Product · code · evidence · one local system
The local software factory.
Start with an idea or an existing repository. PAELLADOC keeps discovery, product decisions, delivery, AI work, code and proof connected as one living system—on your Mac.
Free for personal use · no account · Apple Silicon
- 01Discover
- 02Define
- 03Plan
- 04Build
- 05Verify
- 06Learn
One context. Every decision stays attached.
Not another coding tool
Software is more than the code that happens to exist today.
It is the problem, the evidence, the choices, the specification, the work, the code, the tests and what you learn after shipping. Most tools own one slice. PAELLADOC keeps the entire chain alive.
Three spaces · one product memory
Enter where you are. Move without losing context.
You do not have to choose a permanent persona. Begin without code, operate as a product manager, or open the full engineering system. The same product graph follows you.
Build without code
Describe the product in plain language. PAELLADOC structures the brief, plans the work, runs the builders and shows what is complete, blocked or verified.
Idea → blueprint → working repositoryRun the product lifecycle
Frame problems, gather evidence, design experiments, track metrics, refine stories, plan sprints and turn learning into the next decision.
Discovery → decision → deliveryConnect product to code
Read an existing repo into product artifacts, link stories and acceptance criteria to code KGs, orchestrate agents and require evidence before work closes.
Repo → knowledge graph → validated changeThe factory
One loop from uncertainty to evidence.
Each stage produces context for the next one. Nothing important is trapped in a prompt or abandoned in a planning document.
-
01
Decision
Discover
Turn an uncertainty into hypotheses, evidence, an experiment and a decision.
-
02
Contract
Define
Keep PRDs, epics, stories and acceptance criteria as a traceable product contract.
-
03
Plan
Plan
Refine scope, estimate work, manage dependencies and run visible sprints.
-
04
Change
Build
Route work to your chosen AI engines in isolated worktrees against the real repository.
-
05
Proof
Verify
Use gates, tests, screenshots, traces and video. No evidence means no green light.
-
06
Memory
Learn
Feed outcomes, code history and decisions back into product memory for the next loop.
The connective tissue
A product graph above a code graph.
PAELLADOC does not flatten your product into tickets. It links intent to implementation: discovery evidence to decisions, decisions to artifacts, artifacts to stories and criteria, criteria to code, and code to validation.
Open any capability later and recover what it does, why it exists, where it lives and what proves it works.
- LOCAL
- SQLite, repositories and evidence stay on your machine
- BORN-LINKED
- Product artifacts keep provenance to code
- MODEL-OPEN
- Use Claude, Codex, Gemini or local engines
The execution layer
The agents work. The product contract decides what done means.
Every story can run with its own developer, model, branch and isolated context. Fleet shows initial scope, recovery work, verification and bottlenecks without hiding them inside one opaque queue.
Real application · real projects · local orchestration
- 01Bring the AI subscriptions and local models you already use.
- 02Keep concurrent work isolated with git worktrees.
- 03Recover failed runs without losing the product thread.
- 04Close acceptance criteria only when observable evidence exists.
Open by design
Local does not mean closed.
The factory runs on your computer, but its boundaries are intentionally open where portability matters.
Your models
Route work across providers or local engines. The product memory survives the switch.
Your code
Work in ordinary repositories and branches. Keep using the tools around them.
Your data
Chats, artifacts, decisions and evidence live locally in inspectable storage.
Your extensions
Add methods, decision packs, skills and UI surfaces through the plugin SDK.
Core source-available · plugin SDK Apache-2.0 · explicit local permissions
The field notes
The factory is built in public, not marketed in theory.
Product experiments, architecture decisions, benchmarks and uncomfortable lessons stay available. The archive remains part of the proof.
I asked the Codex app to build an app twice. Neither made it to tomorrow.
./read →Product knowledge graph: read your product's shape, not the repo's feature list
./read →Product management in the AI era: building got cheap, deciding got expensive
./read →You do not have a Claude Code problem. You are the runtime.
./read →Plain answers
What PAELLADOC is—and is not.
- What is a local software factory?
- A local system that connects product discovery, specifications, planning, AI execution, code knowledge and validation instead of treating them as separate tools and conversations.
- Do I need to know how to code?
- No. The Vibecoder space gives you a plain-language cockpit. Product and Engineering spaces reveal more control when you need it.
- Does PAELLADOC replace Claude, Codex or Gemini?
- No. Those are engines. PAELLADOC supplies the durable product context, repository, orchestration, gates and evidence around them.
- Is PAELLADOC cloud software?
- No. PAELLADOC is a local-first macOS application. It can call the AI providers you choose, while product data, repositories and evidence remain on your machine.
- Is everything open source?
- The core is source-available and the plugin SDK is Apache-2.0. Your data stays portable, and extensions use explicit local permissions.
Start the factory
Your next product loop can live on your machine.
Start from an idea. Point it at a repo. Or use the product workspace before another line of code is written.
Alpha v0.2 · signed, notarized, stapled and obfuscated · Apple Silicon
Current core features for personal individual use stay free. Read the promise →