[Project name]
Prepared for [Client company]: a data operations plan covering scope, quality, timeline, and investment.
01Executive summary
[Client company] is building [short description of the AI system]. To reach production quality, the team needs [volume] of accurately [labeled / collected / evaluated] [data type] delivered by [deadline].
AnnoTara AI proposes a managed engagement that starts with a calibrated pilot, then scales to full production under our 5Q™ quality framework, with a target accuracy of [98%] and transparent weekly reporting.
02Objectives
- Deliver [N] production-ready items that follow the client's guidelines.
- Reach and hold a measured accuracy of [98%] or higher on audited samples.
- Surface ambiguous cases and edge cases early, and feed them back into the guidelines.
- Keep data secure and access limited to trained, NDA-bound team members.
03Scope of work
| Workstream | Details |
|---|---|
| Service | [Data annotation, e.g. bounding boxes + attributes] |
| Data type | [Image / video / text / audio / multimodal] |
| Volume | [Pilot: 500 items · Production: 20,000 items] |
| Label taxonomy | [Classes and attributes from the client guidelines] |
| Languages | [English, …] |
| Tooling | [Client platform or AnnoTara-managed tool] |
| Deliverables | [Format: COCO JSON / CSV / JSONL, plus QA reports] |
Out of scope unless agreed in writing: model training, raw data acquisition not listed above, and changes to the taxonomy after production starts.
04Approach
Guideline review
We study the guidelines you shared, list open questions and edge cases, and agree on a final version with your team before any labeling begins.
Team qualification
Annotators are trained on your guidelines and must pass a qualification test against gold-standard items before they touch production data.
Pilot and calibration
A pilot batch of [500] items is labeled, reviewed, and scored jointly with your team. Findings are folded into the guidelines and the QA plan.
Production
Work is delivered in weekly batches with multi-level review, ongoing calibration, and a shared issue log for new edge cases.
05Quality plan: AnnoTara AI 5Q™
Review process: [two-stage review + 10% expert audit]. Batches that miss the target are reworked at no extra cost.
06Timeline
- Kickoff & guideline review[Week 1]
- Team training & qualification[Week 1–2]
- Pilot batch & calibration[Week 2–3]
- Production batches[Week 4–10]
- Final QA & handover[Week 11]
07Investment
| Item | Qty | Unit price | Total |
|---|---|---|---|
| Setup & guideline review | 1 | [$0] | [$0] |
| Pilot batch | [500] | [$0.00] | [$0] |
| Production annotation | [20,000] | [$0.00] | [$0] |
| QA & reporting | — | Included | Included |
| Estimated total | [$0] | ||
Payment terms: [50% on pilot approval, 50% on final delivery, net 15]. Prices in [USD], excluding applicable taxes.
08What we need from you
- Final annotation guidelines and 20–50 gold-standard examples.
- Access to data and tooling, or approval to use AnnoTara-managed tools.
- A named point of contact for questions and pilot feedback within [2 business days].
09Next steps
Reply to confirm the scope or request changes, sign below, and we'll schedule the kickoff. Questions? Write to ai.annotara@annotara.app.