Board Presentation Outline — PRIME PRODUCTS AI-First Transformation

Audience: Prime Products Ltd board. Presenter: CEO, supported by TPL. Duration: ~60 min + discussion. Goal: four explicit board decisions (slide 17).

Source content: strategic transformation report; program detail in 02-program. 18 slides.

---

Slide 1 — PRIME PRODUCTS 2027: An AI-First Company

• Vision: by mid-2027, every PRIME PRODUCTS process is documented, every role works with an AI assistant, repetitive work runs through agents — humans decide, AI does the legwork.
• A company supplying ships since 1929 now equips itself for the next century.
• 12-month, CEO-led, company-wide program; owned infrastructure; measurable results.
• This session ends with four concrete decisions we ask of the board.

Visual: Single bold vision statement over a Piraeus/port photograph; “1929 → 2027” timeline motif.

Speaker notes: Open with continuity, not disruption — PRIME PRODUCTS has reinvented itself before (trading house → S.A. → multi-line safety and supply group). Frame AI-first as the next reinvention, done deliberately and from strength. State up front that you will ask for specific decisions at the end.

Slide 2 — Why Now

• Healthy company, right moment: strong balance sheet (~50.6% equity), rebounding pre-tax profit — we can invest from strength.
• But revenue is drifting (−2.2% y/y) and net margin is thin (~1.5% on €23.7M): we need leverage, not more cost.
• Competitors and marketplace platforms are moving on AI in quoting and catalog operations; speed of quote wins provisioning orders.
• On-prem LLM technology is now mature and affordable at our scale (~50 users) — the window to lead rather than follow is open.

Visual: Two-panel chart — revenue trend (slightly declining) vs. pre-tax profit (rebounding); a “window of opportunity” marker at 2026.

Speaker notes: The message is “act from strength before we must act from weakness.” At 1.5% margin, standing still means any cost inflation eats the entire profit. Preempt the “why not wait a year” question: the constraint isn’t technology maturity, it’s the 12–18 months it takes to build capability — competitors started their clock already.

Slide 3 — Our Operating Challenges Today

• Knowledge-heavy commercial work done by hand: quotes and tenders assembled from SoftOne, mailboxes, spreadsheets, and memory — under port-call deadlines.
• Everything twice: Greek-English documentation burden on every customer interaction, tender, datasheet, and training package.
• Tribal knowledge at risk: decades of know-how in long-tenured heads, largely undocumented.
• Reporting by heroics: management numbers assembled manually, days per cycle.
• 15 departments, 7 sales segments, 3 sites plus bonded warehouses — coordination cost grows faster than headcount.

Visual: “A day in the life of a quote” swimlane showing manual hops between systems, mailboxes, and people.

Speaker notes: Keep this affectionate, not accusatory — these are the normal pains of a successful 109-person company that grew on expertise. The point is that each pain is exactly the shape of problem modern AI solves. Use one concrete anecdote from discovery pre-work if available.

Slide 4 — The AI-First Opportunity

• At ~1.5% net margin, €100k of freed cost/capacity ≈ the profit of €6–7M of new revenue — productivity is our highest-yield investment.
• AI now reads, writes, retrieves, and drafts at professional level in both Greek and English — precisely our documentation and quoting workload.
• Run on our own hardware: no data leaves the building, no per-use fees — economics improve with adoption.
• Four levers: faster quoting/tendering, automated bilingual documents, agents for back-office work, company knowledge on tap.

Visual: Lever diagram — “€100k productivity = €6.5M revenue-equivalent” as the headline number.

Speaker notes: This is the pivotal economic slide; slow down here. Walk the margin arithmetic explicitly — board members remember one number, make it this one. Emphasize that ownership of the infrastructure inverts cloud economics: heavy usage costs approximately electricity.

Slide 5 — Strategic Objectives (12 Months)

• Production on-prem AI platform for ~50 daily users, Greek + English, ≥99% business-hours availability.
• Quote/tender turnaround −50% in pilot sales segments; ≥10 production use cases and ≥5 agents live.
• ≥80% of identified core processes documented in the company knowledge base; ≥70% weekly platform adoption.
• ≥15% time freed in top-3 back-office functions — redeployed, not cut; zero confidential-data incidents.

Visual: Objective scorecard — 8 objectives as tiles with target values (full list in report §2).

Speaker notes: Stress that every objective is measurable and will be reported monthly — this program will not hide behind activity metrics. Note explicitly the “redeployed, not cut” phrasing; it is a governance commitment, not a slogan, and the board will be asked to stand behind it.

Slide 6 — Target Operating Model

• Every process documented in the knowledge base — undocumented work can’t be assisted or safely handed over.
• An AI assistant in every role’s daily flow, bilingual, where the work happens.
• Agents handle repetitive multi-step work — registered, scoped, monitored.
• Humans approve everything external-facing: AI drafts, people sign. Non-negotiable.
• Measure everything: every use case ships with a baseline and a metric.

