The countdown to one of the most thrilling events on the Formula 1 (F1) calendar (that is, the Singapore Grand Prix) is officially underway. Representing the pinnacle of motorsport, the F1 car is a thrilling showcase of how cutting-edge engineering and precise design come together in the relentless pursuit of speed and performance.
In many ways, F1 is also a fitting metaphor for how organisations approach agentic AI. Just as a huge part of an F1 driver’s success is determined by their car, an AI agent’s performance is also informed by data. A driver’s skill, while crucial, can’t overcome the limitations of a slow car; similarly, the best AI agents are ineffective without trusted, unified data.
Without trusted data, AI outputs are often inaccurate, overly generic, or simply lacking the specific context needed to drive meaningful action. It’s no surprise that nearly half (48%) of Singaporean workers say it’s difficult to get what they want out of AI right now. The reality is that many organisations, in their rush to embrace the latest AI advancements, are overlooking a critical element for success: the ability to scale their solutions and consistently deliver reliable outputs.
Context turbocharges the AI engine
Businesses have been told time and again: “Your AI is only as good as your data”. They invest significant effort into cleaning and structuring their data. Yet, many still struggle to achieve meaningful and useful outputs from their AI agents.
The problem? AI agents don’t just need data; they need context or the deep, nuanced understanding of the business, embedded in an organisation’s enterprise knowledge.
See also: Bank of Singapore uses AI agents to cut source of wealth report time to one hour
This “enterprise knowledge” isn’t just the sum of a company’s data stored in intentionally created, curated, and maintained databases. While structured information – such as customers’ names, contact details, financial transactions, and product information – is essential, it represents only part of the picture.
True enterprise knowledge also encompasses a vast, often untapped trove of unstructured information that includes everything from documents, emails, customer interactions, internal guides, and even the nuances of teams’ knowledge. Without harnessing both structured and unstructured knowledge, even the most advanced AI agents will fall short of delivering meaningful, trusted, and actionable outputs.
To return to our F1 analogy, giving an AI agent data without context is like sending a driver onto the track alone with the car. With only the car’s dashboard for guidance, the driver is effectively racing blind. But enterprise knowledge is the race engineer's voice in the driver’s ear, providing real-time updates on car performance, track conditions, changing weather, and race strategy. Having that crucial context gives the driver a full picture of the race to make the strategic decisions needed to win.
See also: The babble about a looming AI bubble
Consider a customer service agent tasked with resolving a billing dispute. Raw transactional data might tell the agent what a customer purchased and when. But without enterprise context, such as the customer's interaction history, seasonal purchasing patterns, and even sentiment from previous email conversations, that agent won’t be able to grasp the specific context to provide a helpful solution.
Where organisations stall on track
This context gap is where many organisations stumble in their AI journey. Enterprise knowledge is often notoriously difficult to activate, locked away in hundreds of disconnected systems and buried in unstructured formats like Slack messages, PDFs, meeting recordings, and support tickets. With the average enterprise juggling over 897 applications, 71% of which are disconnected, it’s no wonder agents struggle to get a full picture of the business, let alone offer useful or trusted outputs.
Data silos mean there's no single source of truth. Without it, agents can't reason effectively, understand nuance, or make confident, informed decisions. Instead, they risk making superficial or even incorrect choices, which erodes trust and limits their ability to drive real value.
The power of unified data
The only way for agentic AI to truly succeed is to connect all of this disparate data and infuse it with real-world business context. When AI agents have access to the full picture, they’re able to act more intelligently, adapt to dynamic scenarios, and deliver meaningful outputs.
Platforms like Salesforce Data Cloud connect both structured and unstructured data sources into one unified, integrated platform. Its zero copy technology allows organisations to access and query their data in real-time, without the need to extract, transform, and load their data across multiple sources – a process which can be costly and time-consuming. All of this data grounds Agentforce, Salesforce’s suite of AI agent tools. It ensures that AI agents are grounded in all your business data, while providing robust security, governance, and compliance.
To stay ahead of the latest tech trends, click here for DigitalEdge Section
When done right, agentic AI can be a game-changer. Sales teams can send personalised messages based on real-time customer insights. Service agents can resolve issues faster and with more empathy, and customers can get product recommendations that feel intuitive and relevant.
Getting to the chequered flag
A skilled F1 driver can’t win with just a competitive car; they also need real-time insights and strategy from their race engineer. The same is true for AI agents: they require trusted, unified data and crucial business context to deliver their full potential. As the AI race heats up, organisations that successfully connect their data and ground it in a real business context will gain a decisive edge, powering their way to the chequered flag in this AI Grand Prix.
Gavin Barfield is the vice president and chief technology officer for Solutions at Salesforce Asean
