Tool steps
Run one connected tool directly with configured arguments. No AI is involved in a pure tool step.
Resource
Botchi automations are deterministic pipelines for repeatable operations. They can call tools, ask AI for structured work, hand off to agents, branch on conditions and record every run.
Automation
Morning support triage
The builder is a recipe, not a blank chat box. Each step has a type, a configured payload and a place in the run trace.
Run one connected tool directly with configured arguments. No AI is involved in a pure tool step.
Use one AI call where language or classification helps, optionally returning structured output for later steps.
Delegate a sub-task to an account agent with its own instructions, tools and knowledge.
Continue, skip or stop based on deterministic rules over earlier outputs, without asking a model to decide.
Run automations manually or on a wall-clock schedule with weekdays, month days or fixed intervals.
Pass run input, earlier step outputs and anchored time tokens like Today or Yesterday into later fields.
Repeat a step for each item in an earlier result, with a max-item cap so large lists fail loudly.
Every run records step status, output, error and Sparks usage so teams can inspect what happened.
An AI automation is a repeatable workflow that runs as a step-by-step pipeline. It can combine direct tool calls, AI prompts, account-agent handoffs and deterministic conditions.
An autonomous agent decides how to approach a task. An automation follows a recipe in the same order every time, which is better for predictable business operations.
Yes. Botchi automations can run manually or on a schedule such as weekdays, selected days of the month, a wall-clock time or fixed intervals.
Yes. Each run records a step-by-step trace, including status, outputs, errors and Spark usage.