The cause almost always lies in the foundation beneath it: incorrect data, systems that don't communicate with one another, facts scattered across spreadsheets, inboxes and tools that were never designed to work together.
Salesforce's State of Data and Analytics report, based on more than 7,600 data and IT leaders worldwide, reveals that 84 per cent say their data strategy requires a thorough overhaul before their AI ambitions can succeed. Yet the vast majority of organisations still lack a unified data strategy. It is in this gap that AI experiments run aground.
This article is all about the fundamentals. Because before an AI agent can carry out a meaningful action, it must first answer one question: what is actually true here? If the answer is unclear, the agent will either take a guess, ask a clarifying question, or stop. None of those outcomes is what you had in mind.
The data problem is not a technical problem
Every system you use leaves data in its wake. A CRM for customer contacts, a financial system for invoices and turnover, project tools for planning and execution, spreadsheets for everything else that doesn't fit elsewhere. Each system is maintained by different people with different definitions. Without a data strategy, that collection grows without anyone determining which source takes precedence.
Who owns a customer? The account manager in the CRM or the contract in the financial system? And what exactly does 'project margin' mean: the margin at the time of quotation, upon completion of the project, or the figure after adjustments? In most organisations that have grown over the years, this is the order of the day.
AI actually makes that ambiguity worse. If the underlying data provides three different answers to the same question, the agent will produce three different outcomes, depending on how it's configured and which source it happens to consult at that moment. That is of no use to anyone.
A single central source of truth
Organisations where AI really delivers results have one thing in common: a single place where all relevant business data is consolidated and remains consistent.
At CBEE Remarkable, we work on this every day within our niche: the lead-to-cash process on the Salesforce platform. From initial contact right through to invoicing, everything runs on the same platform, using the same data and the same definitions. Every step in the process draws on the same source.
If that isn't possible, or if an organisation deliberately opts for a different landscape? Then we bring the separate sources together to create a complete view of the customer, with unified data in Salesforce Data 360. The goal remains the same: a single source of truth that systems and people can rely on.
That foundation is created through the painstaking work of bringing data together into a single system, with consistent definitions, accessible to everyone who needs it. All the AI work that follows is built on that foundation.
What “grounding” actually means
In AI terminology, this is called 'grounding'. It means that an AI agent operates on the basis of your data: your customers, your projects, your appointments, your margins.
Grounding ensures that a generic assistant understands your organisation. An answer that might be correct in any context becomes an answer that is correct for your specific situation. That difference determines whether AI is of any use to you in day-to-day practice.
Salesforce Data 360 is built on precisely this principle. It connects the silos that already exist and makes them available to AI in a single, consistent format. That gives the agent a technical foundation so it no longer has to guess which source is correct. Which definition takes precedence is still human work: together with the organisation, we lay that down in the data model. It doesn't appear automatically the moment the platform goes live.
What this means for your organization
When preparing for AI, the most important question is what state your data is in today. Is your customer data complete and consistent? Do your financial and operational systems speak the same language? Is there a single authoritative source for the figures that matter?
If the answer to any of these questions isn't a resounding 'yes', then that's where you need to start: data management is the foundation on which AI runs. Getting your data in order is the first step. Next comes the question of how AI learns what that data means in your context.
Would you like to know how solid your data foundation is? On 9 and 10 September, you'll find us at Data Expo at the Jaarbeurs Utrecht. Pop by our stand, and I'll go through it with you together with the team from CBEE Remarkable.
This blog post is a contribution from CBEE Remarkable, an implementation partner specializing in Salesforce and Certinia. CBEE helps organizations in professional services, IT, and Health & Life Sciences digitize and connect their commercial and financial processes, from lead to cash. Find more inspiration at www.cbeeremarkable.nl or visit CBEE Remarkable during Data Expo.
Author: Chris Boonstra
