I mean this is the standard enterprise setup. When they used Microsoft or Google it was the same fear it’s just that the use of Anthropic products by them is high
Is this a response to Goldman banning it lol? Not sure
I think we’re not far away from that point. Maybe in a few years. I find as a dev I read my code less and less. I can still communicate with the agent by explaining the code it’s likely getting wrong, finding perf…
I think people just need to use their time to have fun. Is that not what our use is right now? We’ll find new work to do, new problems to explore, new ways to have fun. Our original use was hunting and gathering. We…
Nope that costs too many tokens lol (also performance is worse compared to their text counterparts from what I’ve noticed….)
Evals and benchmarks lol. Make the eval to see what the AI should be doing, benchmark to see what the whole landscape is doing. Refine and repeat and that’ll make it better
It’ll make sure agents automatically deployed will be much more secure than otherwise.
The health idea of yours is really interesting imo lol. I’ve been playing with two: 1. Knowledge Graph that auto parses vital context out 2. A security feature for agentic memory to secure chunks that don’t determine…
Yeah my company moved to its own version called Gitspoke which has been working a lot better and the UI for reviewers is also nicer
I wonder how much of this is AI related vs shipping bad features and then having their companies getting hacked. Their AI on CoderPad was notoriously bad
I mean this is the standard enterprise setup. When they used Microsoft or Google it was the same fear it’s just that the use of Anthropic products by them is high
Is this a response to Goldman banning it lol? Not sure
I think we’re not far away from that point. Maybe in a few years. I find as a dev I read my code less and less. I can still communicate with the agent by explaining the code it’s likely getting wrong, finding perf…
I think people just need to use their time to have fun. Is that not what our use is right now? We’ll find new work to do, new problems to explore, new ways to have fun. Our original use was hunting and gathering. We…
Nope that costs too many tokens lol (also performance is worse compared to their text counterparts from what I’ve noticed….)
Evals and benchmarks lol. Make the eval to see what the AI should be doing, benchmark to see what the whole landscape is doing. Refine and repeat and that’ll make it better
It’ll make sure agents automatically deployed will be much more secure than otherwise.
The health idea of yours is really interesting imo lol. I’ve been playing with two: 1. Knowledge Graph that auto parses vital context out 2. A security feature for agentic memory to secure chunks that don’t determine…
Yeah my company moved to its own version called Gitspoke which has been working a lot better and the UI for reviewers is also nicer
I wonder how much of this is AI related vs shipping bad features and then having their companies getting hacked. Their AI on CoderPad was notoriously bad