Relevance AI has done an impressive job of maintaining its identity as a truly agentic platform, resisting the common trend of drifting back toward traditional workflow builders, which is genuinely commendable. At the same time, we’ve recently seen them reintroduce some lightweight workflow-style capabilities within tools, adding structure and reliability where pure agentic decision-making can sometimes be unpredictable. It’s a balance we’re personally very happy to see, and it meaningfully strengthens the platform. The pace of improvement this year has been outstanding: faster, smarter agents, a growing set of integrations, and a team that listens closely to feedback and ships updates that actually matter. While the agent-first approach can still create occasional friction, the platform is becoming more intuitive, more stable, and more powerful with every release. Relevance is capable of producing sophisticated results for technical users while still being approachable for those with limited experience. The main caution is to keep an eye on token usage during testing, as agent simulations can consume more than expected. But overall, if you’re interested in where accessible, agent-driven automation is heading, Relevance is absolutely worth exploring.