OpenAI just released Canvas for Custom GPTs, including Python execution for graphs and free user access. While this addresses one major need, GPTs still require several key improvements to reach their full potential. Here are four critical features that would make GPTs substantially more capable. First, GPTs need a Recursive Mode. Currently, GPTs can only handle tasks that fit within a single response. This severely limits their ability to work on longer projects that require multiple steps or iterations. A proper recursive mode would let GPTs break down complex tasks into manageable chunks and work through them systematically. Second, Multi-Agent Collaboration would transform how GPTs operate. Imagine multiple specialized GPTs working together on a project – one handling research, another writing, and a third reviewing and editing. This kind of collaboration could dramatically improve the quality and efficiency of AI-assisted work. Third, Advanced Voice Mode with tool calling would make GPTs more accessible and powerful. Current voice interfaces are basic. Adding tool-calling capabilities would let users verbally trigger complex actions and workflows, making GPTs much more practical for hands-free use. Finally, being able to use o1 or a lighter model like GPT 4o mini would be an amazing feature. For complex enterprise GPTs, use o1, for fun joke ones, use GPT 4o or 4o mini. With OpenAI’s 12 days of shipmas underway and Canvas now released, we might see more updates addressing these needs. For more on recent OpenAI developments, check out my coverage of their planned releases at https://adam.holter.com/openai-plans-12-days-of-ai-releases-including-sora-and-new-reasoning-model/ Which features do you think GPTs need most? Let me know in the comments.
Menu
