Meta hiring MongoDB's CEO to lead enterprise AI makes sense. Getting agents to work inside a company is harder than making them look smart in a demo. Permissions, failed API calls, retries, audit logs, and recovery all need to work. A better model won't fix broken infrastructure.
We believe superintelligence will create significant new opportunities for all people and businesses. Meta already serves billions of people at scale and helps hundreds of millions of businesses reach customers. Today we are starting the next major pillar of our business, Meta Enterprise Platform, to help businesses use AI to grow and transform in new ways as well.
@ZenModeJon@Google I am Maaz from Pakistan.
I like to talk about AI systems, agents, evals and RAG.
I like football and cricket. Inwould like to say I am fan of Man Utd, though it is depressing.
100 followers in 8 days.
Small number in the grand scheme, but still cool to see people interested in what Iβm thinking about.
Iβm an AI Engineer writing about production AI systems: agents, evals, RAG, and what it actually takes to ship this stuff.
Looking to connect with more engineers, builders, and founders working on the same problems.
@elshayib_ A good model isn't enough if the tools keep failing or developers have to fight the API to get anything working. The surrounding infrastructure matters just as much.
@CodeWithTamara Prompting matters, but I'd rather see how someone manages context, handles failed tool calls, and tests whether the agent actually works outside a demo.
@sebuzdugan An agent that needs to update one record shouldn't have permission to modify the entire database. Give it access to what the task requires, nothing more.
@rileybrown The real test is whether an agent can finish a task, recover when something breaks, and know when to ask for help without needing someone to watch every step.