Skip to content

roadmap.md

Latest commit

 

History

History
91 lines (60 loc) · 2.9 KB

roadmap.md

File metadata and controls

91 lines (60 loc) · 2.9 KB

Roadmap

This roadmap is intentionally high-level. It describes the direction of pea without locking the project into premature implementation detail.

Guiding Direction

pea exists to simulate stakeholders for the Epic Systems Engineering Judgment Workshop.

The product direction is shaped by a few consistent priorities:

  • make ambiguity feel realistic
  • reward strong requirement discovery instead of fast implementation
  • support critique-based learning loops
  • keep pea focused on the agent/service side of the system
  • leave the learner-facing workshop experience to the separate workshop app

Phase 1: Live Cohort Support

Near-term, pea should be useful in facilitated live cohorts.

That means focusing on:

  • reliable stakeholder simulations for instructor-led exercises
  • a small set of strong scenario archetypes
  • instructor visibility and control over agent behavior
  • straightforward integration with the workshop app

At this stage, realism matters more than breadth.

Phase 2: Reusable Scenario Library

Once live exercises are working well, the next step is to make scenarios more repeatable and reusable.

Priorities include:

  • scenario definitions that can be reused across cohorts
  • agent configurations that capture different stakeholder perspectives
  • prompts and controls that make exercises easier to run consistently
  • a clearer library of workshop-ready exercises

Phase 3: Self-Paced Readiness

As scenarios become more stable, pea should support self-paced learning experiences used by the workshop app.

This phase emphasizes:

  • simulations that hold up without a facilitator in the loop
  • stronger consistency in how ambiguity and constraints are revealed
  • better support for critique after the conversation phase
  • scenario quality that remains useful outside live delivery

Phase 4: Studio-Grade Judgment Practice

Longer term, pea should support more advanced critique and judgment training.

This includes:

  • richer stakeholder dynamics and conflicting incentives
  • more nuanced risk and rollout conversations
  • scenarios that expose silent regressions and hidden tradeoffs
  • improved instructor tooling for refining and evolving simulations

Non-Goals For Now

This roadmap does not assume pea becomes the full learning platform.

It is not meant to absorb:

  • cohort management
  • curriculum delivery
  • learner progress UX
  • the full workshop application experience

Those belong to the workshop app and facilitation layer, not the stakeholder simulation service itself.

How We Should Evaluate Progress

Progress should be judged by whether pea makes workshop exercises better at teaching engineering judgment.

Signals that matter:

  • participants have to ask sharper questions to succeed
  • scenarios consistently surface hidden constraints and tradeoffs
  • critique discussions become more concrete and rigorous
  • instructors can meaningfully control and improve simulations over time