-quality data
AI's effectiveness is directly contingent upon the quality of data it is trained on. Organisations must prioritise building and maintaining robust data ecosystems. This means building a strong data management and data governance capability in your organisation.
4. Build cross-functional teams
The myth that a single team can own AI is precisely that, a myth. Achieving meaningful results with AI requires a collaborative approach that harnesses a broad spectrum of expertise. Encourage collaboration between data scientists, IT, business units and domain specialists to integrate diverse insights into AI projects.
5. Invest in employee training
Successful AI adoption isn't just about technology; it’s about empowering your people and building trust. Providing staff with regular training not only demystifies AI, but also fosters a culture of innovation and adaptability, crucial for leveraging the full potential of AI systems.
The don’ts: Pitfalls to avoid in AI deployment
1. Don’t rush AI deployment
Implementing AI in a rush without a clear strategy is a straight path to wasted investments. Companies should take a phased approach, aligning AI projects with their overarching business goals and continuously refining their strategy based on outcomes and learnings.
2. Don’t underestimate complexity
Far from being plug-and-play, AI agents require a profound understanding of integration within existing business processes. Underestimating this complexity can lead to incomplete deployments that fail to deliver expected value.
3. Don’t neglect human oversight
AI should augment human decision making, not replace it. AI agents still require lots of human supervision and direction. Decision making augmented by AI should still involve critical human oversight and monitoring to guide and validate AI-provided outputs.
4. Don’t overlook ethical concerns
In the rush to harness AI’s power, companies must not bypass ethical considerations. Bias, fairness and transparency should be part of AI governance frameworks, subjected to rigorous testing and review to ensure technology operates within acceptable ethical boundaries and is trusted by its users.
5. Don’t underestimate costs
AI systems require ongoing investment, not just in the software itself, but in infrastructure, training and maintenance. Misjudging these costs can lead to budget overruns and resource constraints, hampering the long-term success of AI initiatives.
The path forward
The ripple effects of adopting agentic AI have the potential to redefine entire industries. Organisations that master AI integration will position themselves as pioneers, setting the pace in their respective fields. On the other hand, businesses that fail to harness AI effectively may find it challenging to catch up with their forward-thinking counterparts.
So, how is your organisation making sure it reaps the return on its AI investments? By following these strategic dos and don'ts, you can start crafting a roadmap that leads not only to successful AI implementation, but also a transformative business future.
Agentic AI – realising value with agents
Contact us
Partner Technology Consulting & Innovation, PwC Belgium
Tel: +32 495 59 08 40
