The InsightGrid App Stack

The tools we use — and why they work together

InsightGrid runs on a deliberately small, modular set of tools designed to survive change.

This page outlines the tools currently used behind the scenes — what each one does, why it was chosen, and where sensible alternatives exist. It isn’t intended as a prescription or a “best tools” list. Think of it as a working example: a real stack shaped by practical constraints, evolving needs, and a bias toward resilience over novelty.

If a different combination suits you better, that’s entirely expected. The principles matter more than the tools.

Stack Philosophy

Before tools, there are constraints.

The InsightGrid stack is guided by a few simple rules:

  • Tools should be loosely coupled — replacing one shouldn’t break the rest
  • Data should remain human-readable and portable wherever possible
  • Local-first and cloud tools should complement, not compete
  • Automation should remove friction, not introduce complexity
  • No tool should become a single point of failure, whether cognitively or technically

This philosophy allows the stack to evolve without needing constant reinvention.

The Stack by Layer

Knowledge & Thinking Layer

Primary tool: Capacities

This is where thinking happens — ideas, notes, connections, and long-term context.

Capacities is used as an object-based knowledge system rather than a traditional notes app. Concepts, tools, projects, and ideas are treated as first-class objects instead of nested documents. This supports clearer thinking over time and reduces the tendency for knowledge to collapse into unreadable archives.

Capacities works particularly well for:

  • Long-term thinking and synthesis
  • Linking ideas across domains
  • Maintaining context over months and years

It is less suited to:

  • Highly structured task management
  • Heavy real-time collaboration
  • Pure markdown-only workflows

Alternatives you may prefer: Obsidian, Notion, Anytype

Each can support similar outcomes with different trade-offs.

Cloud Storage & File Layer

This layer is intentionally split between private cloud and self-hosted infrastructure.

The goal is not ideological purity, but resilience: sensitive data is kept under direct control where possible, while cloud services are used where they meaningfully reduce friction or risk.

Private Cloud Storage

Primary tool: Proton Drive

Proton Drive is used as a secure, privacy-first cloud storage layer for files that benefit from off-site redundancy and easy access across devices.

It fits well into the InsightGrid stack because:

  • Files are end-to-end encrypted by default
  • The company’s incentives align strongly with privacy and user trust
  • It integrates cleanly with a broader encrypted ecosystem

Proton Drive is not treated as a full collaboration suite or a knowledge system. It is simply a secure, low-friction place for files that shouldn’t live exclusively on a single machine or server.

You may prefer alternatives such as iCloud Drive, Google Drive, or Tresorit depending on ecosystem and collaboration needs.

Self-Hosted Cloud Storage

Primary tool: ownCloud

ownCloud provides a self-hosted cloud layer for files where control, structure, and long-term ownership matter more than convenience.

It is used for:

  • Personal and family document storage
  • Structured archives
  • Files that should remain accessible regardless of third-party policy or pricing changes

Self-hosting introduces responsibility — backups, updates, and security matter — but it also restores a degree of digital sovereignty that commercial cloud platforms cannot offer.

For InsightGrid, ownCloud acts as a long-term anchor: slower than commercial cloud services in some areas, but far more predictable over time.

Alternatives include Nextcloud or Syncthing-based setups, depending on preferences and technical comfort.

Automation & Glue

Primary tool: n8n

Automation sits between tools, not inside them.

n8n is used to connect systems where repetition would otherwise introduce friction, whether that be content pipelines, summaries, alerts, light enrichment, and administrative glue work.

Crucially, automation is optional. Nothing in the InsightGrid stack requires automation to function. This ensures the system remains understandable and operable even if automation is removed entirely.

Where automation exists, it is designed to be:

  • Transparent
  • Easily disabled
  • Non-destructive

Alternatives: Zapier, Make, native integrations

Each is valid depending on comfort level and scale.

Infrastructure & Hosting

InsightGrid uses a pragmatic mix of self-hosted and managed infrastructure.

The goal is not maximal control, but sensible ownership:

  • Self-hosting where portability matters
  • Managed services where reliability outweighs sovereignty
  • Clear exit paths wherever possible

Infrastructure decisions are revisited periodically, but never impulsively.

AI & Assistance Layer

AI is used as an assistant — not an authority.

It supports:

  • Summarisation
  • Pattern recognition
  • Administrative acceleration

It does not replace thinking, writing, or judgement. AI output is always downstream of human intent and subject to review. Where AI is removed, the system continues to function, just more slowly.

How the Stack Works Together

None of these tools are remarkable in isolation.

What matters is how loosely they are connected. Thinking can move to a different tool without breaking back-ups or archiving. Archiving can change without affecting automation. Automation can be removed entirely without collapsing the system.

This modularity is intentional. It reduces risk, cognitive load, and future regret.

What This Stack Is Not

This stack is not:

  • A recommendation for beginners
  • A claim of optimality
  • A rejection of simpler or more opinionated systems

It reflects one way of working — shaped by writing, research, automation, and long-term thinking — not a universal solution.

A Living Reference

This page is updated when a core tool changes or a meaningful workflow is replaced.

It exists to provide transparency, not certainty.

Last reviewed: December 2025