Visual: Concentric model diagram — knowledge base at the core; assistants, agents, human approval, measurement as rings.

Speaker notes: This slide defines “AI-first” so nobody leaves with the Hollywood version. The human-approval principle matters doubly for our defense and public-sector customers. Point out the model adds a thin layer (champions, knowledge manager) but restructures nothing.

Slide 7 — What Stays the Same

• SoftOne ERP stays — we integrate with it, we do not replace it.
• No headcount-reduction program — freed time is redeployed to sales, service, and quality.
• No customer-facing autonomous AI — a named person approves everything that leaves the company.
• Daily operations continue undisturbed: pilots run alongside existing processes until proven.

Visual: Simple two-column “Changes / Stays the same” table.

Speaker notes: Deliberately placed early — this slide removes the three fears (ERP replacement trauma, layoffs, robots emailing customers) that otherwise contaminate every later discussion. These are formal out-of-scope items in the project charter.

Slide 8 — 12-Month Roadmap

• Q1 Foundations & Discovery: governance live, company-wide discovery, baselines, hardware procured, knowledge base + Mission Control online.
• Q2 Platform & First Pilots: LLM platform in production, M365/SoftOne integration, sales-quoting pilot plus 2–3 departments, first agents.
• Q3 Scale: rollout to all 15 departments, ≥10 use cases live, training waves, knowledge base ≥60% coverage.
• Q4 Optimize & Institutionalize: KPI targets hit, operating model handed to PRIME PRODUCTS roles, year-2 portfolio.

Visual: Quarter-level timeline bar with milestone diamonds; detail in roadmap.

Speaker notes: Emphasize monthly visible increments — no six-month silent build. First business value (sales quoting pilot) lands in Q2, not at month 12. Q1 discovery is where employees are heard, which doubles as the change-management opening move.

Slide 9 — Investment Areas

• Infrastructure (capex): on-prem GPU server, networking, resilience — one-off, ~4–5-year life. [€XX–XXk placeholder]
• Platform & integration: LLM serving stack, M365 + SoftOne connections, security. [€XX–XXXk placeholder]
• Transformation services: discovery, department pilots, agents, knowledge base, Mission Control. [€XX–XXXk placeholder]
• People: training & change program, plus internal time (champions ~10–15%, interviews, training).
• Run cost after go-live: modest — power, support, licenses; no per-token cloud fees.

Visual: Cost-block bar with capex/opex/services split; placeholders clearly flagged “final in stage offer.”

Speaker notes: Numbers are ranges by design; binding figures come with the staged commercial offer (stage-offer structure) after discovery sizes the work. Highlight internal time as the honest hidden cost — the board should approve it consciously, which is decision 3 on slide 17.

Slide 10 — On-Prem AI Infrastructure

• NVIDIA GPU server on PRIME PRODUCTS premises serving open-weight LLMs for ~50 daily users, Greek + English.
• Customer, commercial, and defense data never leaves company-controlled infrastructure — a hard requirement for our military and public-sector business.
• Owned asset: predictable capex, no usage-based fees, full control of models and data.
• Integrated with what we already use: Microsoft 365, Teams, SoftOne ERP.

Visual: One-box architecture sketch: users → assistants/agents → on-prem GPU platform → knowledge base + SoftOne + M365 (detail in architecture overview).

Speaker notes: Keep it non-technical: “our own AI, in our own building, speaking both our languages.” The confidentiality argument is decisive for the defense segment — some of our customers could not accept cloud processing at all. Technical detail exists in 05-technical for anyone who wants depth.

Slide 11 — Department Transformation Plan

• All 15 departments in scope; sequenced by impact and readiness, not all at once.
• High impact first: sales (quoting/tendering), customer service, accounting, procurement, reporting/BI, compliance/quality.
• Each department gets: an AI champion, a discovery-based use-case backlog, pilots with baselines, training.
• Department heads own their transformation; TPL and the platform enable it.

Visual: Heat-map table — 15 departments × impact level × primary AI lever (from report §7).

Speaker notes: The heat map shows this is a whole-company program, not an IT project — every board member’s area is on the map. Sequencing protects operations: no department takes on change during its peak load without its head’s agreement.

Slide 12 — Governance: CEO-Led

• CEO as sponsor: sets direction, unblocks, chairs monthly steering — 2–3 hours/week.
• Steering committee (CEO, department-head representation, IT lead, TPL transformation lead) meets monthly: progress, KPIs, decisions, risks.
• One AI champion per department (~10–15% time) carries the transformation into daily work.
• Clear decision rights and escalation paths — decisions in days, not months.

Visual: Governance organogram — board → CEO/steering → transformation leads → champions; cadence table alongside (full model: governance model).

