The PE 100-Day AI Authority Plan for Portfolio Companies

# The PE 100-Day AI Authority Plan for Portfolio Companies

After close, the unsigned AI call becomes an operating risk — not a diligence footnote. The question is no longer “should we buy?” It is: in the held PortCo, who may spend, contract, change models, and stop — before use cases scale?

Christian Pobbig and Beyond Chiefs work from Hamburg on AI Executive Search in DACH. This piece is an authority plan for the first 100 days after close. The pre-close signature test stays on the diligence page.

After close, the unsigned AI call becomes an operating risk

Institutional post-close logic starts with Value Creation Plan alignment, a short list of levers, a KPI dashboard, and a clear sponsor liaison (Bain: A PE fund just bought your company). Bain also states that new owners replace somewhere between 50% and 70% of PortCo CEOs during the hold — cite as Bain’s claim about CEO turnover pressure, not as AI-authority proof and not as a DACH rate.

Without a signed decision-rights model, every pilot programme scales ambiguity with it. Diligence may have found the unsigned call; the first 100 days must name it.

Institutional playbooks mobilize — few publish a thresholded authority matrix

Bain’s GenAI field notes (survey of private investors representing $3.2T AUM, Sep 2024): most PortCos still in test/development; nearly 20% have operationalized use cases with concrete results. Leading firms use fund-level capabilities, governance protocols, vendor ecosystems, and portfolio learning (Bain GPR 2025 field notes). Case percentages from Vista/Apollo/Hg anecdotes do not travel here as market standards.

BCG PE survey (n=100 senior PE investors): competing priorities (90%) and unclear ROI (76%) block the digital-to-AI journey; successful patterns assign C-level ownership, budget decisions, and integrated roadmaps — and launch 100-day digital squads immediately post-acquisition (BCG digital-first PE).

BCG’s 100-day cost reset: the first 100–180 days are the credibility window; conditions include sponsor–management alignment from day one, accountability inside a 100-day cadence, and confronting unclear decision rights early (BCG 100-day cost reset). Case savings figures are not republished as BC stats.

PwC: assess PortCo GenAI readiness (tech, data, governance, culture, strategy, hold) and assign governance/risk responsibilities before rinse-and-repeat deployment (PwC GenAI for PE PortCos). Readiness is an input to authority — not this article’s spine.

Optional: post-close confirms roles, decision rights, and operating cadence in the first 100 days (CLA) — one sentence, not a PMI essay.

Four decisions that must be explicit by Day 100

BC frame (INFERENCE):

1. Spend / budget — who may commit AI opex/capex within which bands
2. Vendor / contract / data terms — who approves vendor, retention/training use, location, exit
3. Model / architecture / change — who approves a production model change
4. Override / stop — who may halt or downgrade; trigger; log; who is notified (sponsor/board)

Title (CAIO/COO/CFO/CTO) is secondary. Thresholds and escalation are primary.

Day 1–10 / 11–30 / 31–60 / 61–100: what must be signed — not which use case ships

Practice windows (not law):

| Window | Holder decision | Minimum artifact |
|---|---|---|
| Day 1–10 | Who is accountable at fund, board, CEO/COO/CFO, and function? What is live/shadow? | Authority charter draft; live/shadow inventory; what may continue vs must pause |
| Day 11–30 | Who owns spend, procurement, and data/model constraints? | Spend bands; vendor rule; data/retention guardrails; exception owner |
| Day 31–60 | Who selects models and approves production change? Who owns the risk boundary? | Decision-rights matrix; model-change and vendor-exception register; VCP-linked use-case register (use cases test rights) |
| Day 61–100 | Who may launch, override, downgrade, or stop — and who is informed afterward? | Override/kill protocol; board-ready authority view; handoff to ongoing cadence |

No invented dollar bands as “market standard.” Bands are PortCo/fund design.

Fund guardrails vs PortCo execution

Fund standards and PortCo execution must be written: what PortCo may decide, what is fund standard, which exceptions exist, and where they escalate. Without that boundary, every vendor or model change becomes politics.

Use cases test rights — rights are not a use-case list

Institutional playbooks prioritize a short list of strategic use cases. In this spoke, use cases are tests of the four decisions. Scale a use-case list first and leave decision rights open, and you scale ambiguity.

Optional bridge (global, not PE, not DACH): IBM IBV — 68% of 1,000 C-suite say AI adoption slowed because decision rights and escalation are unclear (IBM Where AI breaks).

Boundary to other Beyond Chiefs pages

| Page | Timing / question |
|---|---|
| Live diligence (`ai-leadership-due-diligence-pe`) | Pre-close: who holds the call in the target — before underwrite |
| This spoke | Day 1–100 after close: who may spend / contract / change / stop in the held PortCo |
| Art. 4 / reporting line / organ liability / hiring | Other URLs — boundary only here |

No diligence scoreboard as hero. No “should we buy?” No CAIO hiring brief.

FAQ

### Does every PortCo need a CAIO by Day 100?

No. By Day 100 it needs signed thresholds for spend, vendor, model, and override — regardless of title.

### Is an approved-vendor list enough?

No. Without contract terms on retention/training use, location, portability, and exit, the list is cosmetics.

### What does “human oversight” mean here?

A named override/stop with trigger, log, and notification to sponsor/board — not a slogan.

---

Draft inventory. Not live. No service CTA.