Skip to main content

Custom Agent Overview

Custom Agents and Out-of-the-box Agents in ai12z are link between your copilot and external systems, enabling seamless two-way communication for dynamic, interactive workflows. Whether you need to query Shopify in real-time via GraphQL, integrate with specialized ticketing solutions like Tessitura, or connect to custom internal microservices, Custom Agents transform your copilot into a true extension of your digital operations.

Unlike competing solutions that rely solely on retrieval-augmented generation (RAG), ai12z harnesses these agents for action-oriented tasks, empowering your copilot to not only answer questions but also perform essential real-world operations—such as managing reservations, processing transactions, or updating inventories.

Custom Agents unlock new possibilities, beyond what the out-of-the-box Agents provide in ai12z. By integrating any digital system—such as a CRM, a marketing automation platform, or even bespoke backend services—your copilot can orchestrate tasks and workflows that were previously out of reach. With this extended connectivity, organizations can harness the full power of their unique digital infrastructures, enabling the copilot to perform specialized operations—from automating intricate internal processes to managing highly customized user interactions.

Extensible and Built for Integration​

ai12z is designed to work the way you need it to. Our platform is a favorite among agencies and developers, who rely on its versatility to integrate with any backend services or technologies, like Shopify, Salesforce, or custom APIs. Need to connect to Shopify’s GraphQL to sync with a client’s inventory or e-commerce system? No problem. Want to pull data from a bespoke RESTful API to power personalized user experiences? We’ve got you covered.

Whether you’re a developer looking for robust tools or an agency integrating a new client site, ai12z fits seamlessly into your workflow, giving you the power to connect, customize, and deliver.

Block Diagram​

carousel flow diagram

Overview Video​

Key Benefits​

Extensibility​

Custom Agents enable you to connect with virtually any backend system supporting REST APIs or GraphQL. This flexibility allows organizations to leverage both in-house systems and third-party platforms, ensuring your copilot remains at the center of your digital ecosystem.

What truly makes a Copilot powerful is its ability to use—and even extend—Agents. Here are some key ways Agents can supercharge your AI-driven workflows:

  • Seamless Integration: Agents can call JSON REST or GraphQL APIs to communicate with external systems like ticketing or eCommerce platforms.
  • Interactive Widgets: They can be HTML components (for example, a Google Chart or a mortgage calculator) embedded directly in your application’s interface.
  • Custom Data Processing: Agents can run Python scripts on data from any source before handing it off to the ReAct LLM—perfect for advanced analytics or data transformation.
  • Specialized LLM Systems: Agents themselves can be specialized Large Language Models, each optimized for a particular domain or use case.
  • Flexible Output Paths: Although the ReAct LLM orchestrates the Agents, an Agent’s response can bypass the LLM entirely and go directly to a UI component—like sending real estate listings straight to a carousel slider without passing through the LLM.

This multi-faceted approach ensures your Copilot can deliver exactly the functionality you need, in the format you want, all while remaining flexible enough to integrate new capabilities over time.

Full Customization​

By defining custom parameters, data mappings, and specialized logic, you can tailor integrations to your precise business needs. From CRM workflows to on-demand content retrieval, you maintain full control over each integration.

Two-Way Communication​

Move beyond mere data retrieval. Custom Agents empower your copilot to take meaningful action, including booking tickets, updating systems, or executing multi-step processes—directly and in real time.

Seamless Integration​

From Shopify’s GraphQL APIs for real-time inventory updates to bespoke RESTful endpoints for specialized data, Custom Agents transform the copilot into a unified hub for all your digital services and workflows.

Scalable and Future-Proof​

Built with evolving business requirements in mind, Custom Agents effortlessly adapt to shifting data needs. They deliver richer, context-driven interactions that grow with your organization, adding new capabilities without overhauling existing infrastructure.

How Agents Work with the LLM and the data flow​

  1. LLM Parmaters

    • The ReAct LLM analyzes the user’s request, reviews all available agents and their parameter descriptions, and decides whether an agent call is needed. If it proceeds with a call, the ReAct LLM supplies the necessary parameters—for example, those used to call REST or GraphQL APIs, or parameters used by Python or the HTML template.

  2. Data Sources: Seamlessly connect to any external endpoint.

    • REST API: REST APIs are often used for client-server and microservice architectures
    • GraphQL: GraphQL is a data query and manipulation language for APIs that allows a client to specify what data it needs ("declarative data fetching").
    • None: For example when creating an HTML widget, it does not always require external data

  3. Data Sources Filtering

    • JSONata: (Optional) Filter and transform incoming data for precise control over the final output.
    • Python: (Optional) Perform advanced data manipulation or custom logic before returning results.

  4. Handle Response - Processing

    • Reasoning LLM: If further analysis or natural-language processing is required, the Reasoning LLM ensures that the data is properly interpreted and enriched.
    • Bypassing the LLM: For rapid responses, you can send data directly to clients (e.g., in a carousel list or slider, or a dynamic Form), bypassing the ReAct LLM to accelerate response times.

  5. Response Delivery

    • Carousel: Present data visually in a slider or list format—ideal for product listings or event schedules.
    • Form: Dynamically generate forms for seamless data collection.
    • Template (HTML): Leverage customizable HTML widgets (e.g., mortgage calculators, Google Charts) for a more interactive user experience.

  6. Action Capabilities

    • Two-Way Communication: Execute real-world actions, like booking reservations or updating records.
    • Workflow Orchestration: Chain multiple steps together to build complex workflows, all accessible via a single copilot interaction.

Example Use Cases​

  • Shopify Integration: Access real-time product inventories, process transactions, or display product carousels directly in your copilot interface.
  • Ticketing Systems: Query event availability, reserve seats, or update booking details in real time, all from within your copilot.
  • Custom Microservices: Bridge proprietary internal systems for data ingestion and updates, creating a unified command center for operations.

Conclusion​

Custom Agents in ai12z enable your copilot to transcend the limitations of simple Q&A. By integrating advanced filtering, transformation, and automation capabilities with any REST or GraphQL endpoint, your AI assistant becomes a powerful, adaptive partner in your digital strategy. Whether you’re seeking seamless workflows, real-time insights, or action-oriented tasks, Custom Agents provide the foundation for an intelligent, fully integrated copilot that evolves with your business.