AI roadmap

Prioritize 2–5 now items, write a sequenced roadmap, and produce executive deliverables (markdown, visuals). Does not create Tealfabric implementation resources.

Document information

Field	Value
Canonical URL	`/docs/07_ai-agents-and-mcp/model_skills/60_ai_adoption/ai-roadmap`
Version (published date)	`2026-08-27`
Tags	`ai`, `skills`, `playbooks`, `roadmap`

Kind

Sequential procedure.

Depends on

ai-adoption-data-model, ai-opportunity-assessment.

Prerequisites

id	question	search	if_missing_or_ambiguous	default_to_propose	binds_to
`use_cases_schema`	Use cases schema	`ai-adoption-data-model`	Import first	`ai_use_cases`	Tenant `SKILL.md`
`roadmap_items_schema`	Roadmap items schema	`ai-adoption-data-model`	Import first	`roadmap_items`	Tenant `SKILL.md`
`engagement_folder`	Deliverable folder	`ai-adoption-data-model`	Import first	`engagements/current/`	Tenant `SKILL.md`

Import contract

Field	Value
`suggested_skill_id`	`ai-roadmap`
`version`	`1.0.0`
`capabilities.tools`	`describe_datapool_table`, `query_datapool`, `insert_datapool_rows`, `update_datapool_rows`, `request_human_input`, `write_file`

Load chat-visuals in the Trace session for Mermaid (2×2 and timeline). Those diagrams are working-session visuals. The executive pack is the JWT SPA (see user guide) plus {{engagement_folder}}/roadmap.md.

Procedure

Stages: rank → sequence → write → deliverable.

Stage	Quality gate to exit
rank	Scored use cases loaded; advisor confirms 2–5 `now` candidates
sequence	Each `now` item has `experiment` (metric, timebox, skills) and `effort_band`
write	`roadmap_items` inserted; matching use cases `status=ready_to_pilot`
deliverable	`roadmap.md` written; Mermaid in the Trace reply

Horizons: now (2–6 weeks), next, later. Default output is small steps, not an 18-month program.

DoD: 2–5 horizon=now rows; markdown presentable to executives without naming Tealfabric components as the destination.

Tenant SKILL.md (copy this body)

# AI roadmap

Operate `{{roadmap_items_schema}}` and `{{use_cases_schema}}`. Write `{{engagement_folder}}/roadmap.md`.

## Sequence

1. Query scored use cases for the engagement.
2. Propose `now` / `next` / `later`. Prefer high value + high feasibility + low/medium risk for `now`.
3. `request_human_input` confirm the set (especially the 2–5 `now` items).
4. Insert `roadmap_items` with `sequence`, `horizon`, `experiment`, `depends_on`.
5. Update selected use cases to `status=ready_to_pilot`.
6. Write `{{engagement_folder}}/roadmap.md`: strategy recap (cite, do not copy wholesale), 2×2 narrative, sequenced table, experiment cards, `implementation_options` as options not a decision.
7. In the Trace reply, load `chat-visuals` and draw a small Mermaid 2×2 and a now/next/later timeline.
8. Set engagement `status=roadmap`. Tell the advisor the JWT engagement app (`/ai-adoption`) shows the same DataPool rows after login.

## Stop

- Do not create ProcessFlow, WebApps, or Autonomic Agents (that is `ai-tealfabric-bridge`).
- Do not write `implementation_links`.
- Do not put more than five items in `now` without confirm.
- Do not treat in-chat Mermaid as the only executive deliverable.

Optional skill.json

{
  "name": "ai-roadmap",
  "version": "1.0.0",
  "description": "Prioritize now/next/later roadmap items and write executive markdown.",
  "capabilities": {
    "prompt": [
      "2–5 now items. Vendor-neutral experiments. Do not create platform implementation resources."
    ],
    "resources": [],
    "tools": [
      "describe_datapool_table",
      "query_datapool",
      "insert_datapool_rows",
      "update_datapool_rows",
      "request_human_input",
      "write_file"
    ]
  }
}