Partner Demo — Story Line and Scripts
This section is for partners and presenters running the IBM DevOps Solution Workbench demo.
Why this demo
The demo focusses on the value of Workbench as the design layer for AI coding assistants. Demo participants should immediately recognise pain points from their own daily work and experience how Workbench mitigates them.
The demo increases the problem scope step by step, telling one story across three blocks:
- Developer / Team pain — Architecture doesn't fit into a text prompt. A developer explains a new requirement to their coding assistant in plain language; the AI generates code that compiles and looks reasonable — but it doesn't know about a dependency that lives elsewhere in the system, or a decision an architect made and documented in Confluence, an email, or a meeting the developer wasn't in. The code works today. The pain shows up later: at runtime, when a transaction only rolls back the payment adapters it knows about, or in the next review, when the implementation doesn't match what was actually decided.
- Enterprise pain — What works for one developer doesn't scale on its own. Every developer would need to know every rule — compliance requirements, security policies, coding conventions — for every service, every time. In practice, that knowledge lives with a few experts, gets copied from old repos, or gets rediscovered by trial and error. Deviations slip through, and at enterprise scale, a single deviation from a convention is a risk.
- Organizational pain — Business intent doesn't reach the teams building it. A business analyst thinks in user stories; an architect thinks in components and boundaries. They don't share a system or a language, so requirements get lost or reinterpreted on the way down, arriving at the developer as a vague ticket with no visibility into dependencies, risks, or decisions already made elsewhere. This bottom-up approach creates an early "aha" moment for the user persona in the demo, and wraps up with the need-to-buy for the buyer persona.
The story shows how Workbench solves real-world challenges that show up when AI coding assistants lack the necessary context — architecture, service design, ADRs, tech debt — in the right place and accessible:
- Not possible to prompt the relevant architecture as text to a coding assistant
- Architecture mostly not accessible to AI
- Enterprise-wide guidelines not available to AI
- Decisions documented somewhere else, not at a central point, not accessible in the context of the respective service
In this section
Key Messages & Features
What to land with the audience — main features to showcase and the key takeaways.
Scenario — RoboFlow
The reference application the demo is built around.
Demo Blocks
What a block is, how the three blocks combine, and the scripts for each.
Preparation and Clean-up
Environment links, AI assistant settings, and per-block preparation steps.