All case studies

SlideX case study

Working product integration

How SlideX turns an idea into slides with a conversational agent

SlideX helps people create polished presentations with less friction. Heddle supplies the agent-runtime building blocks that let a user explain what they want, refine it over multiple turns, and have the agent create or update the real presentation inside the editor.

SlideX editor showing a presentation beside a multi-turn conversation with the SlideX Agent
A real SlideX conversation asks the agent to restyle a slide, add SDK content, and insert a section header.
Product
Browser presentation workspace
Experience
Conversational slide creation and editing
Heddle role
Conversations, tools, hosted runs, and persistence

The challenge was bigger than adding a chat box

SlideX already had templates, a browser editor, MotionDoc presentation tools, and PowerPoint export. The new agent needed to work with those product capabilities instead of becoming a separate demo.

A useful experience also had to remember earlier requests, expose understandable progress, recover completed conversations, reject stale results, and coexist safely with manual edits to the deck.

From a request to a real deck

The agent works inside the same editor where the user reviews and finishes the presentation.

  • Describe the presentation or change in everyday language.
  • Refine the result with follow-up requests while the conversation retains context.
  • Use MotionDoc tools to create, inspect, validate, and update the actual deck.
  • Return to completed conversations after a refresh or server replacement.
  • Keep manual edits and undo behavior under SlideX's control.
SlideX editor showing a list of saved agent conversations for the presentation
Completed conversations remain selectable instead of disappearing when the visual panel closes.

Heddle fits inside the product architecture

Heddle does not replace SlideX's application. It owns the reusable runtime mechanics inside a SlideX-controlled agent service.

  1. 01Product experience

    SlideX editor

    Conversation UI, presentation state, undo, manual edits, and result review.

  2. 02Runtime + product policy

    Heddle-powered agent service

    Multi-turn execution, run lifecycle, activity events, cancellation, reconnection, and SlideX orchestration.

  3. 03Canonical tools + data

    MotionDoc + Supabase

    Presentation mutations and validation, product records, safe transcripts, sessions, and completed archives.

A clear responsibility boundary

Heddle handles reusable agent behavior. SlideX remains the authority for its users, presentation data, product policy, and experience.

Heddle supplies
  • Multi-turn conversation, continuation, and compaction semantics
  • Model, tool, and MCP execution with structured activity and results
  • Addressable runs, ordered events, bounded replay, and cancellation
  • Browser-safe remote protocol and reconnection behavior
  • Revision-safe session and archive persistence contracts
SlideX owns
  • Authentication, tenancy, and presentation authorization
  • The canonical MotionDoc and conflict-safe presentation writes
  • Product chat history, retention policy, and Supabase adapters
  • Editor UI, undo behavior, and stale-result review
  • Model credential handling, deployment, and operational policy

Heddle absorbs the repeated runtime work

The integration still required real product engineering. Heddle reduced the amount of correctness-sensitive infrastructure SlideX had to design and maintain itself.

SlideX did not have to reinvent

  • Conversation revisions, continuation, compaction, and recovery behavior
  • Run identity, event ordering, replay cursors, cancellation, and terminal outcomes
  • Model and MotionDoc MCP tool orchestration
  • Remote envelope validation and browser reconnection policy
  • Generic repository conflicts, archive append, and failure semantics

The SlideX team stayed focused on

  • What a presentation agent should understand and change
  • How users sign in, choose a conversation, and review work
  • How the canonical deck is saved without overwriting newer edits
  • How the agent fits the existing editor, export, and deployment model

Verified in a working product flow

The case study is based on the real editor integration, not a standalone prototype or a mocked marketing screen.

  • Create a deck, then make constrained follow-up edits while preserving the rest.
  • Keep multiple conversations and return to an earlier transcript.
  • Restore completed conversations and model-facing context after refresh and process replacement.
  • Cancel a run without persisting its proposed change.
  • Protect a newer manual edit when an agent result becomes stale.
  • Report credential, validation, and conflict failures without claiming success.

Build an agent that belongs inside your product

Start with a working conversation, then adopt only the tools, hosting, remote-client, and persistence layers your experience needs.