Co-Founder & CEO, Captions Former Design Engineering Lead, Snap
MAR 27 2025
Core Concept
1 Marketable Feature Per Engineer Per Week
"Every engineer ships a marketable product every week — something a user might subscribe just to use."
Marketable = user downloads the app just for this feature
Ruthlessly cut scope, never cut quality
Complaints after launch = the ranked list of what to build next
Volume of shots on goal beats big infrequent bets
Framework
Two Roadmaps: Public vs. Secret
Public roadmap: every competitor has the same list. Wins here are marginal at best.
Secret roadmap: quarterly company-wide brainstorm — all functions vote and rank. Product filters for feasibility.
Biggest wins always come from the secret roadmap
Eye Contact (built first with Nvidia) — viral in every language, millions of views to this day
10M+
Captions users
$100M+
raised
2 days
to build v1
How the secret roadmap is fed
Quarterly all-hands brainstorm: engineering, design, recruiting, marketing, everyone included. Vote and rank all ideas. Product then filters by technology feasibility. Features are announced directly to users — never telegraphed publicly.
Use virality as a prioritization signal
What goes viral on social = something resonant at its core. Post the concept before building anything. Measure sharing. Then decide whether it is worth a full week.
Deep Dive
Technical Debt as Startup Leverage — and the Scope-Cut Loop
Startups should take on technical debt. That is how you out-execute companies 10x your size.
Every piece of debt charges ~1–2% daily interest: bugs, crashes, maintenance. Let it compound to 80–90% and you are just "keeping the lights on."
Ask: is this a problem for the 50th engineer? The 500th? If yes, defer it. That is leverage against a future hire.
Dedicate a full quarter to paying it down — Captions uses Q4 as the infrastructure quarter every year.
One-way door decisions get real care. Two-way doors: ship fast and fix later, always.
"If the company fails, that engineer will never be hired — and all of this won't matter anyway. So we use that future engineer now."
The Scope-Cut Method
For every element in a feature design, ask: What if we remove this? Is the product still useful? Repeat until the next cut would make it useless. That stripped core = the week-one ship.
Example: "Add image to video" could include BG removal, hue/saturation, cloud picker. Cut all of it. Ship: native camera picker, image drops into video, no extra UI. That is useful. Everything else becomes next week's complaint-driven backlog.
PM owns all the way to marketing
A Facebook ad is just a button into your product — it is part of the same funnel. PMs who stop at shipping miss the full user journey. At Captions, PMs are expected to understand paid channels, localization performance, and creative testing. Brian Chesky renamed all PMs to "product marketing managers" for exactly this reason.
Tactics
Snap's Secret Weapon: Design Engineering
10–12 designers at 6,000 employees — no PMs for years. Designers owned the entire product surface.
Evan reviewed every UI change — kept the whole product in his head, enabling true founder-mode control at scale.
"Design engineering" = one person who can design, build, and ship. A startup within a big company.
Build prototypes into the live app; run limited-area tests (two high schools, one city) before committing 500 engineers for six months.
Internal virality = alignment: share a Slack build, watch it go viral internally — VPs and the CEO come to you before you even request a roadmap slot.
Gaurav's personal motto
"The easiest way to be the best is to be first." Being first sets the standard every competitor must chase. Captions was first to ship Eye Contact AI; every major app has copied it since.
Contrarian
Startup Velocity Myths — Debunked
✗Cut quality to ship fasterINSTEAD →✓ Cut scope, never quality. The feature must work flawlessly — it should just do far less than originally planned.
✗Technical debt is bad engineeringINSTEAD →✓ Strategic debt is how startups out-execute large companies. Use it like financial leverage — deliberately, with a scheduled repayment quarter.
✗Build what users ask for to winINSTEAD →✓ The secret roadmap — ideas nobody asked for — is where game-changing wins come from. The public roadmap only keeps you in the race.
✗Snap succeeded because it had the best teamINSTEAD →✓ Snap succeeded because Evan had unmatched user understanding and kept the full product in his head — a 12-person design team running product for 6,000 employees.