SaaS Rock Sells Structure. Solos Need Speed.

Opinions came pre-installed (someone else's)

SaaS Rock style kits arrive opinionated: this ORM, that auth pattern, these folder rites, those SaaS playbooks. Lovely for teams needing alignment — everyone reads the same map. For solo use, it means a month of deletion: ripping out team workflows, unwiring enterprise patterns, simplifying abstractions built for standups nobody holds.

The opinions weren't wrong. They were someone else's — tuned for organizations, applied to an individual. Every preset removed costs the time presets were supposed to save, plus the archaeology of understanding what breaks when team assumptions leave.

Saasrock alternative thinking: conventions should match headcount. Solo conventions (thin files, one boolean, flat deploys) differ structurally from team conventions — and borrowing theirs costs more than building solo.

Opinionated vs solo-minimal

Three details make borrowed opinions expensive:

1. Deletion requires understanding. Removing team workflows demands comprehending them first (what breaks if roles leave? which tests assume tenants?) — learning costs for knowledge you'll discard. Building minimal skips both tuitions: nothing to learn, nothing to remove. Unused opinions bill twice (learn + delete).
2. Defaults anchor silently. Preset choices become "how things are" without deliberate verdicts (ORM because present, patterns because provided) — defaults chosen by absence of decision. Solo stacks need defaults chosen by presence of judgment. Inherited opinions are unexamined commitments wearing convenience.
3. Playbooks assume standups. SaaS rituals (sprint shapes, review queues, release trains) optimize multi-human coordination — solo, they're process theater with no audience. Rituals without participants waste the scarcest resource (founder attention) on coordination with nobody. Audience of one needs no choreography.

The conventions posts built solo law. This one declines borrowed law.

Saasrock alternative solo doctrine

Minimal law, documented decisions, agent-readable thinness — opinions owned, never inherited.

Step 1 — Adopt the solo test per default.

"Does this default earn its place THIS quarter for ONE human?"
  auth-with-teams → NO (boolean suffices) → delete
  thin routes → YES (review speed) → keep
  invitation flow → NO (no invitees) → delete
  agent guardrails → YES (builders need fences) → keep

Quarterly culling keeps the stack honest (defaults re-earn or depart, documented per verdict). Inherited presets face the test immediately (most fail at solo scale — delete in week one, not month three). Owned minimalism compounds: every kept default carries a reason, every reason survives questioning.

Step 2 — Document decisions, not just code.

DECIDED: raw SQL + provider verbs (not Prisma).
BECAUSE: 2am readability beats autocomplete magic (incident #4 taught this).
REVISIT: team of 3+ engineers, or explicit ORM-shaped pain.

Decision records convert opinions into reviewable history (why, from what evidence, when to revisit) — future-you (and agents) inherit reasoning, not just rules. Five lines per decision, written once, consulted forever. Undocumented opinions decay into mystery; documented ones age into wisdom.

Step 3 — Keep files thin enough to re-decide.

# re-decision cost ≈ file size: 40-line routes re-decide in minutes,
# 400-line services re-decide never (fear preserves them)
# thin files are re-decision friendly: read fully, verdict quickly, change safely

Thinness is opinion liquidity (small files change minds cheaply; large files fossilize them). Solo founders re-decide constantly (new evidence weekly) — architecture must price re-decision low. Thick files protect past opinions with present fear; thin files invite better judgment anytime. Stay liquid by staying thin.

That's the shape of it — opinions owned, never rented.

Borrowed vs owned, year one

BoringPush opinions ship with their reasons attached.

No use pretending otherwise: minimalism demands decision work upfront (every default verdict takes thought) that opinionated kits skip (decide by inheritance, fast). Priced against deletion weeks plus fossilized opinions: thinking early beats unwiring later, and documented decisions outlive both. Own opinions deliberately; borrow nothing unconsciously.

FAQ

Is saasrock alternative just anti-opinion contrarianism?

No — pro-calibrated-opinions: team opinions serve teams (alignment value real), solo opinions serve solos (velocity value real). The argument is fit, not fashion: borrowed opinions cost more than authored ones at solo scale. Respect strong defaults generally; audit them specifically. Opinions are tools — match tool to hands.

Should solos really document decisions (sounds corporate)?

Five lines, not committees: what, why (incident number beats theory), revisit trigger. Corporate docs serve compliance; decision records serve future-you (and agents) at 2am. The audience is one forgetful founder plus tireless models — both need written reasoning. Minimal docs for maximal memory: corporate form, solo function.

How do opinionated kits treat agents differently?

Structure helps agents navigate (maps to read) while opinions constrain them opaquely (rules unexplained). Explicit solo law (a conventions file with reasons) beats inherited opinions (patterns without whys) for agent work: models follow stated rules reliably and inferred conventions poorly. Agents need law with legislative history, not just statutes.

What if my opinions are wrong (solo echo chamber)?

Decision records include revisit triggers (team-of-3, explicit pain, quarterly review) — wrong opinions expire by design, not by accident. Plus outside input (founder peers, incident lessons, this blog's checklists) audits the file. Echo chambers need sealed rooms; records with triggers and peers keep windows open. Wrong temporarily beats borrowed permanently.

Can two founders share solo-minimal, or need team shape?

Two is still solo-shaped (shared everything, no roles matrix needed) — pair programming's oldest lesson: two humans, one context. Past ~3 with distinct duties: reassess per the bridge-quarter pattern. Pairs thrive on minimal law; trios start needing explicit coordination. Headcount three, not two, is the inflection — again.

When are opinionated kits genuinely the answer?

Teams needing alignment day one (3+ engineers, shared context required), enterprises with standards (compliance-shaped opinions, audit-friendly defaults), and learners wanting guided tours (opinions as curriculum). All legitimate, all non-solo. Buy opinions when the organization exists to hold them; author your own until then.

Keep reading

• OpenCode SaaS Starter: Stop the Every-Session Redesign — the file holding solo law
• AGENTS.md Is a Wish List Until You Add Teeth — enforcement for owned opinions
• Take Money 3 Ways, Write Checkout Once — one decision record, three rails