Executive alignment and governed value

The counter is not caution. It is governed pace. You move quickly by making the next decision smaller, clearer, and reviewable. OmniCorp Group uses this lesson at the point where Miriam Ashworth, group chief executive, wants momentum, Rosalind Achebe, board chair, wants oversight, and Priyanka Venn, group chief financial officer, wants value that survives audit. Those three demands are compatible only when the work is framed as accountable outcomes, not as AI enthusiasm.

Definition

The Value Charter Gate is the executive method for converting AI ambition into a governed value claim before money, reputation, or scarce delivery capacity is committed. It requires four lines: the enterprise outcome AI should improve, the accountable executive owner, the evidence that will prove value, and the next decision date. If any line is missing, the gate does not open.

The gate is deliberately strict. It refuses the language of transformation until the enterprise can say what will be better, by how much, for whom, under whose authority, and by when. It is not a business case template. It is the minimum proof that the ambition is governable.

When to Use It

Use the gate before the board receives an AI strategy paper, before an operating company launches an AI programme, before a vendor pilot is announced as strategic, and before an executive committee approves funding for a cross-functional initiative. Use it when the organisation is tempted to declare direction before naming evidence.

Do not use it to manage every experiment. Local teams can still test low-risk workflow improvements inside existing controls. Use the gate when the claim is enterprise-level: customer value, regulated decisioning, workforce redesign, clinical safety, operational resilience, or public trust. Big claims need governed entry.

How to Apply It

1. 1Write one enterprise outcome AI should improve, using a business unit the board already tracks.
2. 2Name the executive owner who can accept the trade-offs, not the sponsor who likes the idea.
3. 3State the evidence standard: baseline, target, control, regulator position, safety threshold, or customer measure.
4. 4Separate known evidence from untested expectation in two columns. Do not blend them in one sentence.
5. 5Set the next decision, owner, evidence package, and review date before any announcement or funding release.

The phrasing matters. Write: AI should reduce manual underwriting rework in OmniCorp Financial by 20 percent while maintaining model-risk controls. Do not write: AI will transform lending. One sentence governs. The other decorates.

Worked example 1 of 3

At OmniCorp Group, Miriam Ashworth wanted to announce a group AI acceleration programme after a strong vendor presentation. Priyanka Venn stopped the draft at the value gate. The first proposed outcome was enterprise productivity. Priyanka rejected it because no ledger could hold it. The revised outcome was specific: reduce manual case preparation time in consumer lending at OmniCorp Financial by 18 percent within two quarters, without increasing model-risk exceptions.

MAMiriam Ashworth

I want the market to hear that OmniCorp Group is moving, but I will not brief a promise we cannot govern.

PVPriyanka Venn

Then the first decision is not the announcement. It is whether lending has enough evidence to deserve group funding.

RARosalind Achebe

Bring the board the outcome, the owner, the evidence, and the next review date. The board can oversee that.

Why This Works

The gate works because it makes ambition accountable before the story hardens. Once an announcement is made, the organisation starts defending the announcement. Teams become reluctant to surface weak evidence because weakness feels like disloyalty. The gate prevents that distortion by making evidence the condition of movement, not an afterthought.

It also gives the board a real oversight handle. Rosalind Achebe does not need to approve model architecture to govern the decision. She needs to know the outcome, decision owner, evidence standard, material risks, and review date. Oversight is not technical theatre: it is disciplined accountability over value and risk.

Definition

The Oversight Rights Map is the method for naming who decides, who reviews, who may stop, and who must be informed for an enterprise AI initiative. It separates board oversight, executive ownership, control-function review, operating accountability, and delivery authority. The map prevents AI work from floating between committees until momentum substitutes for permission.

Decision rights are not titles on a slide. They are permissions attached to named choices. The board approves risk appetite and material commitments. The executive owner accepts trade-offs. Control leaders set non-negotiable conditions. Operating leaders own adoption. Delivery teams build within the boundary. Confusion here costs months.

When to Use It

Use the map when an AI initiative crosses operating companies, touches regulated data, changes customer or patient outcomes, alters workforce roles, or requires board visibility. Use it before the first steering committee, not after the first escalation. The map is most valuable when everyone believes alignment already exists.

Do not use the map to create more committees. A rights map should reduce meetings by making the next decision obvious. If it creates a standing forum with no named decisions, it has become governance decoration.

How to Apply It

1. 1List the next four decisions in chronological order: fund, pilot, expand, embed, or stop.
2. 2Attach one named decision owner to each decision. Use a person, not a committee name.
3. 3Name who has stop rights for risk, safety, legal, security, or customer harm.
4. 4Name what the board sees and what it does not need to see.
5. 5Record the review date and evidence pack for each decision before work begins.

The rule is simple: no decision without an owner, no owner without evidence, no evidence without a date. If a committee owns a decision, nobody owns it. Committees can advise. People decide.

Worked example 1 of 3

OmniCorp Logistics explored an AI dispatch optimisation system. Yusuf Demir, vice president of network operations, owned the operating outcome: fewer late routes and lower overtime. Claire Beaumont, head of dispatch systems, owned delivery. Anders Nilsen, platform engineering lead, owned platform constraints. Daniel Osei, group chief information security officer at OmniCorp Group, held stop rights on security conditions. Rosalind Achebe received the board view: operational resilience, workforce impact, and incident thresholds.

Why This Works

The map works because enterprise AI decisions fail at the seams. A business leader assumes technology approved the tool. Technology assumes legal approved the use. Legal assumes the business accepted the operational trade-off. The board receives a polished update and discovers later that no one could point to the person who had authority to stop. Rights mapping closes that seam.

It also protects speed. A named stop right does not slow the work. It prevents a late-stage veto after twelve weeks and 800 delivery hours. The cost of ambiguity is not philosophical. It is delayed approval, duplicated analysis, and a regulator's letter asking who approved the decision path.

OmniCorp Financial began with a broad ambition: use AI to improve lending. Adaeze Okonjo, chief risk officer, refused to let that sentence enter the executive committee pack. She asked Martin Hollis for the enterprise outcome. The first draft said faster underwriting. Adaeze narrowed it to manual case preparation time in consumer lending, because that process already had baseline hours, rework rates, exception logs, and named controls.

Priyanka Venn reviewed the value line from the group centre. She separated evidence from expectation. Known evidence: consumer lending teams spent an average of 3.4 hours preparing a complex case file, and 14 percent of files returned for missing information. Observed evidence: one internal prototype had reduced preparation time on ten historical files. Untested expectation: the same reduction would hold on live files with current controls. That distinction changed the decision. The committee approved discovery, not rollout.

Bea Karlsson, model risk lead, set the evidence standard. The pilot would need baseline hours per case, rework rate, exception count, adverse-action communication review, and a sample of declined cases. Yuki Tanaka reviewed the customer communication boundary. No output could be used to explain a customer decision without legal review. The evidence standard was not a blocker. It was the condition that made the value claim usable.

Rosalind Achebe received the board oversight line one week later. The board was not asked to approve a tool. It was asked to note the risk appetite boundary, the controlled discovery decision, and the review date. Miriam Ashworth approved the internal message after the gate was complete. The message said OmniCorp Group was testing AI support for lending case preparation under named risk controls. It did not promise transformation. It named governed movement.

Six weeks later, the evidence pack showed a 16 percent reduction in preparation time, stable rework, and two model-risk exceptions that were contained before customer impact. Adaeze recommended a second controlled pilot rather than full scale because the target was 18 percent and the exception pattern needed more review. The decision was defensible because the gate had preserved the difference between evidence-backed value and untested expectation from day one.