Collate Data Assets Model Requirements and Technical Constraints
Audit available proprietary datasets, evaluate third-party foundation models, and detail compute, latency, and cost constraints. Gather all regulatory, compliance, and privacy requirements relevant to the data processing pipeline, including UK GDPR considerations. Document existing technical infrastructure capabilities and internal team skills to ensure feasibility.
Completing this input audit grounds the AI architecture in realistic data availability, cost structures, and legal parameters. It ensures the proposed model strategy is commercially sustainable and legally compliant before building commences.
A detailed inventory of training and validation data, proprietary IP, model selections, and a comprehensive cost model per API call or inference run. The founder must also provide a compliance assessment covering data rights, retention, and processing agreements.
Five questions an expert would ask when reviewing your output
Use these to challenge assumptions, pressure-test your logic, and check the quality of this action's output in the context of the parent task and wider venture development.
- 1
Do you possess full legal rights and explicit user consent to utilise these data assets for model training or context injection?
- 2
What are your projected unit economics per inference run at ten times your current user target?
- 3
How will your choice between fine-tuning, retrieval-augmented generation, or prompt engineering scale as data volume grows?
- 4
What specific hardware or API constraints restrict your choice of foundation models?
- 5
How do you plan to source high-quality ground-truth evaluation data if proprietary datasets are limited?