You can’t teach someone to start a startup. You learn it by doing. Paul Graham’s new essay makes a quietly radical case: the best prep for founders isn’t an “entrepreneurship” curriculum or a business plan competition. It’s getting genuinely good at building things, and building them out of real interest, not because you’re chasing a startup idea. #paulgraham #firsttimefounder #ycombinator #entrepreneurship #startupadvice

They told them to leave for San Francisco. They built it in Ottawa instead. Builders Collective started because a group of students kept watching sharp people get told the same thing: the opportunity is somewhere else. So instead of asking where to move, they asked why they couldn’t just build the community they wanted, right where they already were. Now it’s students shipping real products. An AI resume builder. A webring linking Ottawa’s builders. A farm robot with $225K in letters of intent. The advice isn’t wrong. 57% of U.S. venture dollars went to the Bay Area in 2024. Geography matters. But there’s a difference between “SF might be the best place to scale this” and “you can’t start until you’re in SF.” One is a strategy. The other keeps you in planning for years. Builders Collective proves that you don’t need an ecosystem’s permission to start. You can just build the ecosystem around you. Full story on byvi, link in bio #ottawatech #buildinpublic #studentfounders #startupcanada #builderscollective

A 2-month-old startup, with a garage for a factory, just reminded me how a company is actually supposed to start. Most founders build first and understand their customer later. This team at Tandem Robotics did the opposite. Take their farm robot away for maintenance, and farmers call asking how fast they can have it back. But the robot isn’t the story. It’s how these founders earned the right to build it, before they wrote a single line of code. There’s so much any early founder can steal here, no matter what you’re building. I broke it all down. Swipe to the end, the last slide is the part most founders learn too late. #robotics #organicfarming #startuplessons #farmrobotics #farmrobots

At @shopify Builder Sundays, we watched 8 builders demo what they made. The biggest takeaway? None of them waited for the perfect moment. They put something into the world before they had the answers. 👏👏👏 If you’re building in Ottawa, Montreal, or Toronto, this is a pretty powerful place to start. Full story on BYVI. Link in bio. #BuilderSundays #Shopify #OttawaTech #StartupLife #indiehackers

The POV you all been waiting for. Link in the bio for the full story. #ycstartupschool2026 #ycombinator #techstartups #ycstartupschool #yc

90% of tech startups fail, and the odds only get tighter when you try to cross borders. The founders who break out almost always find international revenue early. Megha Kandoi shares a 12-month playbook for global expansion for startups on a small budget: 1. Pilot where the pain is sharpest, not where the GDP is largest 2. Use “bridge markets” (Netherlands, Singapore, Australia, Germany) to de-risk the leap 3. Borrow distribution through accelerators and channel partners before you buy it 4. Build remote-first with Employer-of-Record tools like Deel or Remote 5. Stack free infrastructure: grants, trade commissioners, credit insurance About Megha: Supported 1,200+ startups at MaRS. Leads @techtoorg – Canada’s largest and most connected tech community. Has scaled companies across North America, Europe, India, and Australia. Read the full playbook at the link in bio. #globalexpansion #internationalexpansion #scalingstartups #bootstrapping #techstartups

The hardest part of building a startup is never the part you plan for. Kalanick’s $1.7B robot kitchen can build 300 bowls an hour, until it has to bag them. Then it drops to 200. A floppy plastic bag beats a machine that weighs food to the gram. That’s the “fries in the bag” problem: the boring step you already handed off, assuming it would sort itself out. It’s usually the one that costs you. Daniel Zhao (Head of Finance at Landbase, previously co-founder of LavaReach) on the friction nobody puts on a slide. Link in our bio to the full article. #scalingstartups #traviskalanick #startupadvice #robotics #foundermindset

An AI can now learn your product by watching your user use it once. On July 21, Anthropic shipped “Record a skill” in Claude Cowork: record a task, talk through it, and Claude turns it into a skill it runs on command. And it points at a whole category of software: products whose value was a nice front-end wrapped around a task people repeat. If that’s you, the interface just stopped being the thing worth defending. Here are 6 things a screen recording can’t capture, and where I’d build now: 1. Proprietary data that compounds 2. Being the system of record 3. Real network effects 4. Structural accountability 5. Distribution and brand 6. Being the tool the agent calls Full breakdown on BYVI (link in bio). #ClaudeAI #agenticAI #SaaS #startupmoat #anthropic

How to get your product recommended by AI shopping agents (ChatGPT, Perplexity & Google AI Mode) AI agents now decide what shoppers buy – and they don’t reward ad budgets. They reward clean product feeds, complete data, real reviews, and third-party trust signals. 8 moves to get into the agent’s recommendations: 1. Get into every feed – Google Merchant Center, Shopify Agentic Storefronts, ChatGPT & Perplexity 2. Make your product data machine-perfect – titles, GTINs, every spec spelled out 3. Add Product, Offer & Review schema (JSON-LD) to every page 4. Turn reviews into ranking fuel – count + rating flowing into your feed 5. Build third-party citations – Reddit, roundups, expert blogs (your Perplexity edge) 6. Exploit the rich-media gap – add video & 3D before everyone else 7. Write like buyers actually ask – conversational, comparison-friendly content 8. Measure what’s working – Search Console’s AI reports, then double down Save this for your next feed audit Link in our bio for the full playbook. #AISEO #AgenticCommerce #ecommercemarketing #ChatGPTShopping #ProductFeed