From Gen AI experiments to agent-powered reinvention
AI exists in most companies. Generative AI tools (copilots/chatbots) are widely in use, layered onto existing workflows in an ad hoc way. In this situation, while Individuals benefit, AI hasn’t been treated as a strategic enabler for business growth. Agentic AI is ready to change all that, bringing in next-gen efficiencies that would have seemed impossible just a few years ago. Columbus has worked with companies where processes that once took hours now take seconds, where AI-driven predictive maintenance of machines has slashed costs, and even where cargo ships can be cleared at customs faster to ensure fresh food produce stays fresh.
Currently, the common approach to using AI is mostly informal. Generative AI is used by people within the business, rather than by the business itself. They call upon it as a nifty assistant to help optimise task completion at the personal productivity level. This is the first phase of the AI revolution, where anybody can take advantage of AI’s simple, but still sophisticated, capabilities. AI is not yet being widely leveraged to reinvent work in a way that impacts the bottom line, even though it’s ready and waiting to.
Given the rapid pace of evolution and progress in any field of technology, the second phase is now here. In this phase, AI is fundamentally about becoming a more data-driven, learning organisation. That involves a change in mindset from viewing it as merely a tool, to embracing it as a core competence.
Companies are endeavouring to make that change, but it appears many are not finding it as easy as they may have anticipated. A report from Deloitte states: “Revenue growth largely remains an aspiration, with 74% of organisations hoping to grow revenue through their AI initiatives in the future compared to just 20% that are already doing so.”i
Why embrace agentic AI?
Agentic AI provides an execution layer across all digital workloads, tasks, and processes. AI agents act as autonomous ‘workers’, bringing an added dimension to productivity that is of far greater business value than the chats and responses, advice and guidance that generative AI offers. They bring fundamental change to processes. The core difference is that the way they operate is based on goals, as opposed to prompts. They can take on specific tasks, and then plan, adapt and learn throughout the execution of the tasks. They collaborate. The people within your business then become supervisors of the tasks. People make the decisions, set the context, and handle the exceptions, but as for the actual work, the AI agent does it.
Maximising the business value of AI: The three foundational pillars
A number of factors contribute to the slippage between aspiration and realisation (or actual implementation) observed by Deloitte. Once you address them, you’re ready to join the second phase of the AI revolution. What is more enticing is that you’re ready to leap ahead of the competition.
Columbus has helped guide numerous businesses through this transition. Common pain points they’ve encountered occur when trying to move pilots or proof of concepts outside of the initial stages to realise and optimise business value. Businesses come up against lack of orchestration, missing memory, missing context for the initiatives, weak integration into core systems and processes. It becomes difficult for them to truly harvest the value that lies in their AI initiatives.
The result of any combination of these stumbling blocks, or even one of them alone, is that experiments or pilots stall as they lack the scalability to make broader adoption possible. Exploits prove initially fruitless due to not having given due consideration to reinventing processes, rather than just trying to put AI on top of them.
To build solid foundations for your AI initiative to evolve constantly, three areas will require attention. Projects will accelerate dramatically as each pillar strengthens. Later projects will benefit from the full foundation and can move from pilot to production much faster:
- Strategy and operating model
To define the vision, and how teams align in achieving it, it’s important to treat AI as part of your business strategy and not uniquely as an IT exercise. Ensuring that your operational setup is scalable will enable you to drive internal adoption and support growth.
Achieving the greatest chance of project success requires not just C-level buy-in, but also AI literacy among executives, middle management, and employees. As you pull your strategy together, make sure all those involved subscribe to the reasoning as to why it’s important to act now. At the same time, evaluate what risks would be involved if you decided just to wait a while longer.
- Data and infrastructure
The technical backbone, your infrastructure, needs to be flexible to enable innovation at pace, and should be addressed across all platforms, pipelines, and access layers.
For the highest degree of success you will find that clearly defined ownership and accountability will serve to expedite your initiative. Then comes the consideration you will need to give to data, since agentic AI is data-driven. Is your data setup scalable and available? Is your data of high quality? The old adage of garbage-in-garbage-out will apply more than ever in the new agentic reality.
- Governance and security
This pillar addresses transparency and traceability; knowing exactly who did what, when, and who has the right to. By establishing a detailed governance framework and tight security rules you’ll be able to ensure trust and meet any compliance obligations relating to your business, and particularly to your handling of data. Once again, ownership and accountability are important to make sure AI is used responsibly.
The AI revolution in action
Here are three real-life user cases providing a snapshot of the advances to their processes that some companies are now able to achieve with agentic AI and supporting technologies:
- Automated document generation and validation
An industrial machinery manufacturer was encountering delays and audit risks in validating its technical documents and audit records. The root of the problem was that all processes were dealt with manually. Columbus developed a Copilot Studio solution using Azure AI to automate document generation, validation, and workflows. At every step, the solution guides users and integrates approvals. The processes are all now faster, engineering workloads have been reduced, and documentation is consistent, traceable, and automatically ready for audit. - Reduced unplanned downtime
A company in a similar field required improved real-time insight into machine health across is sites, to avoid recurrent unplanned downtime and reactive maintenance. An IoT-based predictive maintenance system replaced reactive, manual processes. Using Azure IoT, Data Explorer, and machine learning, Columbus developed an IoT-based predictive maintenance system. The solution has now enabled real-time monitoring, anomaly detection, and failure forecasting, delivering ~200% ROI within nine months. - Reducing search time from hours to seconds
A national energy distributor needed to reduce the amount of time its staff spent on collecting and reviewing data across multiple internal and external data sources. The process was impacting the speed with which the organisation could produce tenders, rulings, and decisions across its many public websites. With Copilot Studio Columbus built an AI agent to automate data collection and search. The agent scrapes public sources, stores content in Azure Blob Storage, and indexes it via Vector DB and Azure AI Search. Relevant documents can now be located instantly.
How Columbus can help you seize the business benefits of AI
The AI team at Columbus approaches AI from a customer-first perspective. We’ve accompanied customers in a wide range of industries through their AI adoption journeys. Our role has been in helping them reinvent entire processes and making sure that their pilots evolved into full-on, scalable solutions. We have helped embed AI across solution areas such as ERP, data platforms and digital commerce, enabling customers to automate workflows, enhance insights and improve interactions with their own customers.
While ensuring scalability, and the creation of the foregoing three foundational pillars, Columbus always give full attention to the funding aspect of the project. Within this approach we make sure that the AI journey is funded through careful selection and development of high-impact use cases. Value is often achieved in areas where sometimes if hasn’t seemed possible.
