| name | business-ai-onboarding |
| description | Onboard non-technical business teams to Codex with a controlled rollout, approval model, trust boundaries, role-based workflows, and operational examples. Use when Codex needs to help a company introduce AI to accounting, operations, admin, wholesale-market, or other back-office teams; design first-week rollout plans; define what AI may draft vs what humans must approve; convert recurring work into safe prompts and operating procedures; or produce practical onboarding materials for managers and staff. |
Business AI Onboarding
Use this skill to turn vague "train the team on AI" requests into a concrete rollout that non-technical staff can actually follow.
Workflow
- Identify the team, the work queue, and the approval bottlenecks.
- Define the trust boundary before proposing any AI workflow.
- Start with low-risk, high-frequency tasks that already have a human review step.
- Assign role-specific use cases instead of one generic AI policy for everyone.
- Build a first-week plan with named owners, deliverables, and review points.
Start Here
Collect these facts first:
- Team names and who approves their work
- Systems touched by each team
- Inputs they receive and outputs they produce
- Tasks repeated at least three times per week
- Tasks that already require manager review, dual control, or sign-off
- Data classes involved: public, internal, confidential, regulated, financial, customer
If the user does not provide this structure, derive it by interviewing them in business terms rather than technical terms.
Trust Boundary
Apply this default policy unless the user gives a stricter one:
- Allow Codex to draft, summarize, reformat, classify, and prepare checklists.
- Require human approval before any message is sent externally, any number is posted to a ledger, any order is committed, any payment is released, or any master data is changed.
- Block fully autonomous use for payroll, banking, tax filing, vendor setup, credit decisions, customer promises, and inventory adjustments unless the user explicitly says the company has approved controls for it.
- Prefer read-only and draft-first workflows during onboarding.
Read references/approval-trust-setup.md whenever the user needs approval matrices, escalation rules, or examples of what AI may and may not do.
How To Choose Pilot Work
Pick work that satisfies most of these conditions:
- Repetitive and text-heavy
- Painful but not legally irreversible
- Easy to compare against an existing human output
- Easy to review in under five minutes
- Useful across several employees, not just one expert
Reject pilot candidates that:
- Require direct system mutation without a review gate
- Depend on undocumented tribal knowledge
- Mix several departments' authority in one step
- Hide errors until month-end or quarter-end
First-Week Execution
Use references/first-week-rollout.md for the full day-by-day plan.
In short:
- Day 1: map roles, approvals, systems, and forbidden actions
- Day 2: choose 3 to 5 pilot tasks and define expected outputs
- Day 3: write and test role-based prompts on real examples
- Day 4: run supervised live work and measure review effort
- Day 5: decide what becomes standard practice, what stays restricted, and what needs more controls
Role-Based Design
Do not onboard "the company" all at once. Build by role:
- Accounting: close support, invoice matching, variance explanations, collections drafts
- Operations: intake triage, shipment exception handling, SOP drafting, issue handoff notes
- Admin: scheduling support, vendor packet prep, meeting summaries, document cleanup
- Wholesale-market teams: buyer outreach drafts, assortment comparison, promo planning prep, stall or order summary normalization
Read references/role-based-use-cases.md when the user asks for department-specific workflows or sample prompts.
Output Format
When executing this skill, produce these sections unless the user asks for something narrower:
- Team Scope
- Approval and Trust Rules
- First-Week Rollout Plan
- Role-Based Use Cases
- Risks and What Must Stay Human
- Immediate Next Actions
Keep outputs operational. Name owners, review points, and sample artifacts. Avoid generic "AI transformation" language.
Example Requests
- "Use $business-ai-onboarding to roll out Codex for a three-person accounting team without letting AI post entries."
- "Use $business-ai-onboarding to onboard office admins and operations coordinators in the first week."
- "Use $business-ai-onboarding to design approval rules and safe use cases for a wholesale produce distributor."
References
references/first-week-rollout.md for the rollout sequence, meeting structure, and adoption metricsreferences/approval-trust-setup.md for authority boundaries, human checkpoints, and red-line actionsreferences/role-based-use-cases.md for accounting, operations, admin, and wholesale-market examples
