What would you like EDS to deliver?
Describe the problem and how you will know it is solved. The agent classifies the request; category rules and an ETA less than 30 days away determine priority. Assign the delivery owner in ADO.
Local preview reads actual ADO tasks. Creating requests or starting agents still requires an authenticated API identity.
Request workflows
Request delivery workflows, highest priority first. Click a request's row to expand its delivery progress and ADO hierarchy. Delivery views include a Portal-only completion milestone after post-validation & rollout. Highlighted work items need owner review; open their ADO links to approve or leave comments.
Loading request workflows...
| Priority | Request | Workflow tracker | Delivery stage | Status / blocker |
|---|
No request workflows yet. Submit a request to begin qualification.
Delivery insights
Demo / mock dataIllustrative data only. These placeholder metrics are a synthetic six-week example, not measurements from ADO, agent runs or production. They do not describe the real workflows in the Request workflows tab.
A preview of the trajectories we want to track: PR creation, phase duration, engineering time saved, bugs identified, and human- or agent-triggered correction cycles. Real telemetry is not connected yet.
PR creation trajectory
PRs created per mock week. Creation does not imply merge or successful delivery.
Bugs identified by phase
Mock findings grouped by the phase where each bug was first identified, across all six weeks.
Engineering time saved
(Human-effort baseline - agent active engineering time) / human-effort baseline × 100%
The denominator uses human effort from the EDS task classification table, weighted by the mock number of completed tasks in each category. Agent time includes active intake, design, code/test and all correction or redesign work. Both sides exclude rollout time and waiting for review. This compares execution time, not measured reductions in human labor.
| Mock week | Human-effort baseline | Agent active engineering | Engineering time saved |
|---|
Human-effort baselines and mock task mix
Source: EDS_Agentic_Delivery_Executive_Table.md, Human effort & basis column. These are the table's estimates/plans, not measured agent savings. Only explicit hourly baselines are included; month-based, range and TBD entries are not silently converted.
| Task category | Source human effort per task | Mock tasks | Weighted baseline |
|---|
Correction & backtrack trajectory
One initiated revision/backtrack cycle counts once, whether triggered by human feedback or an agent-detected error. Repeated comments or retries within the same cycle are not additional cycles. Same-phase human revisions are included; routine code-fix retries are not cross-stage backtracks.
| From phase -> return to phase | Human-directed | Agent-detected | Total cycles |
|---|
- Human feedback: Design -> Request intake. The reviewer identifies missing acceptance criteria; clarify the scope before revising the design.
- Agent-detected error: Implementing & pre-validation -> Design. Implementation reveals an architectural constraint that cannot be resolved with a bounded code fix. Return to design, revise the proposal, obtain fresh approval, then reimplement and requalify.
Demo visualization only: these examples do not claim that automatic cross-stage recovery is implemented or that any live workflow has backtracked.
Time spent in each phase
Average elapsed hours, including review waits, for each mock week's completed-delivery cohort. The same cohort is used for all four phase averages; their sum is its average delivery time. These are not engineering effort or hours saved.
| Mock week | Request intake & qualification | Solution design & planning | Implementing & pre-validation | Post-validation & rollout | Avg delivery time |
|---|