Practical guide · AI opportunity assessment guide and worksheet
Use this practical framework and private browser-based worksheet to compare potential projects by evidence, business value, feasibility, and risk before software or AI models shape the decision.
The starting principle
Effective automation starts with understanding how a company creates value, where information moves, why delays occur, and which decisions require human judgement. Unfoldry combines business consulting, process research, data analysis, and technical implementation to turn that understanding into a realistic recommendation.
A template workflow can be useful as a demonstration, but it starts with a preselected tool and process. Companies rarely share the same data, exceptions, responsibilities, approval paths, or success criteria. Time saved by installing a basic template can quickly be lost while reshaping it around the real operation.
The objective is not to find somewhere to use AI. It is to find the most valuable problem that technology can solve reliably.
Unfoldry method
The free review uses public research and focused questions. Paid scoping adds the operational detail, system access, measurements, and decisions required before implementation.
Review the business model, services, customers, operating context, and existing technology before suggesting a project.
Evidence: company information, service journey, roles, and systemsDocument inputs, outputs, decisions, handovers, exceptions, and the people responsible for each stage.
Evidence: process walkthroughs, examples, and exception casesEstimate volume, cycle time, manual effort, delays, errors, rework, and quality before claiming potential value.
Evidence: operational data and representative samplesCompare process improvement, deterministic automation, AI-assisted work, and no-change options instead of assuming AI is required.
Evidence: multiple solution paths and constraintsEvaluate each opportunity using the same criteria, then challenge uncertain assumptions and high-risk dependencies.
Evidence: weighted score, confidence level, and risk notesSelect the smallest useful project that can produce a measurable result and support a sensible next step.
Evidence: recommendation, boundaries, and success criteriaDecision matrix
Rate each factor from 1 to 5. A score of 5 is favourable, including the risk category where 5 means low and controllable risk. Multiply the rating by the weight to compare opportunities consistently.
| Criterion | Weight | What to examine | A useful question |
|---|---|---|---|
| Business value | 25% | Time, cost, quality, capacity, customer value, or revenue | What measurable result would make this worthwhile? |
| Process volume | 10% | Frequency, repetition, and number of people affected | Does this happen often enough to matter? |
| Data readiness | 15% | Availability, quality, permission, structure, and ownership | Can the system access reliable inputs? |
| Technical feasibility | 15% | Integrations, system limitations, exceptions, and maintainability | Can this work reliably in the existing environment? |
| Operational fit | 15% | Ownership, workflow changes, adoption, and support | Will the people involved actually be able to use it? |
| Human oversight | 10% | Review points, escalation, accountability, and reversibility | Where must a person remain in control? |
| Risk | 10% | Privacy, security, compliance, error impact, and dependency | What happens when the system is wrong or unavailable? |
A score is a comparison aid, not a business case. Weak evidence should lower confidence even when the calculated score looks attractive.
Free assessment worksheet
Name the workflow, define the baseline and success measure, then rate all seven factors. The weighted score is calculated in your browser. Nothing you enter is sent to Unfoldry or saved.
Suitability check
Illustrative assessment
The firm prepares 40 client reports per month. Each report takes about 75 minutes to assemble from structured assessment answers, standard recommendations, and reviewer notes. Follow-up activity is recorded inconsistently, while the proposed website chatbot has no defined knowledge owner or support objective.
High volume, structured inputs, a measurable baseline, and a clear reviewer approval step.
Useful and feasible, but ownership and trigger rules need agreement before implementation.
The intended outcome, source quality, escalation path, and ongoing content ownership are not defined.
Start with reporting. Scope one report type, preserve reviewer approval, test difficult input cases, and measure total preparation time and correction rate against the current baseline.
This example is fictional and illustrates the method. Scores and outcomes are not client results or performance promises.
Common mistakes
The available feature becomes the project, even when it does not address the most important operational problem.
Automation makes unclear steps move faster without resolving unnecessary work, unclear ownership, or conflicting rules.
Predictable validation and routing often need deterministic logic, not a probabilistic model.
A clean demonstration hides the unusual cases, approvals, and recovery paths that determine production reliability.
Without current volume, time, errors, and review effort, impact remains an assumption.
A production system also needs security, ownership, monitoring, documentation, maintenance, and change control.
The next decision
During scoping, we measure the current process, define the desired result and major risks, then provide a fixed project price or objective milestones. Implementation fees, production software and API costs, and optional maintenance are shown separately.
Scoping tests whether the attractive opportunity still holds when real systems, data, exceptions, responsibilities, security, and acceptance criteria are made explicit. If the evidence does not support implementation, the recommendation should change.
Request a company-specific AI opportunity assessmentFAQ
Yes. It is designed to improve internal discussion. A company-specific assessment adds research, focused questions, evidence checks, and an independent recommendation.
No. Confidence, dependencies, sequencing, and strategic timing can outweigh a numerical difference.
No. A conventional automation, process redesign, or decision not to build can be the better outcome.
Send your website and email. We send personalized questions within three working days and prepare the opportunity review after your answers.