Executive Management

Pilot wave 2 (M7). Impact: high. See the hub. The CEO leads the transformation personally; this plan covers executive management as users of AI — the leadership team modeling the behavior it asks of the company.

Current state (assumptions to validate)

Executive management runs a ~109-person, multi-segment business on a thin ~1.5% net margin, where a bad month in one segment can erase the year’s profit. Decisions span very different worlds — vessel-supply operations, tender bids with multi-year implications, e-commerce investment, fleet and warehouse capex — and the information for each arrives in a different shape: SoftOne exports reworked in Excel, verbal updates from department heads, email threads, and periodic accountant-produced financials.

Assumptions — to validate in discovery: executive team ~4–6 (CEO plus segment/function heads); no consolidated management dashboard; monthly reporting assembled manually with lag; management meetings minuted inconsistently or not at all, so decisions and action items evaporate; board/shareholder packs built from scratch each cycle; significant CEO time on drafting formal correspondence (banks, major clients, authorities) in English and Greek. Pain points: decision latency from information assembly, no single source of truth on performance, meeting outcomes untracked, key-person dependency on what was decided and why.

Target AI-first operating model

By M12 the executive team starts the week with an AI-drafted management briefing: KPI movements with narrative (what changed, why, which segment — grounded in the BI semantic layer), exceptions needing attention, and status of prior decisions. Management meetings are recorded and summarized; decisions and action items land in a tracked register and this knowledge base. Board and shareholder packs are assembled from standing sections the AI drafts and executives edit. Ad hoc questions (“how did navigation do vs last quarter?”) go to the natural-language query assistant rather than to an analyst’s Excel queue. Every decision, every external commitment, and every board-pack sign-off remains human — the AI compresses the path to the decision point.

AI use cases

Use case	Pain addressed	Data needed	Complexity	Impact	Pilot
Weekly KPI narrative briefing (movements + causes + exceptions)	Manual assembly, decision latency	BI semantic layer over SoftOne, dept KPI feeds	M	H	Y
Meeting summarization → decision & action register	Evaporating decisions	Teams meeting recordings/transcripts	L	H	Y
Board/shareholder pack drafting (standing sections from live data)	Built from scratch each cycle	Financials, KPI data, segment commentary	M	H	N
Decision-support briefs on demand (e.g. “should we bid tender X” — assemble history, capacity, margin data)	Fragmented inputs to big decisions	SoftOne, tender history, capacity data	H	H	N
Executive correspondence drafting (banks, authorities, key clients, EN/GR)	CEO drafting time	Correspondence history, company docs	L	M	N
Competitive/market briefing digests (ports, shipping market, defense tenders)	No systematic scanning	External sources, tender portals	M	M	N

Process transformation opportunities

• Management rhythm: anchor the weekly executive meeting on the AI briefing — same structure every week, exceptions first — instead of round-table verbal updates.
• Decision register: every management decision captured with context and owner in the knowledge base; the assistant answers “what did we decide about X and why.”
• Reporting cycle: monthly pack produced by refresh-plus-edit, not rebuild; the analyst effort moves to analysis.
• Delegation via KPIs: department heads answer to the same numbers executives see (shared semantic layer with BI), collapsing the reporting theater.

Required data sources

• BI semantic layer over SoftOne (revenue, margin, cash, inventory, AR by segment) — dependency on business-intelligence.md.
• Department KPI feeds per KPI framework; finance close outputs.
• Teams meeting recordings/transcripts; executive mailboxes (correspondence history, opt-in).
• Past board packs and shareholder communications (SharePoint).

Potential AI agents

• Executive Briefing Agent — drafts the weekly KPI narrative briefing; CEO/CFO-level reviewer approves before distribution.
• Meeting Minutes Agent — summarizes recorded meetings into minutes + action register; meeting chair approves the record.
• Board Pack Drafting Agent — assembles pack sections from live data; each section owner approves, CEO signs off the pack.

Automation opportunities

• n8n: Monday-morning briefing generation and distribution to the executive channel.
• Action-item reminder flow from the decision register.
• Monthly pack skeleton auto-refresh on close completion.

Required integrations

• BI semantic layer (read); Teams (recordings, transcripts, distribution); SharePoint (packs, register) — see integrations.

KPIs

KPI	Baseline	M12 target
Weekly briefing produced without analyst rebuild	0	100% of weeks
Management decisions logged with owner & due date	TBD (~0)	100%
Action items closed on time	TBD	≥80%
Board-pack preparation effort	TBD	−50%
Executive assistant weekly active use	0	100% of exec team

Risks

• Briefing narrative asserts a wrong cause and steers a decision → every claim traceable to figures; label inference vs fact; human review before circulation.
• Meeting recording chills discussion → clear policy on what is recorded; sensitive sessions off-record by chair’s call.
• Highest-sensitivity data concentration (payroll-adjacent, strategy, M&A-type topics) → executive corpus strictly scoped; separate access tier per architecture.
• Executives delegate briefing reading to assistants → CEO-led adoption; briefing is the meeting agenda, unavoidable by design.

Training needs

• Executive team: assistant fluency, NL querying, prompt habits — private sessions, early (M4–M5), because leaders must model usage.
• Chair roles: minutes-approval workflow (M7).
• Executive assistant/analyst: briefing-pipeline operation (M7).

Deliverables

• Weekly briefing pipeline (pilot M7, standing from M8).
• Decision & action register in the knowledge base; meeting-minutes workflow.
• Board-pack template with AI-drafted standing sections (first live pack ~M10).

12-month execution milestones

Month	Milestone
M1–M2	Discovery: decision-flow interviews, reporting-pack autopsy, KPI wishlist
M3	Executive KPI set agreed (feeds KPI framework top layer)
M4–M5	Private executive training; assistant in daily use for drafting/Q&A
M7	Wave-2 pilot: weekly briefing + meeting summarization on live management meetings
M8	Briefing becomes the standing meeting anchor; decision register live
M10	First AI-assisted board pack produced
M12	KPI review; executive rhythm fully institutionalized