AI opportunity assessment

From as-is processes and platform strategy, propose vendor-neutral use cases. Score value, feasibility, time-to-pilot, and risk. Confirm before writing DataPool rows.

Document information

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

Kind

Sequential procedure.

Depends on

ai-adoption-data-model, ai-discovery.

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`
`as_is_processes_schema`	As-is processes schema	`ai-adoption-data-model`	Import first	`as_is_processes`	Tenant `SKILL.md`

Import contract

Field	Value
`suggested_skill_id`	`ai-opportunity-assessment`
`version`	`1.0.0`
`capabilities.tools`	`get_platform`, `describe_datapool_table`, `query_datapool`, `insert_datapool_rows`, `update_datapool_rows`, `request_human_input`

Procedure

Stages: load_context → propose → score → confirm_write.

Stage	Quality gate to exit
load_context	`as_is_processes` queried; strategy recap already known from discovery
propose	Each use case has `type` from the allowed list and a capability written without Tealfabric component names
score	Four 1–5 scores: value, feasibility, time-to-pilot, risk; `data_readiness`; `effort_band` S/M/L
confirm_write	`request_human_input` confirm before `insert_datapool_rows`

Scoring (1–5, vendor-agnostic):

Value: time saved, error cost, or headcount avoided as volume grows — aligned with strategy value proposition
Feasibility: data accessible, system of record known, skills available, human review acceptable — independent of Tealfabric
Time-to-pilot: can a timeboxed experiment start in weeks on any stack?
Risk: PII, regulated process, irreversible automation

Now (set later by ai-roadmap) = high value + high feasibility + low/medium risk. Say “do not automate yet” when data or ownership is missing.

implementation_options is optional free text (AWS, Azure, Tealfabric, custom). Do not rank Tealfabric first.

DoD: at least five scored ai_use_cases (or advisor-confirmed smaller set); none named as ProcessFlow / Autonomic Agent / WebApp implementations.

Tenant SKILL.md (copy this body)

# AI opportunity assessment

Operate use cases in DataPool schema `{{use_cases_schema}}`. Read as-is rows from `{{as_is_processes_schema}}`. Load `datapool` and `platforms` if you need to re-cite strategy.

## Sequence

1. `describe_datapool_table` then `query_datapool` as-is processes for the current `engagement_id`.
2. Propose use cases. `type` must be one of: `document_ai`, `conversational_agent`, `workflow_automation`, `decision_support`, `data_integration`, `rpa`.
3. Write `capability` as a business capability (e.g. “classify incoming invoices”), never as a Tealfabric resource name.
4. Score 1–5: `value_score`, `feasibility_score`, `time_to_pilot_score`, `risk_score`. Set `data_readiness` and `effort_band`.
5. `request_human_input` confirm, then `insert_datapool_rows` / `update_datapool_rows`. Set `status=scored`.
6. Set engagement `status=scoring`.

## Stop

- Do not invent systems that discovery did not record.
- Do not write `implementation_links`.
- Do not create ProcessFlow, WebApps, or Autonomic Agents.
- Do not default `implementation_options` to Tealfabric.

Optional skill.json

{
  "name": "ai-opportunity-assessment",
  "version": "1.0.0",
  "description": "Propose and score vendor-neutral AI/data/automation use cases on DataPool.",
  "capabilities": {
    "prompt": [
      "Capability types only. Confirm before insert_datapool_rows.",
      "Do not create platform implementation resources."
    ],
    "resources": [],
    "tools": [
      "get_platform",
      "describe_datapool_table",
      "query_datapool",
      "insert_datapool_rows",
      "update_datapool_rows",
      "request_human_input"
    ]
  }
}