Sentient’s AI chatbot Dobby Plus challenges OpenAI with open-source, user-governed AI model

Sentient, a decentralized AI company, has introduced Dobby Plus, an open-source chatbot now available on X and Hugging Face, according to a June 4 statement shared with CryptoSlate.

This new model challenges proprietary AI systems from major tech firms like OpenAI and Anthropic.

Dobby Plus

According to Sentient, Dobby Plus is built on a decentralized AI framework shaped by input from over two million users.

Rather than relying on corporate directives, the chatbot evolves through continuous community contributions, refining its personality, tone, and decision-making to reflect shared values. This approach shifts the narrative from proprietary control to collective ownership.

Dobby’s design emphasizes a balanced blend of factual precision and emotional intelligence. Sentient attributes this to its unique “data overloading” training method, which enhances technical capabilities (such as math and coding) with soft skills like empathy and conversational nuance.

The company said this approach improves the model’s tone and user engagement while maintaining performance and accuracy.

Sentient revealed that its model currently outperforms Claude, GPT-4o, and DeepSeek R1 in personality, engagement, and creativity categories.

Moreover, Dobby’s open-source framework allows independent audits for risks like memory manipulation and prompt injection.

These are key features for the project as the AI community grows increasingly concerned about safety, especially while major tech companies are reportedly scaling back on rigorous testing.

Decentralized ownership

Meanwhile, the launch of Dobby Plus follows Sentient’s February release of its 70B model, which drew more than 660,000 users.

Each user received a cryptographically fingerprinted copy of the model, establishing a new method of open ownership that extends beyond open-source access.

Sentient sees this as a working example of how community-led models can be monetized fairly and transparently.

Himanshu Tyagi, co-founder of Sentient, said:

“We don’t think AI models should just line the pockets of whoever owns the GPU cluster. We’re building an AI system open to scrutiny, shaped by its users, and shared with those who make it valuable.”

Looking ahead, Sentient plans to develop a protocol for collectively governed AI systems, allowing users to become stakeholders with voting rights and profit-sharing benefits.

Mentioned in this article