Created from a single voice note with Agent Craft
X (Twitter)
Most people add AI tools first, sort control later. I ask one thing…
Most people add AI tools first, sort control later. I ask one thing before a single line of code: does it have an MCP? No MCP, no observability. I write it off. What do you check first? — James, co-founder @ Agent Craft Powered by Agent Craft 🎙️→📱
More content from Agent Craft
- LinkedInBefore onboarding any new AI tool, I ask one question first. Does it have an MCP? If it doesn't, the tool sits outside my agent loop. My agents can't reach it, and I can't drive it from where I actually work. I write it off before we touch a line of integration code. If it does, it goes in read-only. Write access is a separate decision, granted per tool, once I've watched what the agents actually do with it. The tools that fail this test aren't bad tools. They're just not tools I can build on. What's your gate? -- Repost it to your network and follow for more. James Goddard — Co-founder & VP of Engineering Powered by Agent Craft
- TikTokIf you want a real legacy, build systems that outlast you. Juan Mouton on why the best thing a founder can do is make themselves replaceable. #founders #businesssystems #exitstrategy #entrepreneurship
- TikTokA 62% wage premium separates AI-skilled workers from everyone else, and that gap is only widening. The window is open, but it won't stay that way. #artificialintelligence #africantech #aicareer #futureofwork
- LinkedInThe spotless repo tells me almost nothing now. Everyone's codebase looks the same. Everyone can build an API. AI made sure of that. So I had to ask myself: when I hire my first engineer, what am I actually assessing? I was walking around my garden, coffee in hand, turning this over. Then it came back to a conversation with my mentor. He snapped it into focus. The old question was "can they build it?" The new question is "can they think, decide, and direct AI?" That adjustment needs to happen now. Not when you're already mid-hire and have got it wrong. What I'd actually want to see from an engineer today: Can you identify the problem clearly? Can you articulate it? Can you make good decisions under uncertainty? Can you collaborate effectively with AI agents? Most people aren't building anymore. They're getting AI to build for them. That's not a criticism. That's the reality. So when someone hands me a clean repo they didn't really write, I'm not impressed or unimpressed. I'm just uninterested in the repo entirely. The conversation about how they got there tells me everything. Six years in, VP of Engineering at my own startup. Given the choice, I'd hire the engineer who can explain every decision in a commit over the one with the spotless repo they can't account for. -- Repost it to your network and follow for more. James Goddard, Co-founder & VP of Engineering Powered by Agent Craft
- X (Twitter)How do you assess an engineer now? Everyone's codebase looks the same. Everyone can build an API. The real question: can they identify the problem, explain it clearly, and collaborate with AI to solve it? — James, co-founder @ Agent Craft Powered by Agent Craft 🎙️→📱
- TikTokA small coffee shop out-communicated one of SA's biggest government institutions. Juan Mouton breaks down what that actually means for your business. #communication #customerexperience #homeaffairs #businesslessons