Deepseek is a good option, if you trust it being a Chinese model....

Good point! Then it would be nice to mimic an AI crew in n8n and see if outcome is same or even better! Would love to see such a comparison, might dive into it if I find the time 😁

Analysis of DeepSeek DeepThink: Conclusion: While n8n can handle simple AI-driven workflows (e.g., automating GPT-4 responses), it falls short in replicating CrewAI's capabilities for complex, collaborative AI tasks that require adaptive reasoning and role-based agent interactions. The choice between the two tools depends on the nature of the task: - n8n: Ideal for integrating applications and automating linear processes. - CrewAI: Best for AI-native problem-solving involving dynamic agent teams and advanced reasoning. For hybrid use cases, combining both tools (e.g., leveraging CrewAI for AI logic and n8n for workflow execution) can provide optimal results.

Infinite memory, like Letta offers: https://github.com/letta-ai/letta Current AI systems, with their limited context and static prompts, don't make sense for practical use. If I share personal details like my profession, age, hobbies, or family, that information should be stored permanently. Also when I say something changed like hobby, it should store that change. Additionally, I should be able to provide details about my colleagues, work, calendar, and more. For example, if I want to plan a holiday, the AI should know when I typically take time off, my preferences, and suggest destinations accordingly. It could also check flight schedules if I ask it to explore specific options, without requiring me to repeat context every time. By integrating infinite memory management, prompts could be dynamically built based on context (as Letta does). This allows the system to think about the question, determine relevance, and return accurate, personalized responses.