Constructor Detail

• Autonomy

  Autonomy(AgentPlatform agentPlatform, Ranker ranker, AutonomyProperties properties)

Method Detail

• getAgentPlatform

   final AgentPlatform getAgentPlatform()

• getProperties

   final AutonomyProperties getProperties()

• chooseAndAccomplishGoal

  @JvmOverloads() final AgentProcessExecution chooseAndAccomplishGoal(ProcessOptions processOptions, GoalChoiceApprover goalChoiceApprover, AgentScope agentScope, Map<String, Object> bindings, GoalSelectionOptions goalSelectionOptions, BindingsFormatter bindingsFormatter)

  Choose a goal based on the user input and try to achieve it. Open execution model. May bring in actions and conditions from multiple agents to help achieve the goal. Doesn't need reified types because we don't know the type yet.

  Parameters:

  processOptions - process options

  goalChoiceApprover - goal choice approver allowing goal choice to be rejected

  agentScope - scope to look for the agent

  bindings - any bindings to pass to the agent process.

  goalSelectionOptions - options for goal selection, such as confidence cut-off and multi-goal selection, if we want customization.

  bindingsFormatter - formatter to extract intent string from bindings for goal ranking.

• chooseAndAccomplishGoal

  @JvmOverloads() final AgentProcessExecution chooseAndAccomplishGoal(ProcessOptions processOptions, GoalChoiceApprover goalChoiceApprover, AgentScope agentScope, Map<String, Object> bindings, GoalSelectionOptions goalSelectionOptions)

  Choose a goal based on the user input and try to achieve it. Open execution model. May bring in actions and conditions from multiple agents to help achieve the goal. Doesn't need reified types because we don't know the type yet.

  Parameters:

  processOptions - process options

  goalChoiceApprover - goal choice approver allowing goal choice to be rejected

  agentScope - scope to look for the agent

  bindings - any bindings to pass to the agent process.

  goalSelectionOptions - options for goal selection, such as confidence cut-off and multi-goal selection, if we want customization.

• chooseAndAccomplishGoal

  @JvmOverloads() final AgentProcessExecution chooseAndAccomplishGoal(ProcessOptions processOptions, GoalChoiceApprover goalChoiceApprover, AgentScope agentScope, Map<String, Object> bindings)

  Choose a goal based on the user input and try to achieve it. Open execution model. May bring in actions and conditions from multiple agents to help achieve the goal. Doesn't need reified types because we don't know the type yet.

  Parameters:

  processOptions - process options

  goalChoiceApprover - goal choice approver allowing goal choice to be rejected

  agentScope - scope to look for the agent

  bindings - any bindings to pass to the agent process.

• chooseAndAccomplishGoal

  @JvmOverloads() final AgentProcessExecution chooseAndAccomplishGoal(GoalChoiceApprover goalChoiceApprover, AgentScope agentScope, Map<String, Object> bindings)

  Choose a goal based on the user input and try to achieve it. Open execution model. May bring in actions and conditions from multiple agents to help achieve the goal. Doesn't need reified types because we don't know the type yet.

  Parameters:

  goalChoiceApprover - goal choice approver allowing goal choice to be rejected

  agentScope - scope to look for the agent

  bindings - any bindings to pass to the agent process.

• chooseAndRunAgent

   final AgentProcessExecution chooseAndRunAgent(String intent, ProcessOptions processOptions)

  Choose an agent based on the user input and run it. Closed execution model: Will never mix actions and goals from different agents. Doesn't need reified types because we don't know the type yet.

• runAgent

   final AgentProcessExecution runAgent(Object inputObject, ProcessOptions processOptions, Agent agent)

• createGoalSeeker

   final Autonomy.GoalSeeker createGoalSeeker(String intent, GoalChoiceApprover goalChoiceApprover, AgentScope agentScope, GoalSelectionOptions goalSelectionOptions)

  Indicate which goal we'd use for this intent, and what agent we'd create. Dry run capability available externally.

• createGoalAgent

   final Agent createGoalAgent(Object inputObject, AgentScope agentScope, Goal goal, Boolean prune)

  Open mode. Create an agent to accomplish this goal from the given user input

  Parameters:

  inputObject - any input object

  agentScope - scope to look for the agent

  goal - the goal to accomplish

  prune - whether to prune the agent to only relevant actions