A founder messaged me last week: 'I have $10k and an idea. What do I build first?' Wrong question. The right question is: what do I build that proves people will pay, before the money runs out? Here's the sequence I'd follow.
Step 1: A landing page that sells ($0–1.5k)
Before a single feature exists, you need a page that explains the outcome you deliver and collects emails. Not a 'coming soon' page — a page with an offer, a price, and a buy button. If nobody clicks 'buy' on the fake door, the product behind it doesn't matter yet.
Step 2: The smallest sellable MVP ($4–6k)
One core flow, done extremely well. Not ten features — the one thing your user pays for. In 2026, an AI-assisted senior developer ships this in two to four weeks. The scoping conversation is where most of the value is: every feature you cut is a week of runway you keep.
Step 3: One AI feature that saves you time ($1–2k)
- Support chat trained on your docs — kills repetitive questions
- Smart onboarding that adapts to the user — lifts activation
- Content automation for your marketing — buys back your evenings
Pick one. The point isn't to impress investors with 'AI-powered' — it's to buy back your time so you can spend it selling.
What I would NOT spend on
A logo suite and brand guidelines. A mobile app before the web app proves demand. A 'growth marketing agency' retainer. Custom illustrations. Any feature a user hasn't asked for twice. That list is where $10k budgets go to die.
Spend your first $10k on proof, not polish. Polish is what you buy with the revenue the proof generates.
If this is the situation you're in, this is literally what I build. The MVP sprint exists for exactly this budget — and the founding rate means it costs less than the numbers above.