12-Month Roadmap — M1 to M12

Calendar projection of the work-breakdown-structure.md. M1 ≈ September 2026 (assumption — to confirm at kickoff); month numbers, not calendar dates, are authoritative. Per the minimal-disruption mandate, M1–M3 asks of Prime Products staff are interviews only (≤2 h per person); heavier engagement starts with pilots in M6 and stays capped (see change plan § minimal disruption).

13 phases (P0–P12) across the program’s foundations, platform, knowledge/visibility, and value-delivery tracks. The published site (prime.tpl.one) renders this as a waterfall timeline against the M1–M12 axis with a live now-line — see it on the homepage or directly on this page. The month ranges below are the source of truth for that view (update both together if phase timing changes):

Phase	Track	Months
P0 Preparation & governance	Foundations	M1
P1 Discovery & requirements	Foundations	M1–M3
P2 Process map & AI opportunities	Foundations	M2–M4
P3 Infra architecture	Platform	M2–M3
P4 Hardware procure & deploy	Platform	M3–M6
P5 LLM stack implementation	Platform	M4–M7
P6 Obsidian KB design & rollout	Knowledge & Visibility	M2–M9
P7 Mission Control site	Knowledge & Visibility	M1–M12
PX Cylinders Pilot — order desk (13 weeks, brought forward)	Value Delivery	M5–M7
P8 Department pilots	Value Delivery	M6–M8
P9 Agents & automation	Value Delivery	M7–M10
P10 Training & change mgmt	Value Delivery	M4–M11
P11 Full rollout	Value Delivery	M9–M11
P12 Optimization & governance	Value Delivery	M11–M12

PX — the Cylinders Pilot, brought forward. Scoped on the 6 July 2026 call (REQ-014, interview note) and offered as TPL-PRIME-2026-002 (€51,000 fixed, 13 weeks: discovery 2 wks → build 6 wks → supervised run 4 wks → go/no-go 1 wk), the calibration-gas cylinders order-desk pilot runs ahead of and independently of the P8 department-pilot phase — its own engagement with its own go/no-go on measured time-to-close. Shown on the timeline as a dashed bar at ~M5–M7 (start is triggered by the open items from the call, not by a program month; offer currently in TPL review). Its results feed the P8 pilot-selection and the wider go/no-go evidence base.

---

M1 — Program launch & discovery start

• Main focus: governance stood up, discovery underway, Mission Control v0 live.
• Key activities: kickoff, steering formation, budget envelope (WBS 0.1–0.7); questionnaires finalized, interview calendar set, executive interviews (1.1–1.3); Mission Control blueprint + v0 build (7.1–7.2).
• Deliverables: signed charter, governance model, discovery instruments, prime.tpl.one v0 live.
• Decision gates: steering ratifies charter and budget envelope.
• Dependencies: CEO availability for kickoff and all-hands.
• Department involvement: executive management only (interviews); all staff receive one announcement — no other load.
• Infrastructure milestones: none (design starts M2).
• Knowledge base milestones: vault repo initialized (this repo already carries program docs).
• Mission Control: v0 launched — overview, roadmap, news, FAQ.
• Training: none yet; comms channel #ai-transformation opened.
• Board reporting: kickoff confirmation; program calendar circulated.

M2 — Discovery in full swing; architecture design begins

• Main focus: department interviews; infrastructure design; KB conventions.
• Key activities: interviews across commercial/operations/back-office departments (1.4–1.6); systems inventory + data assessment start (1.8–1.9); process-mapping method set (2.1); reference architecture + BoM drafting (3.1–3.3); KB blueprint and templates (6.1–6.2).
• Deliverables: first interview-note batch, sizing assumptions, draft BoM, KB blueprint.
• Decision gates: board approves hardware budget (target end M2, hard limit M3) — WBS 3.7.
• Dependencies: systems inventory (1.8) feeds sizing (3.1); IT time for tenant/SoftOne access.
• Department involvement: sales, procurement, customer service, marketing, logistics, warehouse, finance, accounting — one 90-min interview each. Interview-only load.
• Infrastructure milestones: architecture doc and BoM drafted; site survey scheduled.
• Knowledge base milestones: blueprint + template library approved.
• Mission Control: roadmap page updated with confirmed dates; first news post (“what discovery found so far” teaser).
• Training: none.
• Board reporting: hardware budget decision pack (BoM options at €[TBD] price points, TCO, lead-time risk).

