Professional services firms are adopting AI fast, but it’s not working to its full potential. Organizations in this sector nearly doubled their use of GenAI over the past year, but “most respondents say that AI is not yet central to their workflow,” according to a new Thomson Reuters Institute report. If nearly twice as many services firms are using AI now as in 2025, why isn’t AI at the core of their workflows yet?
The technology is ready. The organization’s data may even be ready. Often, the missing element is context. Context is what makes the difference between accurate, reliable AI outputs and outputs that feel like they come from an inexperienced (but very enthusiastic) intern. Even companies that meticulously standardize their data and rethink their processes before implementing AI can run into this problem if context isn’t an explicit part of their AI strategy.
How does context improve AI performance?
AI is powerful but it can struggle to analyze massive amounts of raw data. Without a domain-specific context layer, AI agents embedded into systems can’t correctly interpret the data they access.
For AI to be a good data analyst, it needs access to layers of abstraction based on the specific system where it’s embedded. Then, when a user asks for a specific KPI, the AI knows what to produce.
To test this idea, my team and I ran five rounds of an experiment in which we asked an AI agent to answer business-specific questions of increasing complexity using either raw data or data with a context layer. We started with simple queries like looking up a number. Then we asked more sophisticated questions like, “based on our open deals and probabilities, what revenue would we expect in 2026?”
For very simple tasks, the raw data and domain-specific context layer results were pretty much equivalent. But when we asked our most complex questions, the model using raw data spent 200 seconds trying to generate a response before giving up. The model using a domain-specific context layer consistently provided a correct response in 55 seconds.
What is a domain-specific context layer?
A domain-specific context layer abstracts an organization’s raw data from disparate systems into business metrics. Then the AI constructs queries to that context layer, which defines metrics and data in the same way that human users will describe them and provides a high-level view of the business. This approach delivers reliable outputs while avoiding inaccuracies that result from insufficient grounding.
Professional services firms that build a context layer to support their AI gain an operational asset rather than an underperforming investment. For example, if the leadership team wants data to inform a hiring decision, they can ask an AI with context-layer access for evidence that might support hiring a new creative-team member in the next three months. The AI can then access and analyze pipeline, current work, capacity, utilization targets and other factors to give an answer that can inform decisions.
What a domain-specific layer is not: feeding a services firm’s raw data into a public model like Claude or ChatGPT so the AI can figure it out. When organizations try this, it fails quickly because this approach doesn’t provide the model with the necessary vocabulary to understand the meaning behind the data. A domain-specific context layer does. Designing and building a domain-specific context layer also requires expertise that’s likely outside the scope of in-house teams for most services organizations. Working with a third-party engineering partner or solution provider is the most practical way for services firms outside the AI space to implement a contextual layer for their AI.
Organizational outcomes with a context layer
When professional services firms can use AI this way, they can go beyond getting answers to simple questions. Their AI investment can provide real business value by answering complex questions quickly and accurately. That can drive operational efficiency and margin improvements that make those firms more appealing to potential investors or buyers.
Services firms that develop a domain-specific context layer also improve their competitive advantage. The Thomson Reuters Institute survey found that while 57% of professional services firms are using public AI tools like ChatGPT, only 35% are using industry-specific GenAI or agentic AI solutions. Context-specific AI can improve internal AI adoption rates and drive efficiency improvements across the organization before competitors resolve their AI output reliability issues.
AI with a domain-specific context layer can also demonstrate the outcomes and ROI that CEOs need this year. Eight in ten CEOs say AI failures this year could cost them their jobs, and most say they feel pressured by boards to show AI ROI soon. That’s easier to do when the AI understands exactly how the business works so it can answer questions correctly.
AI ROI depends on more than technology and data readiness, although those matter, too. Giving AI the context it needs to work properly is the step that turns an investment with an unclear ROI into a tool that delivers real results.
As Partner at Form & Function Consulting, Adam Ribaudo draws on over 15 years of experience transforming complex customer data into strategic, measurable outcomes for enterprise clients. Adam has a proven history of architecting data solutions from the ground up, notably during his tenure as a Vice President at Velir, where he built and scaled a key revenue-contributing data division focused on advanced behavioral analytics, CDPs, and cloud data warehouses. Adam previously founded and led the consultancy Noise to Signal for over five years, collaborating directly with executive leadership to rationalize tangled data ecosystems and design bespoke architecture that directly drives tangible marketing and product growth.
