How They Do ItBETA
TREN

COMPANY PLAYBOOK

Supabase

Supabase combines office-free asynchronous work with location choice, product speed with small design gains, and community scale with open contribution paths.
Operating system
  1. Async context
  2. local workspace choice
  3. annual in-person connection
  4. small kaizen release
  5. design consolidation
  6. open contribution and paid-contributor path.

System principles

  • Separate where and when work happens from the outcome.
  • Make design a post-speed consolidation system rather than a delivery gate.
  • Connect community contribution to an open path for progression.

Tensions and trade-offs

  • Remote flexibility enables global participation but requires intentional context and social connection.
  • Kaizen increases speed while accepting temporary design inconsistency and rework.

PRACTICES

How Supabase builds its operating system

01
Remote

Asynchronous flexible work

The team works fully remotely and asynchronously, and people choose where they work.

How it works
Notion, Slack and Discord carry context and connection across time zones.
Why it is used
Enables global talent and personal time flexibility.
Trade-off
Coordination gaps emerge when goals and response expectations are unclear.
Best fit
Global teams with strong self-management and written communication.
Open original source
02
Remote

Global coworking membership instead of an office

Instead of a permanent headquarters, a global coworking membership gives people access to local workspaces.

How it works
A reliable workspace outside the home becomes part of individual location choice.
Why it is used
Supports workspace access in an office-free model.
Trade-off
Local access, quality and usage are uneven.
Best fit
Office-free teams working across cities and countries.
Open original source
03
Culture

Annual one-week company offsite

The distributed team meets in person for one week each year.

How it works
People who work asynchronously all year build shared physical time and social connection.
Why it is used
Strengthens trust and cross-team weak ties.
Trade-off
Creates travel, accessibility, visa and coverage costs.
Best fit
Teams able to fund and inclusively plan an annual global gathering.
Open original source
04
Product

Kaizen first, design consolidation later

Product teams ship small improvements without waiting for design approval; Design consolidates afterward.

How it works
A quick solution ships, then the design team moves the production result toward a more coherent whole.
Why it is used
Removes design as a delivery bottleneck.
Trade-off
Creates temporary inconsistency and rework.
Best fit
Reversible interface decisions with strong product observation.
Open original source
05
Product

Small design system from proven patterns

Only components that have actually been used several times enter the design library.

How it works
Abstraction happens after a repeated pattern is proven in production.
Why it is used
Limits component and variant bloat.
Trade-off
Standardization arrives later and early versions can diverge.
Best fit
Fast product teams able to observe production patterns.
Open original source
06
Participation

Progress from open to paid contribution

The community contributes code, documentation and support; strong contributors can earn stipends and direct team access.

How it works
A visible path is published from volunteer contribution to paid contribution and possible employment.
Why it is used
Provides a path to recognize and sustain community work.
Trade-off
Selection fairness and the boundary between volunteer and paid labor need explicit governance.
Best fit
Companies with an open-source product and active contributor community.
Open original source
07
CI/CD

Restrict infrastructure changes to a controlled pipeline

Corrective actions announced after the 12 February 2026 incident: AWS SCPs to block account- or region-wide changes outside the controlled pipeline, plus an IaC blocklist for risky resources.

How it works
After an internal monitoring-service deployment changed region-wide VPC Block Public Access, two preventive controls were defined for privileged network changes.
Why it is used
Reduces the risk that one service can interrupt customer traffic across a region.
Trade-off
Stale rules and exceptions can also block legitimate infrastructure changes.
Best fit
Multi-region AWS infrastructure carrying customer traffic.
Open original source
08
CI/CD

Detect regional network failure before production and from outside

The postmortem commits to extending pre-production coverage to every production region and continually probing external connectivity; completion is not claimed.

How it works
Regional parity and external connectivity checks were designed together to catch traffic loss missed by internal health signals.
Why it is used
Makes regional configuration drift and customer-visible unavailability visible earlier.
Trade-off
Regional test environments and continuous external probes add cost and can cause false alarms.
Best fit
Services with multiple production regions and external access dependencies.
Open original source

Primary sources

Supabase CareersCompany-published careers page
Source details
Company publication: YesAccessed: 2026-09-05Visible update: Live page reviewed 2026-09-06Licence: No reuse licence displayed; treat page prose as copyrighted/all rights reserved.Archive: Public HTML URL; eligible for an external web snapshot (no snapshot is claimed).
Why we'll stay remoteCompany-published remote-work article
Source details
Company publication: YesAccessed: 2026-09-05Visible update: 2023-08-05Licence: No reuse licence displayed; treat page prose as copyrighted/all rights reserved.Archive: Public HTML URL; eligible for an external web snapshot (no snapshot is claimed).
Contributing to Supabase docsCompany-owned GitHub documentation guide
Source details
Company publication: YesAccessed: 2026-09-05Visible update: 2026-09-04 (visible page history)Licence: Apache-2.0 repository; verify prose reuse scope before republishing.Archive: Public GitHub page; commit permalinks, raw files, and repository archives are capturable (no snapshot is claimed).