M3 — Discovery closes; hardware ordered

• Main focus: finish interviews, order hardware (critical path), synthesize findings.
• Key activities: remaining interviews — IT, compliance, HR, administration, reporting/BI (1.6–1.7); KPI baseline capture (1.12); requirements registered as REQ-NNN (1.10); discovery synthesis (1.11); vendor RFQ, selection, PO issued (4.1–4.2); site prep starts (4.3); model/stack selection ADR (3.4); security design (3.5).
• Deliverables: discovery report, populated requirements register, PO with delivery ETA, serving-stack ADR.
• Decision gates: steering accepts discovery report; hardware vendor selected and PO approved (if budget gate slipped from M2, it must close here — every week of PO delay slips platform GA).
• Dependencies: 6–10-week hardware lead time starts ticking at PO date.
• Department involvement: IT (systems work), remaining departments’ interviews only. Last interview-only month.
• Infrastructure milestones: hardware ordered; interim dev sandbox stood up (4.4, no business data).
• Knowledge base milestones: vault infrastructure (git + sync + access model) working (6.3).
• Mission Control: discovery-findings summary published (sanitized); “hardware ordered” milestone post.
• Training: none; champions pre-identified informally during interviews.
• Board reporting: discovery synthesis presentation; requirements register walkthrough.

M4 — Opportunity design & pilot selection; software work starts

• Main focus: use-case portfolio scored; pilots chosen; LLM software work begins on interim environment.
• Key activities: process maps completed (2.2–2.4); use-case long-list + scoring (2.5–2.6); department one-pagers (2.8); serving engine on dev sandbox (5.1); champions program launch (10.1); curriculum development starts (10.2); seed KB content migration begins (6.4).
• Deliverables: scored use-case pipeline, top-10 shortlist, 15 department one-pagers, champion roster.
• Decision gates: pilot departments selected (2.7) — steering picks 2–3 based on pain, data readiness, champion strength.
• Dependencies: hardware in transit (lead-time weeks 2–6); scoring needs completed process maps.
• Department involvement: dept heads review their one-pagers (1 h each); 15 champions nominated (~2 h/week commitment starts); pilot departments notified.
• Infrastructure milestones: serving stack running on interim environment; hardware delivery ETA tracked weekly.
• Knowledge base milestones: first ~20 seed documents migrated (ISO 22000 procedures, key SOPs).
• Mission Control: use-case pipeline page (sanitized top-10); pilot announcement.
• Training: champions guild first meeting; curriculum outlines drafted.
• Board reporting: portfolio review — where the value is expected, pilot rationale.

M5 — Hardware lands; platform build on-prem

• Main focus: installation, burn-in, migration of stack from sandbox to on-prem; first-cohort training.
• Key activities: delivery/installation/burn-in/base platform (4.5–4.7, assumes ~8-week lead time from M3 PO); stack migration on-prem + Greek/English model evals (5.1–5.2); chat UI with SSO (5.3); RAG pipeline v1 against seed KB (5.4); champion + pilot-user training (10.3); pilot KB content sprints start (6.5).
• Deliverables: acceptance report, model-eval report, chat UI in restricted preview, trained champion cohort.
• Decision gates: hardware acceptance sign-off (4.6). Contingency: if delivery slips to the 10-week edge, burn-in compresses into early M6 and GA gate carries a risk flag (see risk register RSK-01/02).
• Dependencies: everything downstream of 4.7; RAG needs seed content from 6.4.
• Department involvement: pilot departments — champions in training (2 h labs) + content sprints (~3 h/week for champions); everyone else untouched.
• Infrastructure milestones: on-prem platform powered on and accepted; models served locally.
• Knowledge base milestones: pilot departments’ KB sections in active build; RAG indexing live.
• Mission Control: “the machine is here” post with photos; platform status widget added.
• Training: first hands-on labs (champions + ~25 pilot users).
• Board reporting: infrastructure milestone confirmation; spend vs CapEx budget.

