Georg Ell is CEO of Phrase, a language tech leader, brings 20+ years of experience in the tech industry.
The most appealing assumption in enterprise AI right now may be that speed and fluency are the same as effectiveness. The economics are compelling. But a pattern is emerging that should concern any executive responsible for growth: Companies are scaling AI and watching performance flatten precisely in the areas expected to drive it.
I have seen versions of this pattern across different industries. After moving aggressively toward AI-driven operations, some major companies have since adjusted their approach as concerns emerged around quality, customer response and implementation. The question that every enterprise scaling AI now faces is how to ensure that organization-specific judgment about what works for a given audience is part of the system from the start.
The pattern of scaling AI and flattening performance becomes most costly in global communication because language is where cultural and market nuance concentrate. Teams can create content in every target language, campaigns can run on schedule and product information can be accurate, but companies may still see customer engagement soften and support volumes rise in markets that should be performing. The content is not wrong. It just does not reflect what the business has learned about what works in that market.
What The Models Don’t Know
Many enterprises are building on the same foundational models, but competitive advantage increasingly depends on the organizational intelligence those models are given to work with.
Global enterprises, with their knowledge of different markets, hold some of the most competitively valuable intelligence, but much of it is not structured in a way AI systems can use. Built over years of customer interaction, it is distributed across regional teams, customer support, marketing, legal and language operations. A large language model has broad linguistic capability but no access to how your product is really used in Japan, which messages build trust in Germany or where regulatory sensitivities sit in Brazil. That understanding is earned through experience. Making it systematically available to AI is becoming a discipline in its own right, one described as “language intelligence” embedded across enterprise systems.
Language intelligence is not simply about adding context. It requires the ability to evaluate whether AI-generated content meets brand and market standards. It also must keep improving as customer behavior evolves. Without that evaluation loop, AI increases content volume faster than organizations can build intelligence from it. The result is content that is fluent and consistent yet detached from the nuance that drives customer behavior. The markets meant to fuel growth quietly underperform instead.
Where Leaders Should Start
You may be tempted to start with technology, but the organizations I see making progress start with a different question: What does the business already know, and why can’t AI use it? Most global enterprises have accumulated significant understanding of what works in different markets, but it is often distributed across functions with no connection to the AI systems generating content.
The first leadership decision is determining which knowledge has the greatest effect on customer response and business risk, and how it can be structured in ways that AI systems can reference when producing content. Brand standards, regulatory requirements and market-specific messaging guidance often exist in static documents that are disconnected from the AI systems producing content. Moving them into centralized repositories that AI tools can access at the point of content creation is the enabling step. The technology varies, from enterprise knowledge management systems to dedicated language intelligence platforms (full disclosure: my company built such a platform) to custom integrations with existing AI workflows. What matters is the principle. AI can only apply what it can access, and most organizational knowledge about markets and customers is currently locked in formats designed for human readers.
The second decision is accountability. Until someone senior owns the connection between what the organization knows and what AI produces, the gap will persist regardless of technology.
Building Intelligence Into The System
PwC’s 2026 Global CEO Survey found that just 12% of CEOs say AI has delivered both cost and revenue gains, with those seeing returns two to three times more likely to have embedded AI across the full value chain.
For content that crosses markets, that means building feedback loops into how AI operates. The organizations making progress are integrating market performance data into content workflows so that customer responses automatically inform the next cycle. They are using AI-driven quality estimation to flag material that needs human review before it goes live.
This works by defining what good content looks like in each market, and using AI to score output against those benchmarks. Content that meets the threshold moves forward, while content that falls below it gets routed to human review. The organization sets the standards, and AI applies them at scale.
Regional teams also need a reliable way to update this guidance as market conditions and customer expectations change. This turns local knowledge from occasional input into reusable enterprise intelligence.
None of this requires a single platform. The determining factor is whether intelligence is embedded in the system or applied around it.
A Strategic Opportunity
Language intelligence is not a one-time implementation. It is a capability that appreciates with use. It’s an asset that strengthens over time and that competitors cannot replicate even with access to the same models.
According to the Conference Board, the percentage of S&P 500 companies disclosing AI as a risk jumped from 12% to 83% between 2023 and 2025. Boards have registered the exposure. The strategic opportunity behind that risk has received far less attention.
Language intelligence is following the same arc as data governance. Companies that treated data as a procurement problem eventually discovered that it was a competitive one. The same is happening with how companies communicate across global markets.
For most leadership teams, this issue has not yet entered the strategic conversation. That will change as AI becomes the primary layer through which global companies engage their customers. Every enterprise operating internationally is generating intelligence already. Whether it becomes part of the systems now responsible for customer engagement will determine which companies improve with each interaction and build an advantage that generic models cannot supply.
Forbes Business Council is the foremost growth and networking organization for business owners and leaders. Do I qualify?
