Home/Sample assessment
A look inside the deliverable
Explore the assessment for Cedar Supply Operations, a fictional 32-person wholesale distributor. The details demonstrate the method; they are not customer results.
Fictional training example · Not client results
Cedar’s operations sponsor wants to reduce the time spent responding to routine product questions. A draft assistant using approved product sheets is worth evaluating after tool approval, access, baseline, and fallback work. An agent should not receive authority to change orders or prices.
Fictional wholesale distributor
Considered in the example
Prerequisites remain open
The example uses Microsoft 365 and an order-management system. Interviews, workflow notes, and attestations are fictional. No live systems were examined.
Domain
Score / 3
What it means
Business alignment
1.8
The baseline is estimated, not measured.
AI use & tool control
1.0
Vendor review is unverified: one unknown.
Data & access
1.6
Permissions need cleanup.
Governance & ownership
1.0
An acceptable-use policy is absent.
Risk & human oversight
1.2
Action permissions are unknown.
Operational readiness
1.4
The manual fallback has not been tested.
The overall average is 1.3 when rounded from 40/30. It does not clear the workflow gates. Vendor review and action permissions are the two unknowns; the missing policy is an observed gap.
Workflow
Directional score / 5
Decision
Draft answers from approved product sheets
3.95
Prepare first
Summarize internal operational notes
3.10
Prepare first; lower priority
Autonomously change orders and prices
3.00
Defer; transaction authority is uncontrolled
Product-question drafting has the strongest directional score. Data approval, vendor review, and tested fallback remain open. Autonomous transaction changes fail the risk and oversight gate, regardless of their potential value.
The operations supervisor would own the proposed experiment. Staff review every answer and its source before external communication. Only approved, non-sensitive product sheets are in scope. No customer records, mailbox-wide access, order updates, or autonomous sends.
Illustrative economics: an interview estimate of 120 questions at six minutes each equals 12 hours per week. A hypothetical four-minute total handling time would release four hours if volume and quality hold. This is not achieved savings, a forecast, or cash ROI; setup, training, and software costs are excluded.
First, measure actual time, volume, corrections, and exceptions. Proposed evaluation: at least 50 eligible questions across two weeks, median draft-plus-review time of four minutes or less, at least 95% factual accuracy before correction, 100% human review, and zero unauthorized disclosures or actions. The sponsor must approve these thresholds.
Any unauthorized disclosure or action stops the experiment. Repeated unsupported answers trigger suspension and review. The fallback is the existing manual response process.
When
Action and owner
Evidence to advance
Days 0–15
Sponsor and MSP inventory AI use and agree interim rules.
Named tools and written prohibited actions.
Days 0–30
Operations supervisor measures the current workflow.
Time, volume, quality, and exception log.
Days 0–30
MSP and sponsor review sources, permissions, tool handling, and reviewer responsibilities.
Completed pilot gates and processing review.
Days 0–30
Supervisor tests the manual fallback.
A successful return to the existing process.
Days 31–60
Supervisor runs the draft-only experiment if approved.
Evaluation log, incident log, and agreed thresholds.
Days 61–90
Sponsor decides to retain, revise, or stop.
A decision memo that includes total cost and effort.
If prerequisites are late, defer the experiment. Preparation is not quietly moved to after deployment.
The fictional evidence set comprises a sponsor interview, a sanitized workflow map, an approved-software list, a high-level access attestation, a supervisor interview, and a proposed evaluation charter. The charter is a proposal, not an approved control. Future roadmap records are not evidence already reviewed.
All organizations, interviews, figures, and findings on this page are illustrative. No client approval or implementation exists. A real report requires traceable evidence, permitted processing, MSP review, and a client decision owner. This assessment does not certify security, compliance, legal sufficiency, or financial outcomes.
Method:
Nerode readiness methodology
· Example prepared September 20, 2026; website adaptation September 29, 2026.
Tell us about your MSP and an AI question your clients are asking.
Discuss your first assessment ↗
Clear thinking. Practical AI readiness. Built for the MSP–client relationship.
A division of Panopticon Media House.
© 2026 Nerode Labs · Panopticon Media House