LegalOn achieves cost efficiency without compromising AI development speed

LegalOn Technologies, a global provider of AI-driven legal and business solutions, has managed to cut AI costs by 65% while maintaining development speed. The company achieved this by strategically matching models like GPT-6 (Astra, Luna) and GPT-6.1 Sol to specific tasks and aligning budgets with each business’s growth stage. Yuta Tokitake, Senior Engineering Manager at LegalOn, emphasized that new businesses prioritize speed over cost efficiency to drive growth, while mature businesses focus on cost optimization. To control costs without slowing development, LegalOn introduced budget caps at the department, group, and individual levels. It also restricted Fast mode by default, allowing it only when necessary. Teams used parallel task execution to maintain performance. The company also developed a new metric to evaluate AI investment based on customer value rather than just development speed or usage volume. LegalOn is now building a pipeline to calculate this metric, aiming to make ROI in terms of customer value visible. LegalOn is also working to turn individual AI expertise into an organization-wide capability by creating a knowledge base to share best practices. The company is rethinking how work is divided between people and AI, and has started emphasizing AI skills in hiring. AID CoE and the security team are collaborating to optimize performance, cost, and risk, creating a flexible framework for experimentation across the organization.