Thanks for sharing. I’m Zhijie Chen, founder of Verdent and formerly Head of Algorithms at TikTok. Verdent is built for larger, production-grade tasks: it plans before coding, runs agents in parallel, tests results, and explains diffs. Our north star is a full Plan Code Verify loop, and we are building toward that. Curious what kinds of tasks you would actually trust an AI to handle end to end.

▲

boburumurzokov 7 hours ago | parent [-]

Hi Zhijie Chen,

Do you have a benchmark of how many tokens Verdent agents can save for complex asks? For example, I stopped using Cline because it consumes quite a lot of tokens for bigger projects just to analyze and provide context.

	▲	ZJChen 5 hours ago \| parent [-]
		Hi, atm we can say that it's moderately less than Cline but we need more testing to give exact numbers. We have plans for testing by users and 3rd parties in future, for full transparency.