M6 — Platform GA gate; pilots launch

• Main focus: harden, load-test, and open the platform to pilot departments.
• Key activities: integrations v1 — SharePoint read, Teams bot skeleton, SoftOne read-only (5.5); telemetry (5.6); security hardening + pen check (5.7); GA readiness review (5.8); pilot design finalized and pilots begin (8.1–8.3); KPI dashboard wiring on Mission Control (7.3).
• Deliverables: hardening report, load-test results at 50-user profile, pilot plans with success criteria.
• Decision gates: go/no-go platform GA (5.8) — steering decision on load test + security review. No-go path: 2-week remediation sprint, gate re-run early M7.
• Dependencies: GA gates pilot start; telemetry gates KPI dashboards.
• Department involvement: 2–3 pilot departments go hands-on (target ≤4 h/person/month incl. training); IT deep in ops handover prep.
• Infrastructure milestones: platform GA; monitoring/alerting live; p95 latency baseline recorded.
• Knowledge base milestones: pilot dept sections usable; retrieval-eval target met for pilot corpora (5.4).
• Mission Control: live KPI dashboards v1 (adoption + platform health); pilot kickoff stories.
• Training: pilot-user labs completed; office hours start weekly.
• Board reporting: GA decision minutes; M6 KPI targets first formal read vs kpi-framework.md.

M7 — Pilots running; agent foundations

• Main focus: pilot iteration velocity; start building agents.
• Key activities: pilots B/C fully active, weekly reviews (8.4–8.6); agent runtime + registry (9.1–9.2); KB↔RAG feedback loop operating (6.7); IT platform-ops training (10.5); curriculum for all-staff waves finalized (10.2).
• Deliverables: weekly pilot metrics packs, agent runtime, agent registry live.
• Decision gates: none formal — steering reviews pilot interim metrics.
• Dependencies: agent work (9.1) needs GA platform + integrations (5.5).
• Department involvement: pilot departments (steady state ~2–3 h/person/month); IT in ops training.
• Infrastructure milestones: agent runtime deployed; capacity trending reviewed.
• Knowledge base milestones: retrieval-miss triage running monthly; content-gap list worked by champions.
• Mission Control: pilot metrics visible (WAU, queries/user, first time-saved figures).
• Training: IT ops certification; lunch-and-learn series starts (open to all, voluntary).
• Board reporting: pilot interim readout — usage, early KPI deltas, blockers.

M8 — Pilot evaluation; rollout decision

• Main focus: honest pilot evaluation against M1–M3 baselines; first automations ship.
• Key activities: pilot evaluation (8.7); automation wave 1 — 3–5 quick wins into production (9.3); rollout wave plan (11.1); all-staff training wave design (10.4 prep); management track sessions (10.6).
• Deliverables: pilot evaluation report (KPI deltas, CSAT, lessons), wave-1 agents live, rollout wave plan.
• Decision gates: pilot results accepted & full rollout approved (8.7) — the program’s central value checkpoint. Criteria: measured time savings in ≥2 of 3 pilots, WAU ≥60% in pilot depts, CSAT ≥ target.
• Dependencies: rollout plan needs pilot lessons; wave-1 agents need pilot friction data.
• Department involvement: pilot departments (evaluation interviews); dept heads of wave-1 departments briefed; management track (2 × 2 h sessions).
• Infrastructure milestones: straight-through metrics live per agent (9.5).
• Knowledge base milestones: pilot dept SOP coverage ≥ target; wave-1 dept content sprints scheduled.
• Mission Control: pilot results page (the honest version — including what didn’t work); rollout schedule published.
• Training: management track delivered; wave-1 training calendar published.
• Board reporting: pilot evaluation + rollout go decision; mid-program budget review €[TBD] vs actuals.

