Skip to main content
The Complex Task Automation Agent allows you to design conversational flows that handle processes with multiple possibilities. Its main strengths are the ability to integrate tools tailored to your business needs and the flexibility to define instructions directly in the prompt, allowing you to adjust the agent’s behavior without adding unnecessary complexity.

​
Is it for you?

  • ✅ Ideal if you need
  • ❌ Not ideal if
  • You have a complex process you want the AI to manage.
  • There are multiple possible outcomes within a single conversation.
  • You need to define different exit or escalation rules.
  • You want to advance part of a sales process.
  • You want to reduce response time to prospects.

​
Use cases

  • Advanced lead qualification with multiple exit routes.
  • Automated delivery of post-sale information.
  • Communication with attendees after registering for a webinar.
  • Helping users choose a product from a diverse set of options.
  • Automatic follow-up on abandoned carts.

​
Basic configuration

For the Complex Task Automation Agent to work as expected, the initial configuration requires detailed context and several iterations. Think about edge cases, test with real examples, and if something doesn’t behave as you expect, tweak the prompt until you get the desired result. Remember that this agent can call tools, and your instructions must be specific enough to guide their use.
  1. Context and objective
    • Explain what the agent needs to achieve and in what scenario it will be used.
  2. Tone and style
    • Indicate the desired tone (for example, professional and clear).
  3. Available knowledge
    • Specify when to consult knowledge sources and when not to.
  4. Available tools
    • Include the name, purpose, parameters, preconditions, and limits of each tool.
    • Define usage rules: when to call it, what to validate beforehand, and what to return after.
  5. Task description with explicit rules
    • Steps to follow, priorities, completion criteria, and success metrics.
    • What to do if information is missing and how to ask for it.
  6. Capabilities and limitations
    • What the agent can and cannot do (and must not claim to do).
  7. Concrete examples
    • Include examples of correct inputs and outputs, plus one or two counterexamples.
  8. Reasoning
    • A reminder to think step-by-step before answering.
  9. Escalation and exits
    • When to finish, when to escalate, and what data to log.
Note on initial complexity
The basic configuration may still be complex. It requires enough context and testing cycles for the agent to behave as expected. Start with a limited scope, add tools incrementally, and expand coverage as results become consistent. 💡Once you have a prompt that works correctly, head over to the Conversation Builder and add the Automate Complex Tasks block to your flow. TOM Node EN.jpg

​
Tools

The Complex Task Automation Agent can use different tools to perform actions inside or outside the conversation. These tools expand the agent’s capabilities, allowing it to retrieve information, create tickets, or update properties in external systems like HubSpot, in addition to interacting with Treble’s native functions.

​
General

  • Search the Web — Allows the agent to perform an online search to get updated or contextual information.

​
HubSpot

  • Transfer to an agent — Sends the conversation to a human agent within HubSpot.
  • Create ticket — Creates a new HubSpot ticket with the collected information.
  • Read Contact Property — Retrieves the value of a specific contact property.
  • Edit Contact Property — Updates a contact property with new data provided by the user or the flow.

​
Treble

  • Save response — Stores user information or responses as variables within the flow.
  • Transfer to an agent — Escalates the conversation to a human agent in Treble.
  • Send conversation — Sends the entire conversation to another configured flow or agent.
  • Close conversation — Ends the interaction and marks the conversation as resolved.

​
Frequently Asked Questions

It’s an AI agent designed to handle conversational flows with multiple possibilities, integrating tools and advanced logic to automatically execute tasks.
Unlike simpler qualification or response agents, this one can use custom tools, apply conditional rules, and escalate conversations according to your needs.
No. Configuration is done directly in Treble’s interface, though it requires a well-written prompt and iterative testing to achieve the expected behavior.
Yes. The agent can understand voice messages sent by users through WhatsApp and will automatically respond with text in the conversation.
Review the prompt and make it more explicit. You can add more context, outline the steps it should follow, or define the edge cases it should handle.
Yes. The agent can be integrated into a larger flow and can trigger or close conversations, save responses, or call external tools according to your rules.
The agent can use general, HubSpot, or Treble tools, including actions such as creating tickets, reading or editing properties, transferring conversations, or searching the web.
Use the playground to run controlled tests with different inputs. Iterate until the agent responds correctly in all scenarios.
Define clear exit or escalation rules. If the agent can’t resolve a situation, transfer the conversation to a human agent or log the event in a ticket.
It doesn’t learn automatically from each interaction, but you can improve its behavior by iterating on the prompt and adjusting tools or exit conditions.

​
Important considerations

Privacy: Inform users that their data will be stored and ensure compliance with applicable regulations (GDPR, etc.).