Can't take credit for this, but can't find the original post. One of my favourite image prompts so far for chatGPT. Description in the ALT 👇

Has anyone created a Manus equivalent using Cursor and MCP's such as browser-use? I haven't seen any MCP's that bring in a secondary agent that keep the conversation and code progressing with Cursors agent. Don't think Cursor has released any multi-agent comms due to cost

Why does it feel like google gemini 2.5 pro got neutered on latest cursor update? Tools calls failing left right and center for simple things like reading or writing to files.

Does anyone else think that the model naming convention that has been established is absolutely horrendous? Reminds me of how designers would name photoshop files back in the day. Theres a brand play to stick out from the rest here! (Apple intelligence isn’t it 🤪)

Does anything beat Gemini 2.5 Pro when it comes to coding? Gemini 2.5 Pro far surpasses Sonnet 3.7 – It feels pretty AGI'ish, even has the laziness to make it believable. Just wondering if o3/o4-mini actually surpass it when it comes to code.

Has anyone else been experiencing outages with Cursor over the last few days. Seems to just sit there hanging at generating... Restarts, etc don't seem to solve the issue.

Wowza! Figma just killed Illustrator, Framer, Aura and more. Well played! #figmaconfig2025

Google IO is a great example of how fast tech is moving and how cut-throat AI is. Every day people’s products are turned into small ‘features’ by the big players, essentially replacing them overnight. What are AI ideas that you think are niche enough this wont happen?

Gemini 2.5 pro and Claude 4 are all you need. Opus, o3 and more expensive models aren’t worth the cost for development. Convince me otherwise

This is wild. Close your eyes and you can actually feel something from their responses. Not perfect but hell it's come a long way is such a short time! Kudos ElevenLabs

Imagine if Apple blew everyone away at WWDC this year with a on-device llm that was competitive to gemeni 2.5/sonnet 4. So far on-device quantizes models haven’t been able to get anywhere near cloud capable llms.

Large software teams are dead. AI integration is mandatory and DANN© - counter to your previous post, audience is everything. Trusted following and network will surpass any talent in art and development now.

Is there any pipeline that exists where you can go from design system component (figma) to generated front-end code (storybook, react components) that actually look and analyze your components and custom icons properly? (Thinking offset casual game UI style not shadcn everything)

GPT-5 is capable but lazy AF. Needs some tuning.

How can I get an agent to respond to cursor agent in order to have it continue from a stop loop? I find many prompts are “please continue” as the result is 1/20th baked.

OpenAI’s attempt to humanize their realtime ChatGPT conversations with #gpt5 have turned their functional Ember character into that of a male flight attendant personality 🤦🏼‍♂️

Just had Sentry detect a app crash, review the crash, provide a solution for the crash, and deploy a PR. All without having to lift a finger 👆. What a wild world we live in. Nice work Sentry!

It's customer experiences like this that really make a difference in gaining / retaining customers. Huge shout out to Expo - I wish there were more companies out there with this mentality!