The era of "unrestricted experimentation" in artificial intelligence is giving way to disciplined financial management. Businesses are fundamentally revising their strategies as Return on Investment (ROI) becomes a top priority, driven by the escalating costs of advanced language models.
The "Consultant" Strategy
According to technology consultants, a new trend is emerging where the most powerful and expensive AI models serve exclusively as "consultants." In this framework, top-tier systems handle planning and complex decision-making, while individual, simpler tasks are delegated to smaller, more economical models.
Ameya Kanitkar, CTO of Larridin, likens this approach to the operations of a law firm. Just as a high-priced senior partner would never handle basic legal filings, advanced AI models should be reserved for designing complex workflows.
Rationalizing Daily Operations
Executives like Michael Murphy of Adaptovate point out that there is no economic logic in using high-end models for tasks such as:
- Meeting transcriptions
- Basic web searches
- Generating short creative texts
Coinbase CEO Brian Armstrong predicts that within the next 12 to 18 months, 80% of AI tasks will be performed by models that are up to 99% cheaper than today's leading systems. The latter will remain essential only for scientific research, strategic decisions, and the coordination of autonomous AI agents.
This shift is already fostering a new market of companies specializing in helping organizations select the most appropriate model and monitor usage costs.
