My two new books are LIVE! The AI-Native Individual, Made Simple and The AI-Native Company, Made Simple. Practical ways to use AI in your day and your business, with illustrations, diagrams and exercises. Let's go! Get both: lethabolee.gumroad.com/l/plg…
You can end up collecting AI subscriptions before you have a customer. If I were putting together a small software business, this is the kind of setup I would start with. Each tool needs a job.
1. Build: Claude Code + GitHub
Claude Code can help write code, fix bugs and run tests. GitHub keeps the changes and review in one place. Start with one customer problem and a version someone can actually try. You still need to check what the AI builds.
2. Run it: Cloudflare Workers
A place to deploy the app without starting by managing your own servers. I would check the storage, database and usage costs for the product first. Renting a server is also an option, but someone then owns the work of keeping it running.
3. Follow up: HubSpot CRM
Keep the contact, the last conversation and the next step together. A founder should be able to see who is interested, who needs help and which deal has gone quiet. AI can help prepare a follow-up. It should not turn every contact into a target for more messages.
4. Learn: PostHog
Look at where people get stuck, whether they finish onboarding and whether they return. Decide what you need to measure before switching on tracking. Customer information still needs care, even when a tool makes collecting it easy.
5. Deliver and work together: Google Workspace
Gmail, Docs, Sheets, Drive and Meet still have jobs. Customer emails, proposals, shared files and calls do not disappear because you use AI. I would connect AI to a useful piece of that work before replacing the whole setup.
That is a starting point, not a shopping list everyone needs to buy. If a spreadsheet handles your first ten customers, use it. If your existing software already does the job, make it work better before adding another subscription.
For me, an AI-native business should be able to explain what got better for the customer. Fewer tools might be a result. It is not the goal.
What is one tool you pay for that you would struggle to run your business without?
Same in B2B sales. "Can it do X?" on a call is almost never about X. Ask what happened the last time it went wrong and you get the real requirement, and often a bigger deal.
Your next feature idea might already be sitting in a bad review.
Day 9/21: find three real complaints about one job. Ask a potential user about one. Don't ask AI to invent the research.
How did today go? Honest progress welcome. No product-post links here. Keep going.
Sell the feeling, not the product.
People want money for freedom, cars for status, and fitness for confidence.
The real sale is what your product makes possible.