VisionStudio
An LLM-powered web application for authoring, evaluating, and executing product specifications — an integrated workspace that carries the VisionSpec methodology from first draft through LLM-as-a-Judge review to tracked, shipped execution.
Key Capabilities
- Dual methodology selection: choose a requirements methodology (AWS Working Backwards, and others) and an implementation methodology (AIDLC, SpecKit) per project.
- LLM-as-a-Judge evaluation: evaluate specs against profile rubrics, with a context-aware writing assistant to help close the gaps it finds.
- AIDLC workflow: AWS's AI-Driven Development Lifecycle — Inception → Construction → Operations — with generated deliverables and phase gates to track completion.
- Initiative & portfolio execution: track initiatives, phases, and stable roadmap items (RMIs) across programs and repositories, with per-phase completion — surfaced from PRISM.
- Strategic planning: V2MOM cascades from organization to team to project, a visual capability stack, and a timeline roadmap view.
- Maturity Model dashboards: framework-based maturity assessments, including SCALE.
- Performance & token economics: tokens, cost, and cache usage broken down by model, category, and month, with month-over-month trends and shipped-work velocity — alongside DevX signals from OmniDevX.
Product Tour
Spec Definition + LLM-as-a-Judge
Each initiative is defined as a chain of specs — PRD, TRD, PLAN, ROADMAP in the PBHQ-Lite workflow — and every stage is judged in place, carrying a categorical score out of five. The weakest link (a ROADMAP at 3/5 here) is obvious at a glance, and the spec’s rendered markdown sits right below it for revision.
Execution Tracking
From spec to shipped: the Execution tab tracks the initiative’s roadmap items (RMIs) grouped into phases, each RMI with a completion status and its dependencies. Per-phase progress bars show how far Foundation — and everything built on top of it — has actually gotten.
Programs & Roadmap
Step up a level: the initiative you just saw lives in a program — here PRISM Control — beside its siblings, each with its own roadmap items (RMIs). Program-level counts summarize how much is delivered, ready, proposed, or cancelled, so an owner can steer one program at a time.
Initiative Portfolio
All the way out: every program’s initiatives roll up into a single portfolio, each card carrying a stable initiative ID, a lifecycle status — delivery complete, executing, proposed — and a live completion bar. A summary ring tallies the whole book of work: completed, in progress, proposed, and cancelled.
Performance & Token Economics
The build has a bill. The Performance dashboard surfaces total tokens, cache reads, and cost, then breaks them down by model and by category — input, output, cache — with spend-over-time trends. Cache reads dominating the token mix is the tell that context reuse is doing its job.
Monthly Trends & Accomplishments
Zoom out to the month. A running history compares cost and token volume period over period, ties spend back to the models that drove it, and pairs it with what actually shipped — an accomplishments count and weekly velocity — so cost is always read against output, never in isolation.
Why It Matters
As AI compresses the distance between intent and implementation, the bottleneck moves upstream — to how well a spec is written and judged before an agent ever touches code. VisionStudio is the applied tool for the VisionSpec methodology described on this site: it turns strategy, specifications, and delivery workflows into an operating system a ProductBuilder can actually run.