Illustrative report · fictional scenario · no client results
01 / THE DECISION
Would this be worth testing?
Prepare order drafts from customer emails and PDFs for staff approval.
Illustrative recommendation
Test order entry first if the review confirms that sample files can be read and the order system can accept drafts. Use the paid pilot to measure time saved and corrections before deciding on implementation.
A real audit reviews up to three recurring tasks. It may recommend a paid test, a simpler change without AI, or stopping. It does not include building a working tool.
Entering orders
The assumed workload is frequent and repetitive. Product codes and quantities can be checked against the original request.
First candidate for a paid test, subject to data and system checks.
Adding supplier products
This would need a separate review of supplier file formats, required product facts and how often the catalog changes.
Leave outside this order-entry pilot.
Answering delivery questions
This depends on access to current order and delivery records, and your rules for replying to customers.
Review separately if staff spend more time here than entering orders.
02 / THE WORK TODAY AND THE IDEA TO TEST
Less typing. More time to resolve incomplete orders.
Assumed starting point: one shared inbox, recurring PDF formats and an existing product catalog. Staff currently copy order lines into the order system. These are example conditions to verify, not findings from a customer.
| Today: staff do the whole task | What AI would prepare | What staff still do |
|---|---|---|
| Read the email and its attachments. | Read the agreed email and PDF formats and identify order lines. | Check the draft beside the original request. |
| Look up product codes and type quantities. | Suggest matching products and fill in the quantities. | Resolve unclear matches and confirm quantities, prices and stock. |
| Find missing details and prepare the order record. | Mark missing fields, flag possible duplicates and prepare a draft for the order system. | Ask the customer for missing details and approve the completed order. |
Example numbers — assumptions, not measured savings
- Orders handled each month
- 1,000
- Assumed staff time today
- 12 min / order
- Target to test, including checking and corrections
- 4 min / order
- Staff time freed if the target is met
- 133.3 h / month
1,000 × (12 − 4) ÷ 60 = 133.3 hours. The pilot has to establish whether this reduction is realistic across ordinary and difficult orders.
Why it could matter: staff could use the time to resolve missing details and customer questions before dispatch, or handle more orders. Freed time does not automatically reduce payroll or increase sales.
03 / INFORMATION AND SYSTEMS
What must be checked before building.
Can the files be read?
Review representative emails and PDFs, including poor scans, unusual layouts and incomplete requests. Agree which formats the pilot includes.
Not checked in this fictional exampleCan products be matched?
Compare customer product references with your catalog and known alternative codes. Unclear matches must remain marked for staff review.
Product list and matching rules neededCan your software accept a draft?
Your IT contact confirms a supported import file or a direct connection, access permissions and a safe place to test. Compatibility is not assumed.
Order-system check neededWho can see the information?
Agree permitted data, staff access and retention before sharing sensitive information. Use examples with personal details removed where possible.
Data handling and access to be agreed04 / A SEPARATELY PAID PILOT
Know what the test needs to prove.
Test one inbox, the agreed file formats and the path to one order system. Agree the sample size, quality targets, fee and delivery date in writing before the pilot begins.
- 01
Agree the test cases
Select orders with the formats and difficult cases your team actually handles: missing codes, changed quantities, duplicate emails and unclear attachments. Keep a separate set for the final check.
- 02
Measure the complete task
Compare manual entry with preparing, checking and correcting an AI draft. Include orders that still need manual handling and time spent moving information into the order system.
- 03
Keep approval with staff
Your team checks product matches, quantities, prices and stock. The pilot must leave missing information visible and must not submit an order or contact a customer without the agreed approval.
- 04
Make a written decision
Record time spent, corrections, unsupported cases and estimated running costs. Continue only if the agreed quality checks and business case are met. Otherwise stop or agree a different test.
Planning schedule: 6–9 working weeks. Audit 1 week, pilot 2–3 weeks, implementation 3–5 weeks, assuming inputs, access and reviewers are available. Waiting and approvals can extend the calendar. See the assumptions and decision points.
Pilot results: not available. No test has been run for this fictional example. A real pilot report would show measured time, corrections and cases the tool could not handle.
05 / THE BUDGET TO CONSIDER
Allow for the whole project.
Indicative VistulaCode fees for the order-entry example, from audit to implementation and training. You buy each stage separately. These figures help with planning; they are not a quote for your business.
- Automation Audit
- $1,250
- AI Pilot
- $3,000–$5,000
- Implementation & Team Training
- $6,500–$10,000
- Total one-off VistulaCode fees
- $10,750–$16,250
All amounts are in USD and exclude VAT. Client staff time, AI usage, software subscriptions, hosting and ongoing support are additional. The final quote depends on the files, system access and work agreed.
Project + the first 12 months after launch
Illustrative total: $24,550–$30,050 USD net with support at its starting price, or $12,550–$18,050 USD net without support, plus your maintenance cost. Both use an assumed $150/month for AI, software and hosting; the support option adds $1,000/month. These are planning assumptions, not a quote or a maximum price. VAT and your team’s time are extra.
Compare the running-cost assumptionsWithout Monthly Support, your team or appointed provider handles ongoing checks and maintenance. See the responsibilities after handover.
An illustration of value, not a promise of cash savings
If 133.3 hours were freed at an assumed staff cost of $19 USD per hour, their estimated value would be $2,533 USD per month. Subtract an illustrative total running-cost allowance of $1,150 USD and the remaining value would be $1,383 USD per month.
The running-cost allowance is a calculator assumption, not a monthly quote. Include any support you choose in that total; do not count it twice. Replace it with the full costs of your proposed setup. None of these figures are measured client results, and this calculation does not show cash payback.
Try different numbers in the calculator06 / WHAT THE CLIENT DOES NEXT
Agree how to judge the value before you invest.
The manager confirms the task and budget, an experienced team member checks sample orders and the IT contact verifies access to the order system. Staff time is a separate business cost. We agree the work required from your team before each stage.
You can stop after the audit. The pilot starts once you approve its scope and price; implementation is a later, separate decision.