Leveraging AI 278 | How To Build AI Agents: A Step-by-Step Playbook for Business Leaders with Jim Spignardo
22 snips
Mar 24, 2026 Jim Spignardo, Director of Cloud Strategy & AI Enablement at ProArch who builds practical AI agents for sales and operations. He walks through building RFP-focused AI agents, organizing and cleaning data, writing instruction-first workflows, controlling sources to avoid hallucinations, and deploying agents across apps to cut proposal effort and speed decision-making.
AI Snips
Chapters
Transcript
Episode notes
Start With Use Cases Before Granting Licenses
- Start AI rollouts by identifying high-value, low-effort use cases and then assign licenses only to those personas.
- Jim used this approach at ProArch, rolling out Copilot licenses after mapping use cases like RFP responses to justify cost and adoption.
RFP Responses Are High-ROI Automation Targets
- RFPs are an ideal automation target because they both consume senior time and directly drive revenue.
- Isar and Jim call RFP response a perfect bottleneck: heavy effort, low win rate, high ROI if improved.
Prototype The Workflow In Chat Then Auto‑Generate Instructions
- Build agent instructions by first doing the task in a chat, iterating until good, then ask the chat to write the agent instructions.
- Isar and Jim both use regular chat (OpenAI/Copilot) to prototype the workflow and then convert the final process into the agent's instruction set.