Generative artificial intelligence has taken the world by storm, but a persistent problem undermines trust for developers and users alike: hallucinations. When an AI model invents facts or provides plausible but false answers, the damage for businesses and professionals can be significant. Today, a startup called Probably announces a $9 million funding round precisely to tackle this challenge, aiming for a more reliable and deterministic AI.
A funding round that marks a turning point
Probably has closed a Series A round led by deep tech investors, with the goal of developing an architecture that drastically reduces factual errors. The team, composed of seasoned researchers, has already demonstrated that it is possible to achieve accuracy comparable to deterministic systems without sacrificing the flexibility of language models. This approach relies on a cross-verification mechanism that does not simply filter answers but intervenes directly in the generation process. The current landscape shows increasing pressure on the industry: the recent drop in ChatGPT's market share below 50 percent has highlighted how users are seeking more robust alternatives. According to analysis, the demand for reliable systems is growing, pushing startups like Probably to fill a critical gap.
Sponsored Protocol
Beyond hallucinations: the value for enterprises
The impact of hallucinations is not just an embarrassment for chatbots. In finance, healthcare, or legal sectors, a wrong answer can have legal and economic consequences. Probably enters this scenario with a solution that combines probabilistic control and logical verification, reducing error margins to nearly zero. The model not only avoids fabrications but is also capable of stating its own uncertainty, a feature many current systems still lack. This approach echoes methodologies already seen in DevOps, where repeatability and predictability are essential. For instance, platforms like Threads are focusing on personalization, but the real priority for the future of AI remains trust in generated data.
Sponsored Protocol
Probably's technology also stands out for its real-time capability, a must-have for applications such as automated customer support or document analysis. According to the company's CEO, the goal is to make AI not only creative but also accountable. The funding will be used to expand the research team and bring the solution to market within the next twelve months.
An evolving ecosystem
Probably's initiative fits into a broader movement of companies striving to make AI more transparent. While giants like OpenAI and Google work on ever larger models, more agile startups explore alternative paths to ensure reliability. A recent study from Wikipedia on artificial intelligence hallucination emphasizes that the lack of verification is one of the main barriers to enterprise adoption. Probably promises to overcome this barrier, offering a product that could change the rules of the game.
Sponsored Protocol
In conclusion, the $9 million represents a major bet on a future where AI is no longer perceived as an unreliable black box but as a precise and verifiable tool. For developers and enterprises, it is a sign that the path to widespread adoption necessarily passes through trust.
Source: https://techcrunch.com/2026/06/16/probably-raises-9m-to-build-a-more-reliable-kind-of-ai
