optimize() API

GEPA, Pareto, artifact, and optimizer APIs.

Most Used

`optimize`

Convenience optimizer helper that composes AxBootstrapFewShot before AxGEPA and returns an artifact without applying final component changes.

Field	Value
Kind	`function`
Ax concept	`optimize`
API section	Optimizers
Form	`optimize(program, examples, options)`
Returns	AxResult

Important Options

student/client
teacher/reflection client
metric budget
bootstrap

Examples

Rust

let artifact = optimize(&mut qa, train, json!({"maxMetricCalls": 100}))?;

Full reference

`AxBootstrapFewShot`

Few-shot demonstration optimizer that selects successful evaluator rollouts before prompt/component evolution.

Field	Value
Kind	`type`
Ax concept	`AxBootstrapFewShot`
API section	Optimizers
Form	`AxBootstrapFewShot::new(options)`
Returns	optimizer engine

Important Options

quality threshold
max demos
max rounds
batch size

Examples

Rust

let bootstrap = AxBootstrapFewShot::new(json!({"qualityThreshold": 0.7}));

Full reference

`AxGEPA`

GEPA engine with Core-owned reflection, Pareto and bootstrap behavior. Artifacts report original-program baselineScore/baselineScores at the top level and in metadata, independently of the final frontier.

Field	Value
Kind	`type`
Ax concept	`AxGEPA`
API section	Optimizers
Form	`AxGEPA::new().with_reflection_client(reflection)`
Returns	optimizer engine

Important Options

reflection client
budget
metric
candidate count
baselineScore
baselineScores

Examples

Rust

let engine = AxGEPA::new(reflection_client, json!({}));

Full reference

`OptimizerEngine`

Optimizer boundary consumed by AxGen, AxAgent, and AxFlow optimization helpers.

Field	Value
Kind	`interface`
Ax concept	`OptimizerEngine`
API section	Optimizers
Form	`OptimizerEngine::optimize(request, evaluator)`
Returns	optimized artifact

Important Options

request
evaluator

Full reference

`OptimizerEvaluator`

Evaluator callback boundary used by generated optimizers.

Field	Value
Kind	`interface`
Ax concept	`OptimizerEvaluator`
API section	Optimizers
Form	`OptimizerEvaluator::evaluate(request)`
Returns	score/evidence result

Important Options

dataset rows
candidate map
evidence

Full reference

Advanced / internals / full reference

`playbook`

Bind an ACE-backed playbook to a program; agents also expose an agent-bound playbook handle.

Field	Value
Kind	`function`
Ax concept	`playbook`
API section	Optimizers
Form	`playbook(program, student, teacher, options)`
Returns	AxPlaybook

Important Options

student/client
teacher
seed snapshot
online updates
verification budget

Examples

Rust

let pb = playbook(program, student, None::<Rc<RefCell<OpenAICompatibleClient>>>, json!({}));

Full reference

`AxPlaybook`

Persistent playbook with render/update/snapshot operations and agent-bound verified evolve over train/validation task sets.

Field	Value
Kind	`type`
Ax concept	`AxPlaybook`
API section	Optimizers
Form	`AxPlaybook / agent.playbook(...)`
Returns	playbook handle

Important Options

verify
minHeldInGain
epsilon
runsPerTask
maxMetricCalls
maxProposals

Full reference

`optimize_with_client`

Run BootstrapFewShot and GEPA against a live student client. Reports the original baseline before bootstrap. The client-free optimize helper requires pre-scored tasks.

Field	Value
Kind	`function`
Ax concept	`optimize_with_client`
API section	Optimizers
Form	`optimize_with_client(&mut program, Rc<RefCell<Client>>, examples, options)`
Returns	AxResult

Important Options

Rc<RefCell<Client>>
examples
validationExamples
bootstrap
maxMetricCalls

Examples

Rust

let artifact = optimize_with_client(&mut qa, student.clone(), train, json!({"maxMetricCalls": 100}))?;

Full reference

Full Reference

Open the full Rust reference. This page highlights the optimize symbols most relevant to everyday use.