Project Charter — PRIME PRODUCTS AI-First Transformation

Project name

“Program Pharos” (working name — proposal, pending CEO confirmation). A pharos is a lighthouse: fixed on PRIME PRODUCTS’ own ground, guiding vessels safely — an on-prem beacon for a company that has served ships since 1929. Alternative candidates (“Poseidon”, “Kyvernitis” — helmsman) recorded for the naming decision.

Vision

By mid-2027, Prime Products Ltd operates as an AI-first company: every core process documented in the company knowledge base, an AI assistant in every role’s daily flow in Greek and English, repetitive work delegated to monitored agents, and every improvement measured — all running on PRIME PRODUCTS-owned, on-premises infrastructure.

Mission

In 12 months, design, build, and embed the AI-first operating model across all 15 PRIME PRODUCTS departments: stand up the on-prem NVIDIA GPU LLM platform for ~50 daily users, capture the company’s operating knowledge in the Obsidian knowledge base, deliver measurable AI use cases and agents department by department, and hand a self-sustaining operating rhythm to PRIME PRODUCTS roles — with minimal disruption to daily operations.

Strategic rationale: strategic transformation report. Program structure decision: ADR-004. Core needs: REQ-001…REQ-010 in the requirements register.

Scope

• All 15 departments: executive management, sales (7 segments), procurement, logistics, warehouse, finance, accounting, HR, customer service, marketing, product/technical support, IT, compliance/quality documentation, administration, reporting/BI (department plans).
• All PRIME PRODUCTS sites: HQ (Dim. Moutsopoulou 96 & Serifou, Piraeus), Akti Miaouli branch, Perama branch, bonded warehouses.
• On-prem AI infrastructure: hardware procurement through production operation (05-technical).
• Company knowledge base in Obsidian (06-knowledge-base) and Mission Control project website at vanos.tpl.one (07-mission-control).
• AI use-case and agent portfolio with measured baselines (08-ai-portfolio).
• Company-wide discovery, training, and change management.
• Read/write integrations with the existing Microsoft 365 + Teams and SoftOne ERP environment (integrations).

Out of scope

• No replacement of SoftOne ERP. The program integrates with SoftOne; ERP migration is a separate decision outside this mandate.
• No cloud LLM dependency for confidential data. Customer, commercial, and defense-related data is processed only on PRIME PRODUCTS-controlled infrastructure.
• No headcount-reduction program. Freed capacity is redeployed; the program carries no staffing-cut targets, explicit or implicit.
• No customer-facing autonomous AI. Nothing AI-generated reaches a customer, authority, or supplier without approval by a named person.
• No changes to PRIME PRODUCTS legal structure, physical logistics infrastructure, or product-line strategy.
• No general IT modernization beyond what the AI platform requires (assumption on current network/identity readiness — to validate in discovery).

Business objectives

The eight measurable objectives of strategic transformation report §2, in brief: (1) production on-prem platform, ~50 daily users, GR+EN, ≥99% business-hours availability; (2) quote/tender turnaround −50% in pilot segments; (3) ≥80% of identified core processes documented; (4) ≥70% weekly / ≥50% daily adoption; (5) ≥10 production use cases and ≥5 agents; (6) ≥15% time freed in top-3 back-office functions; (7) zero confidential-data incidents; (8) governance and Mission Control operating from month 1. Metrics: KPI framework.

Key stakeholders

Stakeholder	Role in program	Interest / stake
PRIME PRODUCTS Board	Approves budget, mandate, hardware procurement; quarterly oversight	Return on investment, risk exposure, company direction
CEO (PRIME PRODUCTS)	Executive sponsor; chairs steering committee	Program success, company competitiveness, minimal disruption
Department heads (15)	Own their department’s transformation; steering representation	Departmental performance, workload of their teams
Department AI champions	Carry adoption into daily work (~10–15% time)	Practical usefulness, recognition, workload balance
IT lead (PRIME PRODUCTS)	Operates platform infrastructure; security counterpart	Operability, security, support burden
All employees (~109)	Users of assistants and knowledge base; discovery participants	Job security, easier daily work, skill development
TPL transformation lead	Program delivery, methodology, platform build	Delivery quality, staged commercial success
Knowledge manager (PRIME PRODUCTS, to be designated)	Curates the knowledge base	Sustainable knowledge quality
Customers (shipping, defense, public sector)	Indirect — faster quotes, unchanged confidentiality	Service speed, data protection
Suppliers / partners (e.g. UKHO Admiralty)	Indirect — interface processes may change form, not substance	Continuity of cooperation

Governance summary

Full model: governance model. The program is CEO-led: the CEO is executive sponsor and chairs the monthly steering committee, which holds budget-within-envelope, prioritization, and escalation authority. A two-lead structure runs delivery day-to-day — TPL transformation lead paired with a Prime Products-side counterpart. Department heads own departmental change; champions execute it; the IT lead owns platform operations; a security/data owner guards confidentiality. Decision rights, meeting cadence, escalation paths, and reporting rhythm are defined in the governance model and are binding on all workstreams.

