Block, Jack Dorsey's financial services company, announced Dessa, a new AI language model trained from scratch exclusively on human conversation data rather than internet text. The approach aims to create more natural, human-like dialogue patterns by learning from actual speech patterns and conversational norms rather than web-scraped text.

Dessa represents a different training philosophy from mainstream LLMs, prioritizing conversational naturalness over breadth of knowledge. Block's emphasis on training exclusively from conversation data suggests the model is optimized for customer service, interactive applications, and dialogue-heavy use cases rather than general knowledge tasks.

What This Means for Your Business

If you're evaluating AI solutions for customer service, voice assistants, or conversational interfaces, Dessa's training approach indicates there are now alternatives to general-purpose models that may deliver better performance for dialogue-heavy applications. Consider testing conversation-trained models alongside mainstream LLMs to determine which better matches your user interaction patterns.