Ryan Smith

or this: https://www.linkedin.com/posts/djoume_something-is-happening-to-me-im-now-activity-7411876157001879553-9cKQ?utm_source=social_share_send&utm_medium=android_app&rcm=ACoAAAAAjVYB9OAEZJ3PsjRhyrHDygLPLM-1n0c&utm_campaign=share_via

I'd add, while Co-pilot was falling way behind up to the last few months, they are less behind today. They stuck to an AI assistant approach a bit long, while other like Claude Code went more fully agentic. The way I can give a brief or vague prompt and Claude can spin up multiple specialist agents to execute and see complex chunks of tasks all the way through - research, plan/reason, sprint planning, TDD, code (any language or framework), run E2E and debug, then run CICD/DevOps is, like alien technology

Aleks Pamir - Microsoft Copilot opened up enterprise to LLMs. Their security and assurance is why we're here today. Every power user I know that has tried the agentic AI products has Claude Code. They may also use Antigravity (Google) or Codex (OpenAI), or Cursor with Opus 4.5, but it's in the mix. Without knowing the decision-makers, I can't respond to their objections. You may be locked in because Microsoft is great at doing so. However, some light digging around and using the tools will expose advantages - including the SWE bench tests, code quality, error and bug minimization. I was on a call this morning about the verbose attribute which writes more detailed specs in Claude. Alternatively, you could point to Andrej Karpathy (fmr Open AI who coined "vibe coding") who has been posting about his Claude use, or Andrew Ng making comments this week about how impressive Claude Code is executing. A few months ago, it was the parallelization of agents and additional memory structures. What 'appears' to be the advantage since the November release of Opus 4.5 is the extra reasoning bump that is improving code quality.

seeing Teresa Torres' posts on LinkedIn lately give me a lot of hope. There were some dark days there in 2023 and 2024. I recall a number around 400,000 PMs had been laid off from 2022-2024. Salaries were (are still) on a downward pressure from the over supply in many regions. That's changing fast for AI PMs, who are very few, and getting double their prior comp. The new product builder role ships far more value. Now we need to upskill everyone, especially in Canada, where we tend to have a lot of branches and tier 2 opportunities.