# AI workability — Eastern Migratory Caribou Photographic Survey Interpretation

*Buyer: Department of the Environment (ECCC ) · Source: nstenders · Closes: 14 Sep 2026 00:00 UTC*

**AI workability: 62%** (majority AI-deliverable) · notice-only read · bid call: **MAYBE**

- Effort: 400 h total · 247 h AI · 153 h human (low confidence)

## Verdict

AI-assisted majority workability: AI agents can perform most image interpretation and reporting, with a qualified biologist providing final QA and sign-off.

## Bid call

**MAYBE** — AI can perform the majority of the interpretation work, but the tender lacks evaluation criteria and a detailed Statement of Work. Success depends on partnering with a qualified wildlife biologist and clarifying data format, accuracy requirements, and evaluation before bidding.

## Deliverables

| Deliverable | Mode | AI share | Hours |
| --- | --- | --- | --- |
| Imagery ingestion and preprocessing | ai_agent | 90% | 60 |
| Caribou detection and classification | ai_assisted | 75% | 160 |
| Demographic analysis and reporting | ai_assisted | 65% | 100 |
| Quality assurance and final validation | human_only | 10% | 80 |

## Risks

- **[high]** AI detection accuracy on caribou imagery may be insufficient without robust validation. — *Use ensemble models, human review of all detections, and iterative QA with a wildlife biologist.*
- **[medium]** Lack of labeled training data for caribou detection. — *Use transfer learning from similar ungulate datasets and active learning with expert labels.*
- **[medium]** Thin tender documents and unknown evaluation criteria. — *Request the Statement of Work and clarify data format, accuracy requirements, and evaluation before bidding.*
- **[medium]** Tight timeline from publication to closing. — *Start with pretrained models and automated pipelines; allocate expert review time early.*
- **[medium]** Small shop lacks large reference contracts and may face credibility concerns. — *Partner with a qualified environmental consultancy or wildlife biologist to strengthen the bid.*

## Human roles

- **Wildlife biologist / remote sensing analyst** (20%): Final QA, ambiguous image interpretation, demographic conclusions, and professional sign-off.
- **Project manager / contract lead** (8%): Client communication, contract administration, and delivery management.
- **ML engineer (human oversight)** (6%): Model tuning, validation, and handling edge cases that AI cannot resolve.
- **Data analyst (human review)** (4%): Reviewing AI-generated counts and reports for consistency and completeness.

## Why not more AI

- AI cannot provide legally or scientifically binding demographic conclusions without human expert validation.
- Object detection may struggle with calves/juveniles, dense groups, vegetation, shadows, or image artifacts.
- AI only works on provided imagery and cannot fill data gaps or conduct field verification.
- Model performance depends on availability of representative training data and image quality.

## Requirements & compliance

| Ref | Requirement | Status |
| --- | --- | --- |
| Solicitation Number 5000095499 | Supplier must provide accurate and detailed interpretation of digital orthographic imagery from a June/July 2026 eastern migratory caribou photographic survey in northern Ontario, assessing total number of adult caribou, calves and juveniles, and other demographic information. | met |

## Evaluation criteria

- **Not specified in available documents** (unknown): Evaluation likely based on technical approach, personnel qualifications, and price, but no criteria or weights were provided in the notice or extracted facts.

## Key dates

- 2026-08-19T00:00:00+00:00 — Publication
- 2026-09-14T14:00:00-04:00 — Closing Date

## Gaps in source documents

- Statement of Work (Annex A) not included in available documents.
- No evaluation criteria or weighting provided.
- No details on image format, resolution, volume, or delivery method.
- No accuracy or error tolerance requirements for counts.
- No contract duration, value, or security requirements.
- No information on whether the buyer expects a specific biologist certification.

## Assumptions

- Digital orthographic imagery will be provided by the buyer; no field collection is required.
- No security clearance or on-site presence is required.
- A qualified wildlife biologist will be available for QA and sign-off.
- AI models can be trained or fine-tuned using available caribou imagery or similar ungulate datasets.
- The contract value and timeline are sufficient for a small AI-assisted team.

---

*Generated by GovContract from public tender documents. Human review required before acting on bid decisions.*