Loading...
A major communications agency made a custom LLM its public face. An AI interface is not an AI strategy — and the harder question is who owns the decisions above the system.
On August 13, 2026, a major communications agency launched a custom LLM as its primary website interface, backed by Claude via Amazon Bedrock, RAG, a curated knowledge graph, and human moderation. An AI interface is not an AI strategy. As agencies ship AI systems, the harder question is who owns the decisions above the system.
88% of enterprises now use AI. Only 6% see meaningful EBIT impact. The gap isn't technology — it's what sits above it. When an agency builds an AI interface, it demonstrates operating capability. It does not demonstrate strategic accountability.
On August 13, Ruder Finn became the latest agency to make an AI system its public face — replacing its conventional website with a custom LLM interface that answers visitor questions using a knowledge graph and human-in-the-loop review.
This is not a competitive threat. It is a market signal: the bar for "AI-native agency" is now visible, public, and rising.
Three implications for your next board conversation:
1. AI systems are becoming table stakes, not differentiators. When agencies ship conversational interfaces, knowledge graphs, and RAG pipelines as their public presence, "we use AI" stops being a claim and becomes a baseline. The question shifts from "do you use AI?" to "who owns the strategic choices your AI informs?"
2. The gap is widening between AI capability and AI accountability. CMOs now allocate 15.3% of marketing budgets to AI — yet 70% say their marketing processes aren't mature enough to implement it effectively. Every dollar spent on AI infrastructure without a governance layer for the decisions it generates is a dollar that produces output, not outcome.
3. The open seat is above the system. Agencies are building AI interfaces faster than they're defining who owns the strategic direction those interfaces serve. The market rewards execution. It rewards governance even more.
Before your next agency conversation, apply this diagnostic to any AI system your partners build:
| Lens | Question | Red Flag |
|---|---|---|
| Ownership | Who owns the strategic recommendation — the system or a named human? | No named owner |
| Appealability | Can you challenge the AI's output with a human accountable for the answer? | Black-box output with no recourse |
| Memory | Does the system retain strategic context across decisions, or reset each session? | Session-only memory, no institutional learning |
A system that scores "no" on any lens is an interface. A system that scores "yes" on all three is a strategy partner.
Bottom line: When agencies ship AI systems, the market gets more interfaces. The organizations that get more strategy are the ones that demand an accountable human layer above the system — by design, not by accident.
Book a 30-minute demo. Bring a live question and watch the answer get built.