Convert an ambiguous customer request for the portfolio SaaS support copilot into measurable success criteria, technical constraints, and key risks using a discovery canvas.

Assess a set of SaaS feature requirements and justify which are suitable or unsuitable for an LLM-based solution.

Predict how context-window limits, decoding settings, and missing evidence affect output quality, latency, and reliability.

Select and justify prompt-only generation, retrieval-augmented generation, tool use, or fine-tuning for a given SaaS scenario.