Originally published in Forbes Technology Council
As AI agents take on more work across the enterprise, token consumption is becoming an increasingly important part of the AI economics conversation. But what if rising token costs aren’t only a model or prompt optimization problem?
In his latest Forbes Technology Council article, Arango CEO Shekhar Iyer explores a different way to look at the issue: the amount of work agents must do to reconstruct the business context they need.
Customer histories, product information, policies, interactions and other critical business knowledge often already exist across the enterprise. But when that information is scattered across CRM and ERP systems, knowledge bases, product catalogs and operational applications, agents may need to repeatedly retrieve, reconcile and interpret it before they can act.
This is where context efficiency comes in: providing an AI system with the minimum trusted business context required for a task without forcing it to repeatedly discover, reconcile and translate that context at runtime.
From Token Costs to Business Outcomes
Improving context efficiency shifts the conversation from simply asking “How can we reduce token costs?” to asking “Why do our agents need to consume so many tokens in the first place?”
When trusted, governed business context can be reused across applications and agents, organizations can reduce unnecessary reconstruction and allow AI to spend more of its compute on the business problem at hand.
The opportunity extends beyond token savings. More efficient access to business context can contribute to lower latency, more consistent responses, greater explainability and governance, and more efficient use of compute.
“Organizations have spent decades building business knowledge. Agents shouldn’t have to rediscover it for every new task.”
Shekhar Iyer, CEO, Arango
