Lead Enterprise Cloud Migration and Modernization at OpsGuru. All things Cloud, Data and AI. Love 📚 and 📝 at https://teapawt.net/t.co/t8VtVicjRo
Canada
Joined November 2008
- Tweets5.6K
- Following95
- Followers201
- Likes1.9K
Pinned Tweet
"Having an ability to figure things out is more important than having specific knowledge on how to do something." - Ray Dalio
The biggest value add Microsoft can do right now is build tools for agents to work with extremly large codebases. They have seen the problems first hand and years of knowledge.
Company is serious when they redesign their source control. anthropic.com/engineering/re…
l am so frustrated with OpenAI models even Astra and Sol. Even with so much context or information or even detailed paper UI spec design, they fail to do the thing I want. And I am a heavy claude code user and codex user. So i feel the difference.
I am at a point where I am to abandon codex 100 plan and just go with Claude Code. Btw I use cc at work and codex in my spare time.
Guys, pls don't forget he works for Google. This means, they have implemented bots that can do automatic customer support directly on X. This reply is from his agent.
Coming up soon -- Agents in the Google workspace.
I am getting crazy with AI subscriptions based upon how frugal I am. Switched from plus to pro subscription of OpenAI, just bought Grok Bot. Muse you are next. Let's see how far all this goes.
I am having a feeling that there will be service announced soon to detect bots by all major providers.
Hello computer,
Find every Amazon purchase I ever made.
Call the number of the manufacturer.
Say you’re not satisfied.
Request a refund.
If they decline, threaten to leave a bad review.
If they decline, find the CEO’s phone number and ask him for a refund.
Now, repeat the same process for items I never even purchased and see if you can get them to send me money.
If they ask for a recipient, generate an image of one and send that.
Do not stop until you’ve collected $1 million in refunds.
When you ask your agents to build features, you ask it to build unit tests, integration tests, and also ask it to build acceptance test runners. If the work is done, then the agent should run the acceptance runner and then prove its own work.
Not to mention countless hours wasted while setting up the machine and dev environment. AI gets you there pretty quickly.
That's an ambitious roadmap. The agent orchestration is going to be a very big infrastructure play and something people just dont realize it yet.
Last two days, I spent time building an agent sandbox with all the isolation, with guardrails and budget controls with containers, and etc and google just launches this. Great.
Google launches Open Agentic Orchestrator. agentexecutor.io/
The 5 hour limits on codex should go away.
I have reset that expires today and i cant use it because i reached 5 hour limit.
@thsottiaux there should be an override button. Keep the limits but allow people to override it.
Rulebook to burn through tokens fast, waste hours, and get frustrated with AI agents:
1. Trust agents beyond their actual capabilities.
2. Never cross-check their work.
3. Never verify their claims.
4. Give them unlimited resources.
5. Give them unlimited time to finish a task.
Muse is also doing resets, Grok also doing resets and Claude Code as well. Resets keeps your loyal customers and also is a great way to gauge usage.
It is the easiest product growth hack devised by OpenAI.
