A new artificial intelligence startup called Probably has raised $9 million in funding with an ambitious goal — building AI that you can actually trust to tell the truth.
The hallucination problem
If you've used any AI chatbot, you've probably run into the issue. The technology can be impressively articulate one moment and confidently wrong the next, inventing facts, citations, or details that simply don't exist. In the AI world, these confident fabrications are called "hallucinations," and they remain one of the biggest barriers to using AI in situations where accuracy genuinely matters.
Probably is betting that solving this problem is a business worth building. The company says its mission is to prevent hallucinations and factual errors from ever reaching users in the first place, rather than catching them after the fact.
Reliability on par with traditional software
What sets Probably's pitch apart is the standard it's aiming for. The startup wants its AI to achieve accuracy on par with deterministic systems — in other words, the kind of predictable, reliable software that produces the same correct answer every single time.
That's a high bar. Most generative AI today is probabilistic by nature, meaning it makes educated guesses about what to say next rather than following fixed rules. That flexibility is what makes modern AI so powerful, but it's also exactly why these systems can drift away from the facts. Bridging the gap between flexible AI and rock-solid reliability is the technical challenge Probably has set for itself.
Why investors are paying attention
The $9 million raise reflects growing investor appetite for companies tackling AI's trustworthiness problem. As businesses rush to deploy AI in customer service, healthcare, finance, and legal work, the stakes around accuracy keep climbing. A chatbot that invents a fake refund policy or misstates a medical detail isn't just embarrassing — it can be costly and even dangerous.
That's created a market opening for startups focused less on flashy capabilities and more on dependability. If Probably can deliver AI outputs that organizations trust without constant human double-checking, it could become an attractive layer for companies wary of deploying AI in high-stakes settings.
A crowded but critical race
Probably is far from alone in chasing reliable AI. Major labs and a wave of startups are all working on techniques to ground AI responses in verified information and reduce errors. But the company's framing — aiming for the consistency of deterministic systems — stakes out a clear and demanding position in that race.
Whether Probably can live up to its name remains to be seen. Reducing hallucinations has proven stubbornly difficult across the entire industry, and $9 million is modest compared to the billions flowing into frontier AI labs. Still, the funding signals that "trustworthy AI" is shaping up to be one of the defining battlegrounds of the next phase of the AI boom.
Source: TechCrunch
