Startup fundamentals
Why now
A real reason this was hard recently and is newly tractable — tech, regulation, behavior, cost, or infrastructure.
Great ideas have a why now: a reason this was hard or impossible recently and is newly tractable. Without one, you are often competing against patient incumbents on their home field. With a false one ("mobile exists, therefore my clone wins"), you still lose.
As of when this was written, AI is the loudest why now. That does not make every "GPT wrapper" a good idea. It means some workflows where language, classification, or generation was too expensive or too bad are newly automatable.
The big idea
A locked door is not a business. A newly unlocked door might be. Why now is the unlock: a technology, a rule, a habit, a cost curve, or a piece of infrastructure that did not exist last time someone tried.
The analogy breaks when the door is open for everyone, including Google. Then why now is necessary but not sufficient. You still need founder-market fit, a wedge, and something incumbents cannot easily copy — proprietary data, a workflow, or a distribution channel. See competition and moats.
Walking through it
| Driver | Example shift | What becomes possible |
|---|---|---|
| Technology | LLMs, cheap GPUs, mobile, cloud | New UX; automate previously human work |
| Regulation | Open banking, privacy laws, new compliance | Products that wrap new rules |
| Behavior | Remote work, creator economy, QR payments | Distribution and habits flip |
| Cost curves | Storage, bandwidth, inference | Price points that unlock mass use |
| Infrastructure | App stores, Stripe, AWS, Shopify | Startups compose instead of rebuild |
"AI customer support" is a tarpit shape if your only insight is "LLMs can answer questions." It becomes interesting if you have a vertical's ticket corpus, evals that beat generic bots on that vertical's policies, and a sales motion into that vertical. The model is the commodity. The workflow and data are the company.
The construction-reports idea works the same way: tablets on site plus cheap computer vision is a why now. Without those, you were selling clipboards.
Common mistakes
- A fashion why now. "AI is big" is a weather report, not a reason this product can exist now.
- Ignoring that incumbents got the same unlock. If the door opened for everyone, name what they still cannot do.
- Skipping the last-attempt autopsy. If a 2016 company died doing this, what actually changed?
- Confusing cheaper compute with a buyer. Cost curves matter only if they unlock a price someone will pay.
Check yourself
Your pitch is "GPT can write marketing copy." What is missing from the why now?
A reason this wrapper wins — data, workflow, distribution — now that the model is a commodity.
Name one non-tech why now.
A regulation change, a habit shift like remote work, or new infrastructure such as open banking.
Key takeaways
- Why now is a reason the idea is newly tractable, not a trend label.
- Drivers: technology, regulation, behavior, cost curves, infrastructure.
- AI unlocks some workflows. It does not make every wrapper a company.
- If incumbents got the same unlock, you still need a wedge they cannot copy.
Further learning
How to Get Startup Ideas (Paul Graham)How to Get Startup Ideaspaulgraham.comRelated product framing: AI product strategy.
Back to the map: startup ideas.
Related
Startup ideas and why now
Where good ideas come from, which shapes to avoid, and how to judge an idea before you quit your job.
Tarpit ideas
Attractive, crowded shapes — social networks, dating apps, Uber for X — that keep attracting founders and rarely succeed.
Competition and moats
7 Powers, network effects, switching costs and aggregation theory, and what defends an AI company.