M9 — Rollout wave 1

• Main focus: first ~4 non-pilot departments onboard; automation wave 2 begins.
• Key activities: wave-1 onboarding (11.2) — accounts, labs, KB sections, 2 use cases per dept; hypercare (11.6); automation wave 2 development (9.4); power-user enablement starts (9.6); content sprints for wave-2 depts (6.5).
• Deliverables: wave 1 live (≈7 of 15 departments active), hypercare log, wave-2 agent designs.
• Decision gates: wave-1 exit check — WAU and incident thresholds met before wave 2 proceeds.
• Dependencies: wave order per readiness scores (1.11); capacity watch as DAU grows (11.5).
• Department involvement: wave-1 depts (training 2 h + hypercare); pilots in steady state; champions guild now 15-strong and weekly.
• Infrastructure milestones: DAU climbing; p95 latency within target under growing load.
• Knowledge base milestones: wave-1 dept sections live; KB health dashboard (6.6) public.
• Mission Control: department-by-department adoption heatmap added.
• Training: all-staff labs for wave-1 depts (~30 people).
• Board reporting: rollout progress, adoption curve vs plan.

M10 — Rollout wave 2; automation scale-up

• Main focus: next ~4 departments; wave-2 agents into production.
• Key activities: wave-2 onboarding (11.3) + hypercare; wave-2 agents production (9.4); power users shipping first user-built automations (9.6); pulse survey #2; laggard analysis from wave 1.
• Deliverables: wave 2 live (≈11 of 15 departments), 5–8 wave-2 agents running, pulse-survey read.
• Decision gates: wave-2 exit check; steering reviews adoption-laggard action plans.
• Dependencies: hypercare capacity — TPL + champions must not be saturated; if wave-1 hypercare overran, wave 3 shifts 2 weeks.
• Department involvement: wave-2 depts (training + hypercare); wave-1 in steady state.
• Infrastructure milestones: GPU utilization reviewed against scale-plan trigger (11.5).
• Knowledge base milestones: company-wide SOP coverage tracking toward M12 target; freshness/orphan metrics in monthly review.
• Mission Control: hours-saved counter live (agent telemetry 9.5); automation showcase page.
• Training: wave-2 labs; power-user workshop series (agents & automation track).
• Board reporting: adoption + automation ROI early indicators.

M11 — Rollout complete; optimization begins

• Main focus: final wave including branches; declare rollout complete; start transition to operating model.
• Key activities: wave-3 onboarding — remaining depts + Akti Miaouli, Perama branch access, warehouse-floor model (11.4); adoption checkpoint (11.7); optimization phase starts — KPI deep-read prep (12.1), model-refresh runbook (12.2); prompt-skill assessments and certification (10.7).
• Deliverables: all 15 departments live; adoption checkpoint report; certification records.
• Decision gates: rollout complete declared (11.7) — WAU ≥ target company-wide; laggard plans in motion for any department below threshold.
• Dependencies: branch network/access readiness (Prime Products IT).
• Department involvement: wave-3 depts (training + hypercare); all others steady state; champions guild shifts to monthly-improvement agenda.
• Infrastructure milestones: full ~50-DAU load sustained; capacity/scale decision for year 2 drafted.
• Knowledge base milestones: all departments have live sections; orphan rate and freshness at target.
• Mission Control: full-company dashboard; “12 months in 12 stories” series starts.
• Training: wave-3 labs; certification push; refresher clinics for early waves.
• Board reporting: rollout completion; preliminary ROI figures.

M12 — Closure, handover, year-2 direction

• Main focus: measure, hand over, decide what’s next.
• Key activities: KPI attainment review vs M12 targets (12.1); permanent AI governance board constituted (12.3); pipeline + agent-registry handover to Prime Products (12.4); year-2 roadmap with budget sketch (12.5); program closure + lessons learned (12.6).
• Deliverables: KPI attainment report, governance charter v2, handover sign-off, year-2 proposal, closure report.
• Decision gates: program closure & ops handover accepted (12.6); board decides on year-2 proposal.
• Dependencies: honest KPI data — no target-gaming in the final month.
• Department involvement: dept heads in KPI review sessions (1 h each); champions in lessons-learned retro.
• Infrastructure milestones: platform fully operated by Prime Products IT; TPL moves to advisory/escalation.
• Knowledge base milestones: KB ownership with departments; review cycles assigned for all core docs.
• Mission Control: final program report published; site transitions to permanent AI-operations portal.
• Training: train-the-trainer complete — champions run future onboarding.
• Board reporting: final board presentation — outcomes vs charter, ROI, year-2 recommendation.