[+] · compare · paelladoc vs the field

two layers. one question.

Some tools write the code. PaellaDoc is the layer above them: it runs those agents, gates their output against your acceptance criteria, and turns the work into versioned product. Whatever writes the code, the question is the same: when an agent says done, who decided that?

There are two kinds of tool in AI coding. Some write the code: Claude Code, Codex, Cursor, Aider, Cline. PaellaDoc is the layer above, so most of them aren't alternatives, they're what PaellaDoc runs. The real head-to-heads are the other systems on the same layer. PaellaDoc decides done by running the code against criteria you set first. A green build is not a correct feature, and that's the line every row below comes back to.

Same layer, side by side.

  PaellaDoc Devin Conductor Antigravity Superset
Runs onYour machineCloud sandboxYour MacCloud + IDEYour Mac
ModelAny agentCognition's stackAny agentGemini onlyAny agent
Decides “done”Execution gateIts own testsYou reviewIts own browserYou review
Parallel agents in worktrees✓—✓✓✓
Versioned .paella artifacts✓××××
Open pack SDK (method/stack/design)✓××××
Reverse intake of an existing repo✓××××
Telegram remote control✓××××
Multi-repo control room✓××××
No-coder mode✓××××

Model-agnostic and parallel agents aren't unique: Conductor and Superset do that too. The rows that only PaellaDoc fills are the bet: a gate that decides done by execution, and a product layer on top of the code.

The real head-to-heads.

Systems on the same layer as PaellaDoc. These are the honest comparisons.

autonomous · cloud

PaellaDoc vs Devin

Devin grades its own tests in the cloud. PaellaDoc runs locally, model-agnostic, gates on execution, and serves the no-coder Devin can't.

read → parallel · review

PaellaDoc vs Conductor

Conductor runs parallel agents and you review the diffs. PaellaDoc verifies instead of reviewing, and adds a product layer.

read → gemini · ide

PaellaDoc vs Antigravity

Antigravity is Google's agent-first IDE, locked to Gemini and built around the cloud. PaellaDoc is local, model-agnostic, gated.

read → parallel · review

PaellaDoc vs Superset

Superset runs many parallel agents and reviews diffs at scale. PaellaDoc verifies instead of reviewing, with a product layer on top.

read →

The agents PaellaDoc runs.

You don't choose between these and PaellaDoc. You run them inside it: PaellaDoc orchestrates them per task and gates whatever they produce.

runs inside

PaellaDoc vs Claude Code

The agent alone, versus the agent inside a gate and a product layer.

read → runs inside

PaellaDoc vs Codex

Same agent, but PaellaDoc decides done by running the code against your criteria.

read → runs inside

PaellaDoc vs Cursor

Cursor is where you write code. PaellaDoc is the layer that runs agents and verifies their output.

read → runs inside

PaellaDoc vs Aider

The open-source CLI pair programmer, run and verified inside PaellaDoc.

read → runs inside

PaellaDoc vs Cline

The autonomous VS Code agent, with an execution gate and product artifacts around it.

read →

What PaellaDoc adds on top of any of them.

  • An execution gate: done is decided by running the code against your criteria, not by a passing build or a glance at the diff.
  • A product methodology via an open SDK: method, stack, design, and validator packs, with first-class, versioned, comparable .paella artifacts.
  • Reverse intake of an existing repo, Telegram control, and a control room for the hundred repos on your machine.
  • A no-coder mode that builds a whole product for someone who can't read a diff.

Frequently asked.

Does PaellaDoc replace Claude Code or Codex?
No. It runs them. You keep your subscription, PaellaDoc orchestrates and verifies what they produce.
Is PaellaDoc an alternative to Cursor?
Not exactly. Cursor is where you write and edit code. PaellaDoc is the layer that runs agents, verifies their output, and manages product across all your repos.
What's the difference vs Conductor or Devin?
Read the full comparisons: vs Conductor, vs Devin.