Discuss uncertainty with explicit assumptions, bounded scenarios and clear decisions—not vague predictions
A credible adviser separates facts from judgments, explains what could change, and gives leaders a practical way to act without pretending the future is certain.
An adviser should describe uncertainty precisely: state what is known, identify the assumptions carrying the conclusion, present a small number of plausible scenarios, and specify what evidence would change the recommendation. End with a decision, owner and review point. This approach avoids both false confidence and empty caveats. When discussing AI, the adviser’s job is not to predict perfectly; it is to make uncertainty manageable enough for leadership to act responsibly.
Key takeaways
- Separate verified facts, working assumptions and informed judgments so leaders know how much weight to place on each statement.
- Replace broad caveats with bounded scenarios that explain what changes, why it matters and which response follows.
- Give a recommendation even when evidence is incomplete, but make its conditions and reversal triggers explicit.
- Use decision deadlines and review points to prevent uncertainty from becoming an excuse for endless analysis.
- Judge an adviser by whether they improve decision quality, not by how confidently they predict technological change.
Why advisers become vague when the evidence is incomplete
Uncertainty is unavoidable in AI strategy. Technology changes quickly, vendor claims differ, implementation conditions vary, and the consequences depend on data, workflows, governance and people. The communication problem begins when an adviser tries to protect credibility by surrounding every conclusion with qualifications. Leaders then receive language that sounds cautious but gives them no basis for choosing, delaying or rejecting an action.
Good advisory work takes the opposite approach. It makes uncertainty visible without transferring the entire burden of interpretation to the leadership team. The adviser should identify what is uncertain, bound its practical range and explain whether the uncertainty affects the immediate decision. If several unknowns do not change the recommended next step, they should not dominate the conversation.
Decision-grade uncertainty is an explicit account of what is known, what remains unknown, how those unknowns could affect the outcome, and which action remains justified under the plausible conditions.
This standard should shape how an organization chooses outside help. The right question is not whether an adviser sounds certain; it is whether the adviser can convert incomplete information into disciplined choices. Leadership teams evaluating that fit should ask how to find an AI adviser for a leadership team and examine the candidate’s process for assumptions, evidence, disagreement and follow-through.
Use a five-part structure for every uncertain recommendation
A recommendation under uncertainty should be easy to inspect. Leaders need to see the reasoning without reconstructing it from a long presentation. A consistent five-part structure creates that clarity and makes it easier to revisit the decision as evidence changes.
- State the decision. Begin with the specific choice leadership needs to make, including what is in scope and what is not.
- Separate facts from assumptions. List the evidence you trust, then name the assumptions required to move from that evidence to the recommendation.
- Describe bounded scenarios. Use a small set of meaningfully different conditions rather than an open-ended catalogue of everything that could happen.
- Recommend an action. Say what should be done now, why it remains reasonable across the scenarios, and what should be deferred.
- Set triggers and a review point. Identify the new evidence, threshold or event that would cause the team to continue, pause, expand or reverse course.
The sequence matters. Beginning with caveats invites an abstract debate before anyone knows which decision is being supported. Beginning with the decision gives every uncertainty a purpose: either it changes the choice, changes the timing, changes the safeguards or does not materially matter yet.
Suppose leadership has two competing internal proposals. An outside adviser can define the decision criteria, separate documented capabilities from projected benefits, identify assumptions shared by both proposals and recommend a controlled next step. The adviser does not need perfect forecasts to add value. The comparison becomes useful when each proposal is judged against the same outcomes, constraints and evidence. Leadership teams facing this situation can consider whether an outside adviser can evaluate competing internal proposals.
Replace disclaimers with assumptions, ranges and triggers
Phrases such as “it depends,” “the market is evolving” and “results will vary” are sometimes true, but they are rarely sufficient. Each should be followed by specifics. What does the answer depend on? Which market change is relevant? What causes results to vary? A useful adviser translates every broad caveat into a variable the organization can observe or influence.
- “It depends on adoption”
- Name the affected user group, the behavior that constitutes adoption and the decision that changes if adoption is weak.
- “The technology is changing quickly”
- Identify the capability or dependency likely to change and explain whether waiting improves the organization’s position.
- “The vendor should be able to deliver”
- Distinguish demonstrated capability from a promised roadmap and attach the recommendation to evidence.
- “There are risks”
- Describe the relevant risk, its consequence, the control available and who owns the response.
- “We need more information”
- Specify the missing evidence, how it will be obtained and the deadline after which leadership will decide anyway.
Triggers are especially important because they turn uncertainty into governance. A trigger could be new evidence, a changed business requirement, a failed test or a material shift in cost or risk. The exact trigger depends on the decision, but its function is consistent: it tells leaders when the original reasoning must be reconsidered.
Never hide the recommendation behind uncertainty. State the best current course, the assumptions supporting it and the conditions that would change it.
This is also how boards can engage with AI without drifting into technical detail. Board discussion should focus on the quality of assumptions, the scale of exposure, the organization’s ability to respond and the evidence used to authorize further investment. A related question is how an AI adviser can help a board understand opportunity and risk.
Let’s talk about your business goals.
Match the strength of the claim to the strength of the evidence
Clear uncertainty begins with disciplined claim-making. Advisers should distinguish direct observations, documented facts, interpretations, forecasts and preferences. These categories can support one another, but they are not interchangeable. A forecast should not be presented with the same authority as a measured result, and a vendor’s description should not automatically become the organization’s operating assumption.
