Field notes.
Short essays on building software without a team.
§073
2026-10-03
Naman Barkiya
Postgres vs MongoDB for an MVP: which should you use?.
Postgres is the default for almost every MVP in 2026 — JSONB columns cover the flexible-schema case MongoDB used to own, and relational constraints catch data bugs a document store lets through silently. We've shipped Postgres on all 15 products in our portfolio. MongoDB only wins for a single, relation-free, high-write collection.
- mvp
- stack
- database
- engineering
§072
2026-10-01
Naman Barkiya
App Store rejection: 4 reasons and the pre-launch checklist.
Most App Store and Google Play rejections come from four causes: broken or incomplete functionality (Apple's Guideline 2.1), missing or inaccurate privacy disclosures (Guideline 5.1.1 / Data Safety), misleading metadata (Guideline 4.3), and over-broad permission requests. First submissions fail more often than resubmissions, and each rejection costs a full resubmission cycle — 24 hours to two weeks depending on the store and the flag. A 48-hour internal test pass before submission catches nearly all four.
- mvp
- launch
- mobile
- checklist
§071
2026-09-29
Naman Barkiya
The $0-cash MVP: when paying a developer in equity actually works.
Almost never, and the studios who do take equity price it as a bet, not a favor. Real deals cluster around a true technical co-founder (40-50%, pre-product), a studio discounting an invoice for a founder with existing traction (under 2%), or a part-time freelancer with an existing relationship (1-3%) — not a stranger building a full MVP for free stock with no cash and no traction.
- equity
- mvp
- hiring
- freelancer
- cofounder
§070
2026-09-27
Naman Barkiya
Should investors or buyers get source-code access?.
No, not by default. When an investor or enterprise customer asks for source-code or server access, the safe answer is supervised, NDA-gated review on your terms — not a repo invite and not root credentials. Code escrow and third-party audits (Black Duck, Veracode) are the fallback when a flat no isn't realistic.
- due-diligence
- investors
- security
- mvp
- contracts
§069
2026-09-25
Naman Barkiya
How much does it cost to add payments to your MVP in 2026?.
Payment processing on an MVP costs a per-transaction fee of roughly 2-6% depending on the processor, plus 3 days to 4 weeks of engineering depending on whether you use a hosted checkout or build your own. Stripe (2.9% + $0.30) and Razorpay (~2% domestic) are the cheapest but leave tax compliance to you; Paddle and Lemon Squeezy (~5% + $0.50) act as merchant of record and remove that problem entirely.
- payments
- stripe
- razorpay
- mvp
- cost
§068
2026-09-23
Abhiraj Sakargaye
You Already Vibe-Coded Most of It. Here's the Handoff..
A good handoff starts with an intake audit before anyone touches the code, a written decision on what's kept versus replaced, and a mental-model transfer session -- only then does a real team start committing, under branch protection and PR review from commit one.
- ai
- vibe-coding
- process
- mvp
§067
2026-09-21
Naman Barkiya
Subscription, usage-based, or hybrid: how do you price your MVP?.
Use subscription pricing when your value is per-seat and usage is low-variance, usage-based pricing when your own cost scales with consumption, and hybrid — a base subscription plus metered usage — when you're pre-PMF and need to de-risk both a scary price and an unpredictable cost. Most first-time SaaS founders should start hybrid.
- pricing
- mvp
- saas
- startups
§066
2026-09-19
Naman Barkiya
MVP vs POC vs prototype: which one should you build first?.
Build a proof of concept when you're not sure something is technically possible, a prototype when you need to show how it feels, and an MVP when you're ready to find out if real users will pay for it. Skip straight to the MVP when both the technical risk and the UX risk are low, which is the normal case for most SaaS and marketplace ideas.
- mvp
- poc
- prototype
- product-strategy
- startups
§065
2026-09-17
Naman Barkiya
How many developers does an MVP actually need?.
Most first MVPs need two to four people, not a department: one or two generalist engineers who can move across the stack, a designer only if the interface is the differentiator, and a founder who owns the product decisions. Specialization — dedicated devops, QA, or a full-time engineering leader — pays off once real usage creates a specific, recurring bottleneck, not before.
- hiring
- mvp
- team
- engineering
- startups
§064
2026-09-15
Naman Barkiya
AI Code Review Before Hiring a Dev Shop? | SingleBit.
Run three checks before you hand a vibe-coded codebase to a dev shop: scan git history for committed secrets, run a dependency audit for stale or unmaintained packages, and pull a real test-coverage number. Ten minutes with an AI code-review tool catches most of it; a shop's paid audit ($3,000-$15,000) catches the rest.
- ai
- vibe-coding
- security
- mvp
- hiring
§063
2026-09-13
Naman Barkiya
Is Your Product Just an AI Wrapper? A Vendor Lock-In Checklist.
Run this test before you build: if the model vendor disappeared tomorrow, what's left? If the answer is 'an API key and a system prompt,' it's a demo, not a product. A real business owns proprietary data, a workflow, or distribution the vendor doesn't have — check that before writing the first prompt, not after a competitor's changelog makes the decision for you.
- ai
- startups
- vendor-lock-in
- mvp
- product-strategy
§062
2026-09-11
Naman Barkiya
Does Your MVP Need ADA/WCAG Accessibility Compliance? What It Costs.
Every public-facing MVP is already in scope for ADA Title III accessibility lawsuits — no revenue or user-count threshold to cross. Plaintiffs filed 3,117 federal accessibility lawsuits in 2025, up 27% year over year; courts treat WCAG 2.1 AA as the benchmark. Building it in from day one costs $800-$5,000; a demand letter after the fact costs $15,000-$110,000+.
- compliance
- cost
- mvp
- accessibility
- ada
- wcag