> ## Documentation Index > Fetch the complete documentation index at: https://docs.obversa.ai/llms.txt > Use this file to discover all available pages before exploring further. # Software factory, as a workflow > Search calls it a software factory. What ships is a workflow: named roles, a review that sends work back, and a person at the merge, on the engines you already use. Search calls it a software factory. What ships is a workflow: named roles, a review that sends work back, a person at the merge. Obversa is that runtime, on the engines you already use. The factory picture is a belt that turns issues into merged pull requests without a person. The teams that make it work do not run a belt. They run a process: one agent writes, another reviews, the work goes back when the review fails, tests run as commands rather than as agent turns, and a person decides at the end. That process is what you write here, as one file. ## The file The file is `examples/teams/feature-delivery.ts`. ```ts theme={null} import { claude } from '@obversa/engine-claude-cli'; import { codex } from '@obversa/engine-codex'; import { run } from '@obversa/runtime'; import { fromFile, person, stage, workflow } from '@obversa/teams'; /** * A feature, delivered the way a team delivers one. The roles are named once; * every stage is a small block of nouns: who does it, what it writes, who * reads it, where a red result goes back to. Inference happens only where a * role is named; every other stage is a command or a person. */ const team = workflow('feature-delivery', { brief: fromFile('briefs/triple.md'), options: { timeout: '10m' }, roles: { analyse: claude('claude-sonnet-4-5'), implement: codex('gpt-5.6-luna'), 'research-review': [codex('gpt-5.6-luna')], 'code-review': [claude('claude-sonnet-4-5')], approve: person('Ship this change?'), }, stages: [ stage('research-context', { agent: 'analyse', writes: 'team-output/research-context.md', desc: 'Read the workspace and write down what the change touches.', gate: 'The context note is in the workspace and a reviewer has accepted it.', reviewedBy: 'research-review', retry: 3, }), stage('research-requirements', { agent: 'analyse', writes: 'team-output/research-requirements.md', desc: 'Turn the brief and the context note into requirements, one REQ-n per line.', gate: 'The requirements note is in the workspace and a reviewer has accepted it.', reviewedBy: 'research-review', retry: 3, }), stage('plan', { agent: 'analyse', writes: 'team-output/plan.md', desc: 'Write an executable plan from the requirements, one check per REQ-n.', gate: 'Every requirement has a check in the plan.', reviewedBy: 'research-review', retry: 3, }), stage('tests-first', { agent: 'implement', writes: 'test/triple.test.mjs', desc: 'Write the declared test files from the accepted plan before any implementation exists.', gate: 'Every declared test file exists and covers the plan.', reviewedBy: 'code-review', retry: 3, }), stage('implement', { agent: 'implement', writes: 'src/triple.mjs', desc: 'Write the code to the plan and the tests.', gate: 'The source file exists.', retry: 3, }), stage('test', { run: ['node', '--test', 'test/triple.test.mjs'], desc: 'Run the tests; a red run goes back to implement with the output.', gate: 'The test command exits 0.', sendsBackTo: 'implement', }), stage('review', { panel: 'code-review', agree: 1, desc: 'Read the change and the test result against the plan.', gate: 'At least one reviewer has accepted the change.', sendsBackTo: 'implement', }), stage('approve', { input: 'approve', desc: 'Put the verified change in front of a person.', gate: 'A person has said yes.', }), stage('close', { agent: 'analyse', writes: ['team-output/evidence.md', 'team-output/learning.md'], desc: 'Write the evidence of the run and what was learned, from the record alone.', gate: 'Both notes are in the workspace.', }), ], }); const result = await run(team); console.log(JSON.stringify(result.outcome, null, 2)); ``` Every stage names its worker and the files it may write. The review role is a seat from another provider. A failed review sends the work back to the stage that owns it, with the findings, up to the retry you set. The last stage is a person. ## What a factory search usually wants | the search | what you write | | ------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Issues in, pull requests out. | A workflow whose first stage reads the issue and whose last stage is a person at the merge. | | Many agents on one change. | Named roles, one engine each, from more than one provider. | | Quality without a human reading every diff. | A review role that sends work back with findings, and commands for the tests. | | Runs that do not lose work. | A record of every step as it happens. A plain run does not resume; the [supervised runner](/driving/runner) restarts a killed run from its own record, and today it drives a compiled graph rather than a file like this one. | | A place to watch it. | Every run can [serve its own page](/driving/monitor). | ## Things that catch people out * **A product factory is the same search with a bigger scope.** The answer is the same: a workflow with the reviews and the person written in. * **An agent pipeline without a return is a queue.** The step that sends work back is what makes the review count. See [the review loop](/reviewing/review-loop). * **The tests are commands.** An agent that runs and supervises a test suite burns tokens to do what a command does for free. Write `run` stages for them. ## Where to go [Domain-specific harnesses](/use-cases/domain-specific-harnesses) for the same file read as one job's harness; [the review loop](/reviewing/review-loop) for how a return works.