Project roles

Role	Filled by	Commitment
Executive sponsor	CEO (PRIME PRODUCTS)	2–3 h/week + monthly steering chair
Steering committee	CEO, 2–3 department heads (rotating rep.), IT lead, TPL lead	Monthly, 90 min + prep
Transformation lead (TPL)	TPL senior consultant	Core engagement
Transformation counterpart (PRIME PRODUCTS)	To be designated by CEO	~50% time (assumption — to validate)
IT lead	Head of IT (PRIME PRODUCTS)	~30–50% during build, ~20% steady state
Department AI champions (15)	Nominated by department heads	~10–15% time
Knowledge manager	To be designated (PRIME PRODUCTS)	~20–30% time from Q2
Security/data owner	IT lead or compliance head (to be decided)	As needed; approves data flows
Platform engineering	TPL	Per staged offer

Workstreams

#	Workstream	Delivers	Key links
1	Governance	Steering rhythm, decision rights, reporting, risk management	governance model, risk register
2	Discovery	Company-wide requirements capture, baselines, use-case backlog	03-discovery
3	Infrastructure	GPU hardware procured, installed, operational	hardware and sizing
4	LLM platform	Serving stack, RAG pipeline, GR+EN models, security	llm serving stack, rag and data pipeline
5	Knowledge base	Obsidian vault structure, templates, capture process	06-knowledge-base
6	Mission Control	Program website at vanos.tpl.one, KPI dashboard	07-mission-control
7	Department pilots	Sequenced pilots with baselines, starting with sales quoting	04-departments
8	Agents & automation	Agent portfolio, registry, human-approval workflows	08-ai-portfolio
9	Training & change	Training waves, communications, adoption support	change plan
10	Rollout & optimization	Company-wide scale-up, KPI attainment, handover to PRIME PRODUCTS roles	roadmap

Decomposition: work breakdown structure. Timeline: 12-month roadmap.

Assumptions

• Board approves budget envelope and hardware procurement at the approval session; hardware lead time (~6–10 weeks, assumption — to validate with suppliers) is absorbed in Q1.
• The CEO sustains the stated time commitment across all 12 months.
• SoftOne ERP data is accessible for read integration via database/API at acceptable quality (assumption — to validate in technical discovery).
• Existing M365 tenant and network infrastructure can support the platform with modest upgrades (assumption — to validate in technical discovery).
• Department heads release champion time (~10–15%) and staff time for interviews and training.
• ~50 daily users is the correct platform sizing for year 1; company-wide passive access (knowledge base, Mission Control) for all ~109 employees.
• Open-weight LLMs available in 2026 provide adequate Greek-language quality for the target use cases (to verify in platform selection, llm serving stack).

Constraints

• Budget: envelope TBD by board; released stage-by-stage per the staged offer structure.
• Time: 12-month program window from board approval.
• Scale: platform sized for ~50 daily active users.
• Operations: minimal disruption — no pilot may degrade daily ship-supply operations; sequencing subject to department heads’ operational calendars.
• Language: all user-facing capability in Greek and English from first release.
• Confidentiality: on-prem processing only for business data; defense/public-sector data handling per customer and regulatory requirements.
• People: no dedicated full-time PRIME PRODUCTS hires assumed for year 1 except role designations above.

Risks (top 8 — summary)

Full scored register with owners and mitigations: risk register.

#	Risk	Direction of mitigation
1	Adoption resistance / fear of job loss	No-headcount-reduction commitment, CEO communication, early wins
2	Champion/staff overload alongside daily operations	Explicit time allocation approved by board; sequencing by operational calendar
3	SoftOne data extraction harder or dirtier than assumed	Early technical discovery spike; fallback to narrower integration scope
4	Tribal knowledge capture stalls	Dedicated knowledge manager role, interview-driven capture, champion incentives
5	Greek-language LLM quality below expectation	Model evaluation gate in Q1; bilingual prompt/RAG design; model swap flexibility
6	Hardware procurement delay	Order immediately on approval; interim development on TPL infrastructure with non-confidential data only
7	Confidential-data incident	On-prem-only architecture, access controls, audit trail (security)
8	Key-person dependency (transformation counterpart, IT lead)	Documented runbooks, TPL backup coverage, knowledge-base-first working style

Dependencies

• Board decisions (budget, mandate, time allocation, hardware) — gate for program start.
• GPU hardware supplier lead times — gates Q2 platform go-live.
• SoftOne vendor/partner cooperation for integration interfaces (assumption — to validate).
• M365 tenant administration access via PRIME PRODUCTS IT.
• TPL federation conventions and tooling (knowledge-OS model) — provided by TPL.
• Champion nominations by department heads — gate for discovery quality.

Success criteria

The program is successful at month 12 if, measurably (KPI framework):

1. The on-prem platform is in production at ≥99% business-hours availability with ~50 daily users supported.
2. Adoption: ≥70% of target users weekly-active, ≥50% daily-active.
3. Quote/tender turnaround in pilot sales segments reduced ≥50% vs. discovery baseline.
4. ≥10 use cases and ≥5 agents in production, each with measured before/after results.
5. ≥80% of core processes identified in discovery are documented in the knowledge base.
6. ≥15% time freed in the top-3 back-office functions, evidenced against baseline.
7. Zero confidential-data incidents over the program period.
8. Operating model handed over: knowledge manager, champions, and IT operations running the platform and rhythm with TPL in advisory posture.
9. Board has received a year-2 portfolio and accepts program closure at the final quarterly review.