.png&w=256&q=75)
🤓 Latest Submissions
oakplay
By simply describing a messy support scenario aloud, leads can leverage Oakplay to automatically propose connected steps and decision branches. Unlike tools that invent policies, Oakplay explicitly leaves unstated outcomes as visible open routes and commits nothing until you actively accept the draft. Once a playbook is confirmed, teammates can run a case along a specific route to determine the next documented action and generate a specific handoff. If an undefined branch is reached, the run honestly stops to prevent unauthorized policy invention. The final confirmed maps export as SVG files for external documentation. Oakplay ensures your support processes remain accurate, human reviewed, and seamlessly integrated into your team operational rhythm.
30 Sep 2026
Dcnstrct
Dcnstrct helps developers understand what happens behind an action in an unfamiliar application. A simple action such as cancelling an order can cross business rules, database updates, notifications and worker logic. Understanding that behavior often requires switching between the interface, source files and logs, then manually connecting the pieces. Source code explains possible behavior, but developers also need evidence of what actually happened. Dcnstrct connects an action to its observed execution journey and an evidence-linked explanation. Our first version uses a synthetic order application with two contrasting scenarios: cancellation before fulfillment is accepted, while cancellation after shipment is refused. Users enter an interactive workspace, choose a scenario and perform the cancellation. The application executes real backend operations and records an ordered sequence of events. Users can select each step to inspect observed data, relevant numbered source excerpts and a recorded IBM Bob explanation. They can then run the second scenario, compare outcomes and reset their own runs. The accepted path shows the order becoming cancelled, a synthetic notification being saved and a fulfillment worker reading the current state before skipping shipment. The refused path preserves the shipped state and produces no cancellation notification. Notifications are database records, not real emails, and the demo worker executes synchronously. Our target users are developers onboarding to a codebase, maintainers investigating application behavior and technical teammates explaining a workflow. Dcnstrct’s creative distinction is its action-first approach: users explore a concrete outcome rather than begin with a generic code conversation. It brings the action, execution steps, source and explanation into one workspace. Observation remains separate from interpretation, and uncertainty is disclosed. IBM Bob’s recorded explanations are displayed only when the scenario and source revision match, with evidence references validated against the current run. This makes the explanations traceable while allowing visitors to explore the demo without a Bob account. I detailed my experience building on IBM Bob here: https://x.com/_HEISEMMA/status/2104221834824532348?s=20
27 Sep 2026
