Execution
Think→Act→Observe: no workflows to author
How execution works
ViksaAI does not use pre-defined workflows or DAGs. When you give a goal (in chat, via webhook, or on a schedule), the Agent Executor runs a Think→Act→Observe loop until the goal is done, aborted, or paused for your input.
The executor decides at runtime which agents to call and in what order. It can ask you questions or request approval for sensitive actions; when you respond, execution resumes from where it left off.
Think → Act → Observe
Think
Analyze current state (goal, prior task results, mappings) and decide next action: run a task, ask the user, complete, or abort.
Act
Execute the chosen action (e.g. call an agent endpoint) or pause for user input/approval.
Observe
Capture the result (task output, user reply, or error) and feed it back into the next Think step.
Loop
Repeat until goal is completed, aborted, or paused for user input.
Key concepts
| Concept | Description |
|---|---|
| Iteration | One cycle of Think → (optional Act) → Observe |
| Agent Executor | Runs the loop: reasons over state and chooses next action (execute_task, ask_user, complete, abort) |
| Mappings | Context (e.g. mapping_id → payload) passed into the executor for resolving inputs |
| Resume state | Snapshot used to continue execution after user input or approval |
Entry points
The same execution model runs whether you start from chat, a webhook trigger, or a schedule:
Chat
Natural language in the project chat
Triggers
Webhook URL with optional prompt/payload
Schedules
Cron or recurring runs