Speaker notes: CEO leadership is the single strongest predictor of transformation success at this company size — this is why the mandate request (slide 17) is explicit. The structure is deliberately light: one new committee, no new hierarchy.

Slide 13 — Risk Management

• Top risks are human and data-related, not technological: adoption resistance, fear of job loss, champion overload, SoftOne data-extraction quality, tribal-knowledge capture.
• Mitigations built into design: no-headcount-reduction commitment, CEO visibility, training program, early data workstream, on-prem-only confidentiality.
• Live, owned, scored risk register reviewed at every monthly steering (risk register).
• Staged commercial structure limits financial exposure: each stage approved on the results of the last.

Visual: 2×2 probability/impact matrix with top-8 risks plotted.

Speaker notes: Be candid: the technology is well-trodden; the program fails, if it fails, on people and data quality — so that’s where mitigation money and attention go. The staged offer is itself a risk control the board should value: no single large irreversible commitment.

Slide 14 — Expected Business Impact

• Sales leverage: −50% quote/tender turnaround → more quotes out, more tenders answered, at unchanged headcount — directly addressing the revenue drift.
• Productivity: ≥15% time freed in top-3 back-office functions ≈ 3–5 FTE-equivalents of capacity redeployed.
• Resilience: company knowledge captured and searchable — key-person risk hedged, onboarding accelerated.
• Payback logic: at 1.5% margin, either the capacity effect or the sales effect alone plausibly covers program cost within 18–24 months of go-live; they are additive.

Visual: Benefit waterfall — capacity freed + sales throughput + error reduction → payback horizon.

Speaker notes: Anchor again on the margin arithmetic from slide 4. Commit to honesty: every benefit claim will be measured against a baseline captured in discovery — the board will see actuals, not projections, from Q2 onward.

Slide 15 — How We Will Measure Success

• Monthly board-visible scorecard: adoption (weekly/daily actives of ~50), quote turnaround vs. baseline, hours freed, processes documented, use cases live, availability, data incidents (target zero).
• Every use case: baseline before, measurement after — no baseline, no benefit claim.
• Live transparency: Mission Control at vanos.tpl.one shows program status to the whole company.
• Full metric definitions, owners, and targets in the KPI framework.

Visual: Mock-up of the Mission Control KPI dashboard.

Speaker notes: Transparency is a feature, not a risk: staff who can see the program’s honest status trust it. The board gets the same numbers management gets, one click away, at any time.

Slide 16 — Team & Partners

• PRIME PRODUCTS: CEO sponsor, Prime Products-side transformation counterpart, IT lead, 15 department champions, knowledge manager.
• TPL: transformation lead, LLM platform engineering, knowledge-base and Mission Control build, training delivery.
• Operating principle: TPL builds capability into PRIME PRODUCTS — by month 12 the platform, knowledge base, and operating rhythm run on PRIME PRODUCTS roles.
• Commercial engagement staged by phase (stage-offer structure).

Visual: Two-column partnership diagram with named roles (roles, not persons) and the handover arrow across the 12 months.

Speaker notes: Address the dependency question before it’s asked: the explicit design goal is PRIME PRODUCTS self-sufficiency at month 12, with TPL moving to an advisory/maintenance posture. Champions and the Prime Products-side counterpart are the capability-transfer vehicle.

Slide 17 — Decisions Requested Today

• 1. Budget approval: approve the program budget envelope [€XXX–XXXk placeholder range] with stage-gated release per the staged offer.
• 2. Governance mandate: confirm the CEO as executive sponsor and authorize the steering committee with the decision rights in the governance model.
• 3. Department time allocation: authorize ~10–15% time for one AI champion per department and staff participation in discovery and training.
• 4. Hardware procurement authorization: approve procurement of the on-prem NVIDIA GPU infrastructure [€XX–XXk placeholder] to start immediately (longest lead-time item).

Visual: Four decision cards, each with a checkbox — literally designed to be resolved in the meeting.

Speaker notes: Ask for each decision individually and record each in the minutes. If the board wants to stage its commitment, decisions 2–4 plus Q1 budget suffice to start; full-envelope approval can follow the discovery readout. Do not leave the room without decision 4 — hardware lead time gates the entire Q2 plan.

Slide 18 — Next Steps (First 30 Days)

• Week 1: steering committee constituted; program kickoff; hardware order placed.
• Weeks 1–2: champions nominated by department heads; all-hands announcement by the CEO.
• Weeks 2–4: discovery launch — questionnaires to all employees, interviews with department heads and management (discovery).
• Week 4: Mission Control live at vanos.tpl.one; first monthly steering with baseline scorecard.

Visual: 30-day calendar strip with the four milestone rows.

Speaker notes: End with momentum and specificity — approval today means visible motion within a week and the whole company engaged within a month. The CEO’s all-hands announcement is the change-management keystone: the program should be heard first from the CEO, not through the grapevine.