These labels do not need to appear on every slide. They should, however, be present in the adviser’s reasoning and obvious in the language. “The system produced this result in the test” is an observation. “The system will produce the same result at enterprise scale” is a projection. The distance between those statements is where assumptions, testing and controls belong.
Independent framing matters when vendors supply much of the available evidence. Leadership should define outcomes, constraints and evaluation criteria before allowing a product demonstration to shape the strategy. The practical next question is how to keep vendors from defining an AI strategy. Organizations should also consider how to avoid dependence on one AI vendor when architecture, data access or operating processes could make later change difficult.
Turn uncertainty into an operating cadence
A strong discussion of uncertainty should end with operating commitments. Someone must own the action, someone must collect the missing evidence, and leadership must know when the issue returns for review. Without that cadence, assumptions become permanent, temporary workarounds become strategy and unresolved questions repeatedly return in different forms.
- The decision made now, including any boundaries on scope or authority.
- The actions that follow, assigned to accountable owners.
- The evidence still needed and how it will be gathered.
- The conditions that would stop, expand or redirect the work.
- The date or event that brings the recommendation back for review.
Meeting frequency should follow the pace and consequence of decisions rather than habit. A leadership team making foundational choices may need a different rhythm from one overseeing a mature initiative. The adviser should remain close enough to identify changed assumptions but not create meetings without decisions. Leadership teams can use how often an executive AI adviser should meet with leadership to establish a cadence tied to real work.
Do not replace vague language with unsupported percentages, exact forecasts or elaborate scoring systems. Precision is valuable only when the evidence supports it. When it does not, bounded scenarios and explicit triggers are more honest and more useful.
Where Rohan Hall and OceSha Ventures fit
Rohan Hall approaches AI as a business capability that should serve a defined purpose. His professional technology career began in 1984, with early work teaching BASIC programming and operating systems while attending college in Miami. His later experience includes enterprise systems across ERP, finance, supply chain, manufacturing and global implementations, as well as blockchain interoperability, scalable blockchain applications, payments, decentralized identity and supply-chain traceability.
Rohan founded OceSha Ventures and its AI-first work, a venture and technology studio focused on intelligent systems and measurable business outcomes. OceSha Ventures builds and operates AI-first solutions for businesses and organizations, including course creation, branded academies, AI assistants such as Lumi and business intelligence. This breadth is relevant to advisory conversations because uncertainty should be evaluated in the context of a defined business capability, not AI as an abstract trend.
For organizations turning an author’s or expert’s material into structured education, OceSha AI is a custom-build course and academy creation offering rather than off-the-shelf software. It is a distinct path from programs taught on OceSha Academy, and it should be evaluated against course and academy objectives—not against the scorecard for an AI assistant or business intelligence initiative.
Rohan examines the wider relationship between AI and emerging technologies in The Convergence of AI and the Top 10 Emerging Technologies. The book is a route into that broader work; current descriptions of individual ventures and offerings remain on their respective pages. Readers can also visit Rohan Hall’s home page for his ventures, book and podcast to continue exploring his work.
Visit Rohan Hall’s home page to explore his ventures, book and Explainable AI Podcast.
Explore Rohan Hall’s workFrequently asked questions
Should an adviser always give one recommendation?
Usually, yes. Leadership benefits from a clear preferred course, even when the recommendation is conditional. If two options are genuinely equivalent, the adviser should say why, identify the deciding criterion and explain how leadership can resolve it.
How many scenarios should an adviser present?
Use only enough scenarios to capture meaningfully different decisions. Avoid long lists of remote possibilities. Each scenario should change the action, timing, safeguards or level of commitment; otherwise it adds complexity without improving the decision.
What should an adviser say when evidence is weak?
State that the evidence is weak, identify what is still an assumption and recommend the smallest responsible action that can produce better evidence. The adviser should also define what result would justify continuing, changing direction or stopping.
How should disagreements between advisers and executives be handled?
Make the disagreement inspectable. Identify whether it concerns facts, assumptions, risk tolerance, timing or strategic preference. Then determine what evidence could resolve it and which party has authority to make the decision if the disagreement remains.
When should leadership delay an AI decision?
Delay when a missing fact could materially reverse the choice, when the organization cannot control a serious exposure, or when no accountable owner exists. Do not delay merely because every uncertainty has not been eliminated.
What records should be kept after an uncertain decision?
Record the decision, the evidence available at the time, the key assumptions, the chosen safeguards and the triggers for review. This creates accountability and prevents later outcomes from distorting how the original decision is judged.
The bottom line
An adviser should never use uncertainty as cover for avoiding a recommendation. The right approach is to state the decision, separate facts from assumptions, define a few plausible scenarios and explain which evidence would change the course. Leaders then need an owner, an action and a review trigger. Avoid false precision, unbounded caveats and vendor-led framing. The most credible adviser is not the person who predicts AI’s future with the greatest confidence; it is the person who helps leadership make a sound decision now and revise it intelligently when conditions change.
Rohan Hall
Founder of OceSha Ventures · AI architect and author
Rohan Hall is a technology entrepreneur, AI architect and author with years of technology experience, now focused on practical AI across business, education, government and global impact. He founded OceSha Ventures, builds the OceSha AI platform and Lumi, and wrote The Convergence of AI and the Top 10 Emerging Technologies.
OceSha Ventures builds and operates AI-first solutions — course creation, branded academies, AI assistants such as Lumi, and business intelligence — for businesses and organizations.
Sources
About this page. Last reviewed .
It is based on his verified public professional record.
e6a4f84a89960d5505480465b837916a