These threads seem to have become exceedingly vibes-based.
Yes, something may now be cheaper or more expensive or whatever, but there is no way to objectively measure quality (except for "trust me bro" benchmarks). So the discourse is people saying that for them, this or that model was better - which is a very low value data point.
Yes, programming language discussions can be vibes-based and therefore low value too, but sometimes the more concrete aspects of the languages at hand, e.g. language features and tradeoffs are discussed. That is something that I'm not seeing in equivalent AI discussions.. there is a lot of how a particular AI model "feels" to interact with.
It's got to be because AI has become a commodity, people are looking to get the best value for money, and things change based on your specific application (which may be nearly unique) and literally the time of day.
It's a bunch of bread bakers talking about wheat suppliers.
Enterprise AI adoption has reached a point now where FinOps matter, and a harness platform story with a discounted underlying model can be enticing for a number of organizations.
I've seen Gemini land well in a F100 well known for their AI hardware story for that reason, and Alibaba's leadership canned the OSS minded Qwen team in order to build a similar commercial minded approach as well.
At least in cybersecurity, we're also reaching a point where the harness is starting to matter more than the underlying foundation models, and building a harness/bedrock style story while discounting a specific model can play well in upper market deals.
What makes you think so? Is it only because you dont like Musk or do you have some insight into all companies using Cursor you want to share with us? Even if you dont like Musk you should realise that others may not share your sentiments and/or may have similar feelings regarding SamA, DarioA or any of the other CEOs in this field.
Enterprise customers care about things like guardrails and data safety. xAI has always been anti guardrails, and who knows if you can trust them with your data.
At least in the F1000 RFPs I've seen and the decisionmakers I've chatted with, when they talk about AI guardrails what they mean is generic API (eg. can we rate limit, block connections, RBAC/ABAC capabilities, etc) and Data Security (eg. ZDR, encryption at rest/transit, controlled access) controls.
There is a recognition that foundation models and tools leveraging them will introduce some degree of nondeterminism, so the best way to solve that is to leverage preexisting best practice that is used to reduce lateral movement risk by humans (who are similarly nondeterministic in nature).
My company’s security team is very much “no proprietary data or information can be used to train a model”, I just don’t know how you can validate or trust that they aren’t doing just that.
ZDRs, spot audits, and the fact that salespeople can be held personally liable both financially and even criminally for fraud if they sold a contract with a ZDR that was not honored.
