Industry
Startups and venture-backed products
Early-stage products have one job: produce evidence that the thesis holds, before the runway ends. The engineering discipline is knowing exactly where to be rigorous and where to be deliberately, temporarily cheap.
Workflows
The work that actually happens here
Named the way practitioners in this sector name them, because that is the vocabulary a delivery team has to learn before it can be useful.
- Thesis to testable scope
- Converting an investment thesis into the smallest product that produces a real signal.
- Instrumented launch
- Releasing to real users with the measurement in place from the first day rather than added later.
- Iteration on evidence
- Weekly cycles driven by usage data and user conversations rather than a fixed roadmap.
- Investor and board reporting
- Metrics produced from the product itself rather than assembled in a spreadsheet each month.
- Technical due diligence
- Being able to withstand a diligence review at the next raise without an emergency remediation sprint.
Constraints
What changes the architecture
These are the reasons a design that works elsewhere would be wrong here.
Runway sets the timeline absolutely
Scope must fit the months of runway available with margin. This constraint outranks every architectural preference.
Deliberate debt, documented
Shortcuts are correct at this stage — but only where they are chosen consciously, written down and bounded. Undocumented debt is what fails diligence.
Some things cannot be deferred
Authentication, tenancy isolation, payment correctness and personal-data handling are extremely expensive to retrofit. These get built properly the first time.
Diligence will happen
At the next raise, someone competent will read the code, the security position and the licence compliance. Building for that is cheaper than remediating for it.
Integrations
Systems we would expect to meet
- Stripe for payments and subscriptions
- Product analytics and session tooling
- Authentication providers
- Error monitoring such as Sentry
- Customer messaging and support tooling
Trust considerations
What users and regulators need here
- Early users are doing you a favour; be honest about what is unfinished rather than presenting a facade.
- Personal data collected early is a liability that compounds — collect only what the thesis requires.
- Tenancy isolation must be right from the start; a leak at this stage is existential rather than embarrassing.
- Metrics shown to investors must be reproducible from the product.
Architecture
A shape we would actually propose
A starting sketch, not a template. Discovery is what turns this into a specific design with recorded trade-offs.
- Stack
- A deliberately boring, well-understood stack — Next.js, Postgres, managed hosting — chosen so effort goes into the product.
- Non-negotiables
- Authentication, row-level tenancy isolation, payment idempotency and data handling built correctly from day one.
- Deliberate debt
- Shortcuts recorded in an ADR with the trigger condition that will require revisiting them.
- Measurement
- Product analytics and outcome instrumentation present at launch, not retrofitted.
- Diligence readiness
- Dependency and licence scanning, secret scanning and architecture documentation maintained continuously.
Outcomes
What good looks like
- A first release inside runway that produces a real signal
- Technical debt that is chosen and documented rather than accumulated
- Investor metrics reproducible from the product
- A technical position that survives diligence at the next raise
Related
Where to go next
Next step
A startups and venture-backed products project starts with the constraints
The Project Architect asks sector-specific questions about data, regulation and integration before it asks about features.