How to choose between big tech, startups, and founding your own company

Executive overview

Most career advice treats big tech vs. startups as a values question. It's actually a skills question. Each environment teaches a different set of capabilities — and none transfers cleanly to the others.

Jacob Bank left a six-year run at Google (Gmail, Calendar) to found Relay.app, a workflow automation tool for cross-tool tasks. The move forced him to unlearn as much as he'd learned at Google.

The environment you choose should match the specific skills you want to build — not just the prestige or the paycheck.

Choosing the right environment for the skills you want

• Zero-to-one product building: go to an early stage startup, not Google
• Hyperscaling a product: go to a growth stage company
• Operating at billion-user scale: go to Google
• Building a talent network: join a larger company with more people
• Building a peer founder network: start your own company
• Be deliberate — pick the company that will teach you what you actually want to learn

What Google teaches (and doesn't)

• The bar for shipping at 2 billion users is extremely high; that discipline is valuable
• Advantages at Google: data, hardware, algorithms
• Disadvantages: press scrutiny, regulatory pressure, internal bureaucracy, inertia
• At Google, the job is redirecting a "flowing river" of users and requests
• At an early startup, there is no river — you have to find the first drops

What the startup experience teaches

• Early users are people, not dashboard numbers — you know their names and their problems
• Decision-making is fast and reversible; no inertia to overcome
• Intellectual honesty is easier: when something isn't working, you can just change it
• Many skills from big companies must be actively unlearned
• Starting is now easier than a decade ago: Stripe Atlas, Gusto, Rippling remove old friction

How Relay.app came to be

• Bank observed at Google that cross-tool workflows were often more important than within-tool ones
• Getting an email into a CRM matters as much as composing the email in Gmail
• Solving cross-tool workflow problems from inside a big company is structurally hard — incentives push toward existing products
• Relay automates the repeatable parts of multi-tool, multi-person work; humans handle the parts requiring judgment
• The final product idea emerged only after a year of building and discarding five or six experiments

Traction and the three existential questions

• Traction benchmarks: 10 people who love it (YC definition); 10 unaffiliated paying customers (Jason Lemkin / SaaStr)
• Relay had both, plus week-over-week growth in users, revenue, and usage
• The three questions that matter: Are you solving a real problem? Has your product actually solved it? Can you find more users at scale?
• Everything else — hiring, finance, marketing, sales — is difficult but secondary

Burnout is about progress, not hours

• Working 50 hours on something that goes nowhere burns people out fast
• Working hard on something customers love and that grows rarely causes burnout
• Solving each incremental challenge creates energy and motivation for the next one