Cutting-Edge AI Models: The New Business Consultants

AI Models as Luxury Consultants

In the field of artificial intelligence, a new approach is emerging to manage the costs associated with cutting-edge models. AI consultants are now suggesting that these advanced models be viewed as luxury advisors or consultants. The idea is simple: if you find that your budget for AI tokens is consistently exceeded, it's time to rethink your strategy.

These experts propose viewing cutting-edge models as strategic advisors. Allowing these models to handle complex thinking and planning, while delegating execution to less expensive models, could be the key to achieving significant savings. Optimizing return on investment (ROI) in AI spending has become a hot topic, and AI consultants have an interesting trick to reduce costs.

Maximizing the Efficiency of AI Models

Reports have widely covered the idea that users should direct their tasks towards the appropriate AI models to maximize their investment. However, AI consultants go further by asserting that the most expensive cutting-edge models should be treated as the costly consultants they are.

Ameya Kanitkar, Chief Technology Officer of the AI measurement platform Larridin, based in San Francisco, illustrated this idea by stating, "You wouldn't use the most expensive lawyer for routine tasks that less costly lawyers can handle." During a discussion in July, Kanitkar clarified that models like Fable 5, the most powerful and token-hungry model from Anthropic, should be used to create a workflow roadmap, while cheaper and smaller models should execute them.

"The consulting model essentially plans things, breaks down problems into smaller sets, and has the complete context of how everything will work," he added. "Then, the subtasks are delegated to less expensive models."

Lightweight Models for Daily Tasks

Michael Murphy, a partner at the AI transformation consulting firm Adaptovate, shares this vision. He believes it is unwise to use company funds to send "the most powerful model to do something that replaces a Google check or transcribes meeting notes or creates a creative brief."

Murphy asserts that cutting-edge models should be reserved for strategy development, creating initial versions of a new application or website, or for tasks requiring complex thinking. Companies should then determine which "lightweight models" are most appropriate for daily tasks.

It's not just consultants who support this tactic. Industry leaders have also discussed the idea of using cutting-edge models as advisors. In a post on X in June, Brian Armstrong, CEO of Coinbase, stated that he expects "80% of workloads to operate on models that are 99% cheaper within the next 12 to 18 months." The best models should be reserved for "IQ maxxing" tasks, he specified, such as scientific breakthroughs or orchestrating agents.

Optimizing AI ROI Becomes Crucial

Companies are increasingly concerned about not getting proportional returns on their AI spending. Many have abandoned tokenmaxxing, a trend where companies allowed their employees to freely experiment with AI, encouraging them to burn as many tokens as possible. Some, like Duolingo, have even integrated AI usage as a performance criterion.

Today, companies are adopting a more conservative approach, reflecting on how to optimize the value of their spending. Several budgeting tricks have emerged, such as routing models or using open-source Chinese models like Kimi K3 from Moonshot AI or GLM-5.2 from Z.ai.

A new wave of startups is capitalizing on AI ROI, making it a priority for executives. These startups, known as AI routing companies, help direct developers to different AI models and monitor excessive spending, becoming favorites among investors.

The New York-based startup OpenRouter announced in May that it raised $113 million, valuing the company at $1.3 billion. A competitor of OpenRouter, Concentrate AI, has also secured over $5 million in funding.