
Zencoder at present presents two built-in brokers, for unit check technology and coding, and an interface for outlining customized brokers. Customized brokers are basically named, saved instruction prompts for the big language mannequin, aimed toward fixing particular duties.
IDG
Zencoder in comparison with opponents
Zencoder (the corporate) helpfully ready comparability charts for Zencoder (the product) and eight opponents. Not surprisingly, since I by no means actually belief what one vendor has to say about one other’s product even after they try and be honest and balanced, I believe that each one of those comparisons have been designed to tout Zencoder. Some have inaccuracies that understate the rival’s capabilities, and several other appear to be about older, much less succesful variations of the competitors. For instance, the Copilot aspect of the screenshot under is newer than the model Zencoder used to put in writing its comparability.
Zencoder omitted Solver from their comparisons. Given how new Solver is, that’s comprehensible. However, Solver is much more promising than Zencoder, because it not solely reads complete repositories, it may well modify complete repos. You’ll be able to learn my evaluation of Solver right here.

On the left, you may see Zencoder’s rationalization of a C++ sorting perform; on the proper, you may see GitHub Copilot’s rationalization of the identical code utilizing the o1-preview mannequin. On the prime you may see the Zencoder features obtainable in VS Code within the Zencoder v0.5.1 plugin. Total, I like the reason on the best higher, however I’m keen on a documentation model that mixes phrases and code.
IDG
An bold try
Zencoder is, in my view, an bold try and enter the code technology and restore market, but it surely’s nonetheless moist behind the ears. Utilizing a pipeline with error correction to keep away from hallucinations, as an alternative of straight code technology with an LLM, sounds promising, but it surely isn’t ipso facto higher than a mannequin that generates appropriate code proper out of the field and suppresses hallucinations internally. Equally, “grokking” complete repositories ought to present higher context than simply taking a look at no matter pages are open within the editor, however the time it takes to digest giant repos is critical, and it appears to be like like Zencoder redoes that course of from scratch everytime you refresh the repo.