COMPLETE SAMPLE / FICTIONAL BUSINESS
A clear first step.
A plan to build on.
Example: Bayou Project Services, a fictional small project-services office. Every business detail, estimate and outcome below is illustrative. This is not a client case study.
1. The work today
The owner and an office coordinator manage quotes, project updates and customer messages using email, spreadsheets and shared files. They report spending about 12 hours a week on the five tasks below. Their goal is less chasing and drafting, with accurate customer commitments. They can use approved material in an employer-approved AI workspace, but cannot connect the inbox yet.
What has not been verified: Task times are self-reported. File quality, software features, permissions, security requirements and integration costs need checking. This roadmap is a planning tool, not a systems audit.
2. Opportunities, in order
| Priority | AI could help with | Why this order | What it needs / human check |
|---|---|---|---|
| 1. Quote follow-ups | Prepare a follow-up list and draft messages from approved open quotes. | Frequent work with a clear source and a simple manual pilot. No app connection needed. | Current quote register and contact rules. A person checks price, status, dates and every message. |
| 2. Weekly updates | Turn approved task notes into completed work, open work, blockers and decisions. | A stable report format makes comparisons easy after the first pilot. | Current task notes and report template. A manager checks status and missing updates. |
| 3. Meeting follow-ups | Extract decisions, owners, dates and open questions; draft a recap. | Useful with little setup, but notes often omit owners or confirmed dates. | Approved notes. The meeting owner confirms assignments; AI flags missing facts. |
| 4. Customer replies | Draft common replies using approved policies and current project facts. | Useful after reliable source material and writing rules are collected. | Approved examples and current policies. A person checks promises and sends. |
| 5. Internal questions | Find and cite answers in a small approved reference set. | Wait until documents are current and ownership is clear; stale sources create risk. | Named document owners, version dates and access rules. Unknown answers go to a person. |
Ranking rule: Favor repeated work with available sources, a checkable result and low access requirements. Defer ideas with unclear records, permissions or accountability.
3. Estimates to test
| Task | Reported baseline / week | Illustrative pilot target, including review | Potential capacity returned |
|---|---|---|---|
| Quote follow-ups | 5 hours | 3–4 hours | 1–2 hours |
| Weekly updates | 3 hours | 2–2.5 hours | 0.5–1 hour |
| Meeting follow-ups | 1.5 hours | 1–1.25 hours | 0.25–0.5 hour |
| Customer replies | 1.5 hours | 1–1.25 hours | 0.25–0.5 hour |
| Internal questions | 1 hour | Not estimated until sources are checked | Not estimated |
These are hypotheses for this fictional example, not typical or promised results. Do not add them together before checking overlap and actual adoption. At an assumed $40/hour, the first pilot’s 1–2 hours would represent $40–$80 of weekly staff capacity, not cash savings or reduced payroll.
Effort and cost: Allow a planning estimate of 4–8 staff hours to clean the quote register, write rules, test drafts and record results. This excludes custom development. Use an approved existing AI subscription if its features and data rules fit; verify any additional license or integration cost before committing. The optional TMI hands-on setup is $7,500 with scope agreed separately; buying both offers totals $8,497.
Tool and cost guide
Existing tools first: Keep the approved quote spreadsheet as the source of truth. Pilot in an existing employer-approved AI workspace if the account supports the task and permits this data. For this fictional sample, assume the current license is already paid for and suitable; the incremental monthly license cost would be $0. Verify that assumption with the account owner.
If a new license is needed: This fictional budget sets aside $30 per user per month for evaluation. That is an illustrative planning allowance, not a provider quote. Confirm the actual plan price, taxes, file limits, data policy and access controls before buying. Custom integrations, API usage and implementation fees are not included in that allowance.
Setup: The 4–8 staff-hour planning range above covers a manual pilot, not an installed integration. Keep outbound messages under human approval. Do not buy another tool until the existing setup has been tested and the missing capability is clear.
4. First-project brief: open-quote follow-ups
Goal: Reduce the full time needed to prepare accurate follow-ups, without sending incorrect or duplicate messages.
Owner: Office coordinator prepares; owner approves customer commitments and sends.
In scope: A manual pilot on ten approved, non-sensitive open quotes. AI produces a follow-up list and unsent drafts. No inbox connection, automatic sending, repricing or CRM changes.
Inputs: Quote reference, customer alias, current status, confirmed amount, last-contact date, next action and do-not-contact flag. Remove unnecessary personal details.
- Record current preparation, checking and editing time for a comparable batch.
- Confirm the quote register is current and allowed for use in the chosen tool.
- Ask AI to identify due follow-ups and flag missing facts before drafting.
- Compare every draft with the register. Correct errors and save the rule that prevents each error.
- Repeat on a second batch. Record total time and errors before deciding whether to continue.
Pass criteria: Every price, date and commitment is traceable to the source; no restricted contact receives a draft; missing facts are flagged; the owner approves the result; total time including review improves against the baseline. If those checks fail, fix the source or instructions and retest. Keep the existing manual process until the pilot passes.
Stop rules: Missing permission, outdated quote status, conflicting prices, unclear recipients or uncertain commitments stop the affected draft. Ask the owner; do not infer an answer.
5. The next 90 days
| Planning window | Action | Decision / owner |
|---|---|---|
| Days 1–14 | Confirm access rules, clean the register, baseline the current task and test two small batches. | Coordinator records results; owner decides whether the pilot passes. |
| Days 15–30 | Refine the approved workflow, document checks and train the backup person. | Keep manual use only if quality and full task time improve. |
| Days 31–60 | Test weekly updates using the same source-and-review method. Check whether any app connection is worth the added complexity. | Owner and IT approve access and costs before connecting anything. |
| Days 61–90 | Compare results, retire ineffective routines and decide whether to expand meeting or customer-reply support. | Assign a maintainer for sources, permissions and workflow changes. |
The 90 days are a suggested planning horizon, not a delivery deadline or guaranteed implementation schedule.
6. Your choice after the roadmap
Build the first project with your own team, use another provider, or ask TMI to put the agreed starting point in place. Your roadmap is usable on its own.
Your AI Right Hand is TMI’s separate $7,500 hands-on implementation: we learn exactly how you work, build the agreed agents and executive assistants around that work, test them and show you how to use them. TMI arranges implementation scheduling directly with you. There is no setup sales-call booking step.