Crafting a Story-Driven Use Case

Overview

  • Project Name: Unifying Custom Providers in Agent Control Tower (ACT) — GraphQL Use Case Guide
  • Role: Technical Writer and API Documentation Architect
  • Objective: I designed and authored an end-to-end technical use case guide demonstrating how developers can use Boomi’s GraphQL APIs to programmatically register and govern third-party or in-house AI agents that lack native integration.
  • Key Outcome: Transformed a complex set of GraphQL mutations and queries into an approachable, narrative-driven blueprint that helps users understand when and why to use the API via a realistic enterprise scenario.

Problem

  • The Technical Complexity: Boomi’s Agent Control Tower (ACT) provides centralized governance, but companies often run fragmented AI stacks—such as Amazon Bedrock, internal data science models, or niche custom tools that ACT doesn’t natively recognize.
  • The API Onboarding Wall: GraphQL APIs (consisting of dozens of granular mutations and queries) can feel intimidating and abstract to developers if presented merely as an uncontextualized reference list.
  • The Documentation Gap: Users needed to understand how the API pieces fit together sequentially (Authentication > Provider Creation > Agent Registration > Governance Attachment) rather than just reading parameter definitions.

Target audience

  • Primary: Enterprise developers, integration engineers, and platform architects building automation scripts or CI/CD pipelines to manage AI fleets.
  • Secondary: Security and compliance administrators looking to centralize disparate AI models under a single governance control plane.

Solution and technical process

To make the documentation engaging and actionable, I engineered a relatable, end-to-end narrative scenario backed by a structured architectural path:

  • The Made-Up Enterprise Scenario: I introduced a fictional global enterprise running multi-stack AI (Amazon Bedrock, an in-house fraud-detection LLM, and a third-party pilot) to immediately ground the abstract technical problem in a real-world business pain point (lack of visibility and auditability).
  • Sequential “Choose Your Path” Information Architecture: I broke down a complex web of API interactions into three logical, distinct operational phases:
    1. Path 1: Registering custom provider accounts (containers for un-integrated stacks).
    2. Path 2: Registering and versioning agents under those containers.
    3. Path 3: Layering on governance objects (guardrails, tasks, tools).
  • Practical Developer Considerations: I proactively documented crucial technical guardrails, such as handling 5-minute JWT expirations, managing API rate limits (HTTP 429), and leveraging CI/CD pipelines to treat the AI catalog like code.

Tools and technologies used

  • APIs and Protocols: Boomi GraphQL API, REST (for JWT generation), Basic Authentication, Bearer Tokens
  • Platform and Ecosystem: Boomi Agent Control Tower (ACT), Custom Type Provider Accounts
  • Documentation Formats: Markdown, interactive API Explorer, Altair GraphQL Client scripts

Results and impact

  • Bridged the Theory-Practice Gap: Turned a raw list of 28 mutations and 19 queries into an intuitive, story-driven workflow that developers could easily follow.
  • Improved Developer Experience (DevEx): Empowered teams to successfully onboard un-supported AI providers into a unified governance catalog without waiting for native platform updates.
  • Encouraged Automation: Provided clear guidelines on embedding GraphQL mutations into CI/CD pipelines for automated AI fleet management.

Live Artifact: Unify Custom Providers in Agent Control Tower