Created from a single voice note with Agent Craft
LinkedIn
Six years in. VP of Engineering. Co-founder. The hardest thing I have…
Six years in. VP of Engineering. Co-founder. The hardest thing I have ever done was write down everything I didn't know. It had nothing to do with code. It was looking honestly at my own weaknesses. When you step into a role like VP or Founder, the instinct is to prove you belong. To show up with answers. But the real starting point is the opposite. You sit down and list everything you don't know yet. Everything you'd need to learn before you can genuinely cover the role. It stings to write it down. Admitting the gap feels like admitting you aren't ready. That list is the whole job. I built mine early. It's the reason I could walk into a startup, take the engineering lead, and build AI agentic software that actually scales and stays reliable instead of falling over the first time real users touch it. Almost a year in, and the biggest improvements haven't come from a shinier tool. They've come from staying honest about the gap, then deliberately closing it. I still catch myself avoiding the list. Then I open it, add the next thing I don't know, and get back to work. Do you have a list ? -- Repost it to your network and follow for more. James Goddard — Co-founder & VP of Engineering Powered by Agent Craft
More content from Agent Craft
- X (Twitter)The blocker isn't access. Everyone has the tools. The blocker is that nobody wants to look like they don't know something, so they wait until they feel ready. That day never comes. The company I was at handed out Copilot licenses to everyone. Almost nobody touched them. I was one of the first to actually use it. That was my gateway into AI, and eventually into Claude. While I was writing features and running tests with AI in the background, most of my co-workers were still hand-writing every line. The tool was sitting right there in their editor. They knew it. They'd seen the posts, heard the stories. They just didn't open it. I've seen this most with South African developers I've worked with. Sharp people. They have the access. What they don't have is a reason to start. Here's what I think is actually going on. The tools are easy to reach. The real blocker is fear of the learning curve. The part people miss is that these tools teach you while you use them. You don't need a course first. If you sit down in front of Claude and have no idea what you're doing, tell it exactly that. Type "I'm struggling to understand what I'm doing with you" and watch what happens. It'll give you step-by-steps. It'll point you at what to go read. It closes its own gap as you go. The thing you're scared to learn is the thing that teaches you. So the fix is simple. Open the tool today and be bad at it for an afternoon. Look silly for a week. That's the whole entry fee. — James, co-founder @ Agent Craft Powered by Agent Craft 🎙️→📱
- ThreadsMost people think AI content means paste, iterate, repeat until it's okay. Is that your experience too, or have you found a setup where the first output is actually usable?
- X (Twitter)Buying tells you where the market was. Craving tells you where it's going. Most marketing decks read the wrong one. Consumption data is a lagging indicator, and half the marketing decks I see are built on it. Look at what people bought last quarter. Look at what they bought last year. Then pour budget into more of the same. That's the whole plan for a lot of brands right now, and it's backwards. The brands that win are reading want. Here's why the distinction matters when someone's writing a cheque. When an investor tells you "the metrics aren't quite there yet," that's a code phrase. What they actually mean is they can't see how another rand of capital turns into company value over time. They want the return to be likely, not just possible. If your marketing is oriented entirely around historic purchase data, you're showing them the past and asking them to fund the future. Those are different things. You're pattern-matching what already happened while the demand that pays your next raise is forming somewhere you're not looking. Craving shows up before the spend does. It shows up in what customers ask for and can't find, in the problem they keep describing that nobody's solved cleanly. Matching a customer who has a problem to a business that answers it is the whole game. You do that by watching the want. Extrapolating a spreadsheet of receipts won't get you there. A dashboard full of consumption metrics feels rigorous. It reads as receipts. But it's a rearview mirror dressed up as a windshield. I'd rather have one honest read on what a customer craves than a year of clean purchase history. The purchase history is safe and it's late. The craving is messy and it's early. And early is where your next raise comes from.
