AEGIS · Module 6 · Brief
Artifact 6.3
12-Month AI Roadmap
Now / Next / Later roadmap across five categories. Refreshed at every QGR. The roadmap the executive team commits to and the board pack references.
- Client
- [CLIENT NAME]
- Engagement
- [ENGAGEMENT ID]
- Version
- v1.0
- Issued
- 2026-05-18
Delivered by TechFides under the AEGIS Governance Operating Services engagement. This document is proprietary to the client named above. Redistribution beyond the engagement steering committee requires written consent.
Purpose
The 12-Month AI Roadmap is the forward plan that binds every other AEGIS artifact together. It is the place where value hypotheses (Artifact 3.3), governance investments (Module 1 + 2), and workflow design (Module 4) land on a timeline the executive team can commit to. It is reviewed every quarter at QGR and re-baselined every 12 months. Items move between horizons; they do not accumulate forever.
Horizon Definitions
Intent — Three horizons. Fixed meanings. No ‘soon’ or ‘Q-ish’ — the horizon itself is the commitment level.
Months 1–3
Commitment: Active. Resourced. Named owner. In the current quarter plan.
Exit: Shipped, retired, or demoted to Later with rationale at QGR.
Months 4–6
Commitment: Planned. Budget sketched. Owner tentatively named. Prerequisites identified.
Exit: Promoted to Now at QGR, or demoted to Later with rationale.
Months 7–12+
Commitment: Directional. Hypothesis-level. Review conditions named.
Exit: Promoted to Next at QGR, or dropped with a dated decision.
Roadmap Item Schema
Intent — Every item on the roadmap uses this schema — no exceptions. Incomplete items do not enter the roadmap until the fields are filled.
| Field | Value | Guidance |
|---|---|---|
| Item ID | [RM-000] | Stable identifier for the life of the item. |
| Title | [Title] | Outcome-phrased, not activity-phrased. |
| Category | [Governance / Enablement / Workflow / Platform / Compliance] | One category only. Forces sharper framing. |
| Horizon | [Now / Next / Later] | Set and defended at QGR. |
| Owner | [Name] | Single name. Accountable, not advisory. |
| Sponsor | [Name] | Executive who clears blockers and reports at QGR. |
| Outcome | [Observable change] | What is true after this item is done. Measurable where possible. |
| Value Hypothesis | [Value type + estimate + confidence] | From Value & Spend framework (3.3). |
| Prerequisites | [Upstream items / artifacts] | Explicit chains. No hidden dependencies. |
| Risk Exposure | [High/Med/Low + top risk] | From Risk Register (2.3). Very High requires Exec Sponsor approval to enter Now. |
| Decision Gates | [Go/no-go checkpoints] | Pre-declared, not improvised. |
| Stop Condition | [What would make us kill this] | Every Now item has a stop condition. This is non-negotiable. |
Illustrative Roadmap
Intent — Representative sample across the five categories and three horizons. Replace with client-specific items during the Diagnostic synthesis.
Ratify AI Use Policy + AUP v1.0
Owner: AI Governance Lead · Sponsor: General Counsel
Outcome: Policy ratified by Exec, AUP acknowledged by 100% of staff, training currency baselined.
Value: Risk avoidance (audit-ready posture).
Ship WF-03 (support triage) to T2 autonomy
Owner: VP Support · Sponsor: COO
Outcome: Cycle time on triage down 40% vs baseline. ≥80% of in-scope agents run 3+/week.
Value: Productivity + CX — projected $420k / 12 mo.
Launch Builder training cohort + certification
Owner: AI Governance Lead + L&D · Sponsor: CHRO
Outcome: 25 certified Builders across Engineering + Data + Product. Merge protection enforced on prompt library.
Value: Risk avoidance + throughput capacity.
Expand AI Inventory to include embedded features in SaaS tools
Owner: AI Governance Lead · Sponsor: CISO
Outcome: Inventory captures embedded-AI vendor features; vendor risk reassessed for top 10.
Value: Risk avoidance (shadow AI surface area).
Evaluate internal model hosting for CLIENT-RESTRICTED workloads
Owner: VP Engineering · Sponsor: CISO
Outcome: Decision: host vs. continue approved vendor. Costed options on table.
Value: Data sovereignty + unit economics.
Scale WF-03 pattern to Finance AR/AP (WF-0X)
Owner: VP Finance · Sponsor: CFO
Outcome: New workflow specced + approved; pilot running in 6 of 10 target processes.
Value: Error reduction + throughput — projected $260k / 12 mo.
Publish ISO 42001 readiness gap analysis
Owner: AI Governance Lead · Sponsor: General Counsel
Outcome: Gap analysis complete. Decision on formal certification path at QGR.
Value: Customer-facing posture + contract velocity.
Assistant pattern for customer-facing comms (T1 only)
Owner: CMO · Sponsor: CEO
Outcome: Design + red-team suite drafted. Pilot conditional on WF-03 sustainment.
Value: Throughput + CX.
Governance automation — inventory + risk register sync
Owner: VP Engineering · Sponsor: AI Governance Lead
Outcome: Prototype reduces manual inventory upkeep by >50%.
Value: Program efficiency + audit readiness.
Re-Planning Cadence
Intent — The roadmap is a living document. These are the only times it changes.
| Event | What Changes | Output |
|---|---|---|
| QGR (quarterly) | Horizon promotions/demotions. New items admitted. Stop-conditions exercised. | Roadmap v+1 with diff log |
| Annual re-baseline | Full rebuild. Every item re-justified against current strategy. 12-month window reset. | Roadmap v2.0 delivered to Board (6.2) |
| Material incident | Immediate pause of any Now item tied to the incident. Review at next QGR. | Incident RCA + roadmap impact memo |
| Strategy change | Exec Sponsor calls emergency QGR. Items re-scoped, de-scoped, or added. | Roadmap interim version + minute |
| Regulatory shift inside 90 days | Compliance-category items jump horizons as needed. Other work may be displaced. | AI Governance Lead memo + roadmap update |
Anti-Patterns
Intent — Roadmaps fail for a small, recognizable set of reasons. Catch them early.
Now shelf keeps growing
Why it happens: Items added but none shipped or demoted. Now becomes Later in disguise.
Counter: Enforce a Now cap (e.g., 5 items per category). No admits without a demotion or a ship.
Later becomes a dumping ground
Why it happens: Ideas land in Later and are never reviewed.
Counter: Every QGR, review Later. Promote, sharpen, or drop. Nothing stays untouched for 2 quarters.
No stop conditions
Why it happens: Items ship late, pivot often, and no one can tell when to stop.
Counter: Stop conditions are a schema requirement (Section 3). Items without one do not enter Now.
Governance and workflow items trade air time
Why it happens: When the program is pushed, governance slips because it looks optional.
Counter: Governance has its own category and minimum allocation per horizon. Protected at QGR.
Value hypotheses go unchecked
Why it happens: Projected value never reconciles with realized value.
Counter: At each QGR, reconcile projected vs realized against the Value Tracker (3.3). Adjust confidence.
Roadmap does not match the board pack
Why it happens: Two forks of truth confuse the board and the staff.
Counter: The board pack (6.2) pulls directly from this artifact. Any divergence is a pack error and is corrected.
Regulatory & Framework Mapping
Intent — A roadmap is the vehicle by which the frameworks below expect demonstrable program evolution over time.