Multimodal API Evaluation and Prompt Chaining Integration Checklist

Role: Lead AI Solutions Engineer specializing in multimodal pipeline integrations, diffusion model APIs, and automated visual workflows.

Context

• Target Account: {{prospect_company}}
• Evaluated Model Stack: {{evaluated_model_stack}}
• Key Technical Contacts: {{technical_stakeholders}}
• Performance Bottlenecks: {{benchmark_pain_points}}
• Compute and Cost Constraints: {{latency_cost_thresholds}}
• POC Target Date: {{trial_milestone_deadline}}

Task

Create a comprehensive technical follow-up checklist to guide engineering communication with {{prospect_company}}, ensuring rapid resolution of integration blockers, validation of prompt chaining benchmarks, and adherence to {{trial_milestone_deadline}}.

Method

1. Review the performance benchmarks discussed during the demo of {{evaluated_model_stack}} against {{benchmark_pain_points}}.
2. Identify specific API parameter tuning recommendations (inference steps, sampler configurations, CFG scale) to address {{latency_cost_thresholds}}.
3. Outline diagnostic validation steps for prompt chaining workflows, including multimodal input embedding caching and token utilization.
4. Draft specific follow-up inquiry points tailored for {{technical_stakeholders}} regarding environment provisioning and SDK integration.
5. Define measurable acceptance criteria for the proof-of-concept deliverables.
6. Construct a structured technical follow-up checklist verifying all prerequisite logs, sample generation seeds, and latency metrics.
7. Assemble the complete QA and follow-up protocol into clear operational phases.

Constraints

• MUST include specific technical verification items for API response times, cold-start latency, and seed reproducibility.
• MUST format all outputs as itemized checklist entries with clear pass/fail or action criteria.
• MUST NOT propose changes that violate the constraints specified in {{latency_cost_thresholds}}.
• Avoid high-level sales language in favor of developer-centric workflow validation.

Output format

• Phase 1: Technical Diagnostic & Context Checklist (4-5 checklist items)
• Phase 2: Follow-Up Email Architecture Checklist (5-6 technical verification items)
• Phase 3: POC Milestone Gating Checklist (3-4 validation checkpoints)

Self-review

• Confirm every checklist item includes explicit verification criteria.
• Check that technical parameters relevant to {{evaluated_model_stack}} are systematically covered.
• Verify that all variables are properly referenced and contextualized.