- X (Twitter)Two investors wrote me million-dollar checks off nothing but me talking. One of them was too busy to even take my call afterward. Here's how it happened. I'd presented on stage at a Jason Calacanis launch event. Someone recorded it and put the video on YouTube. Later I ran a crowdfunding campaign and dropped that video onto the platform. One of the investors watched it and committed a million dollars off the back of it. When I reached out afterward and offered to hop on a call, he told me he was too busy. Too busy. He'd already sent the money. That one shook me. But it wasn't a fluke, because a second one landed the same way. Tim Draper wrote me a million-dollar check off nothing but a PowerPoint. No incorporated company. No product. No prototype. No customer traction of any kind. An idea, a deck, and me in the room pitching it. Two checks. Two million dollars. Both off nothing but me talking. I met Tim Draper in person for his. The crowdfunding investor I never spoke to. And that's the part that stays with me. The persuasion, the presence, the thing that moved both of them, I was carrying it into those rooms before either check existed. I already had the network. I already had the smarts. I already had the deck. The only thing separating that version of me from the flat, stuck version is whether I believe it on the day. Tim Draper wrote a million-dollar check off a single PowerPoint. What are you talking yourself out of that's better than a PowerPoint? Building: AgentCraft (dot) ai Grow your personal brand in minutes per day with voice notes
- X (Twitter)You will look foolish. Repeatedly. That discomfort is the fastest teacher you'll ever have, and it's the real reason most people never build anything. Six companies from zero. That's the number people fixate on when they ask me how I do it, like there's a secret I'm holding back. There isn't. The method is almost embarrassingly simple. Pick an area you're actually interested in. Could be a customer problem you keep running into. Could be a piece of technology you can see shifting. Could just be an idea that won't leave you alone. Here's the part that trips people up: you don't need to know much about it. Domain expertise is not the entry ticket. It's something you acquire on the way in. Then start building. Whatever that takes. People, technology, spreadsheets, phone calls, doesn't matter. You just start working the problem. And while you build, you socialize it. Heavily. You take it to your target audience, to potential customers, to anyone you think is relevant, and you talk about it long before it's polished. Then you shut up and listen. Not politely. Intently. This is where it gets uncomfortable, and I want to be honest about that part because most people won't say it. Your idea will be half-baked. You'll discover there are already three solutions to the problem you thought was yours alone. Someone who's worked in the space for fifteen years will explain, kindly or not, why the thing you're excited about can't work. Sit in it. The willingness to look foolish while you listen is exactly what lets you learn fast enough to build something real. People who need to look competent from day one never get past the front door. Pursue it doggedly and you will succeed. That's not motivation, it's just what I've watched happen six separate times. And there's a moment I've now lived through more than once that still surprises me. At some point I find myself standing in front of a room of people from that industry, walking them through how my solution solves their problem. They're nodding. They look genuinely appreciative. And these are often the same people who laughed at me a year or two earlier because I clearly didn't know what I was talking about. I still didn't have their years of experience. I had something else. I'd listened harder than anyone else was willing to, and I'd kept moving while they explained why it couldn't be done. If you're sitting in a job right now sensing the traditional path isn't yours, the barrier you think you have isn't knowledge. It's the fear of the foolish stretch in the middle. Start before you're ready. Listen like your company depends on it, because it does. What's the area you keep circling back to but keep telling yourself you don't know enough to build in? Building: AgentCraft (dot) ai Grow your personal brand in minutes per day with voice notes
- FacebookHere's the mistake that keeps AI content tools stuck in an endless revision loop: they treat every piece of content as a blank slate. You paste an input. You get a draft. It's off. You tweak. You paste again. You get another draft. Still not quite you. Twenty rounds later you've spent more time editing than you would have spent writing it yourself. That's the pattern most tools train you into. And it's why so much AI content gets rejected right now, by readers and even by the people building these tools. Agent Craft was built to break that loop before it starts. The fix happens in onboarding, not generation. The team admin establishes the full context, voice, positioning, campaign objectives, audience, once, at the source. Every piece of content then draws on all of that at once, so the first draft lands where it needs to, not after a dozen rounds of correction. The difference is simple. Other tools iterate toward your brand. Craft starts there. If your team is burning hours fixing AI drafts, the problem probably isn't the model. It's what the model was never told.