@mcuban Yes exactly. In our experience LLMs can't be trusted for enterprise data, a human in loop at each step is a real value. AI accelerates, human validates.
@Dan_Jeffries1 A big part of the token economics problem may come not just from large context windows, but from uncontrolled reasoning loops in autonomous agent architectures. Eenterprise AI may need more workflow constrained and deterministic reasoning patterns.
@Cornell Great work. Humans are Biologically Intelligent (BI) species that uses Artificial Intelligence and Mechanical Intelligence (MI) to our advantage.
When Mechanically Intelligent devices truly merge with BI, the real AGI will emerge.
@a16z@Benioff Absolutely. Easy access to the CRM by agents, is welcome every where. At the same time demand for standardised analytics foundation is growing.
#quantamine
@KobeissiLetter The US sanctioned China and the irony is US itself took it's full enterprise force to China for business, but not the otherway around. Interesting times in the US.
@realBigBrainAI@chemanth we have built #Quantamine that works exactly the same way as described, AI is working with human to accelerate human work and validate it's output.
@OpenAI Congrats OpenAI on the Deployment Company launch! Deployment is now the real battlefield.
At @Quantamine, we use AI for Salesforce to Snowflake data models & transformations with full human-in-the-loop.
Models start it. Humans + AI win.
#AI#Enterprise
@mcuban But the inconsistency problem runs deeper than the AI layer. A lot of "inconsistent AI answers" in enterprise settings aren't hallucinations. They're data problems. Two people ask the same revenue question, hit different source tables, get different numbers.