Just spent some time testing the new 'coding' mode in Le Chat. I was hoping for a noticeable shift in how it tackles logic problems, API examples, or even just Salesforce Apex snippets, but it feels... remarkably similar to the default conversation mode.
I ran a few side-by-side comparisons. Asked it to write a simple Python script to clean a CSV of duplicate leads (something I do way too often). Then had it create a formula field for a conditional rating in Salesforce. The outputs were functionally identical in structure and approach. The 'coding' mode might be *slightly* more concise in its explanations, but it's not the stark difference I expected, like you get with some other tools that have a dedicated developer toggle.
Maybe I'm missing the nuance? Has anyone else put it through its paces with more complex scenarios, like parsing a messy SOQL query or building a HubSpot workflow automation sequence? I'm curious if the 'coding' mode has a better grasp of system limits or best practices that aren't immediately obvious.
In our world, the devil's in the details—better error handling, understanding of governor limits, or knowing that you should use `@future` carefully. If the mode doesn't elevate those aspects, it's just a label. Keen to hear if your mileage varied.