Episode 56 · Season 1
AI Agent Strategy: 7 Deliverables to Lock In Before Development
Most AI agent initiatives do not fail in the build. They fail in the planning, where six teams each carry a ninety percent plan and nobody agrees on what success looks like. In this episode of Cirrius Talk: Agentic AI, Tim Harting and Gavin Franklin of Cirrius Solutions walk through the seven deliverables every organization should lock in before building a single agent.
What you’ll learn:
- The three pre-execution phases: Assess, Initial Architecture, and Initializing Execution
- How to build a use case value model with KPIs an agent can actually move
- Running a platform fit assessment so you do not end up paying for agents across six platforms
- Why target state architecture for AI leans on data and integration, not system design
- Building an agent evaluation framework that covers output quality, token cost, and security, not just KPIs
Who it’s for:
CTOs, CIOs, engineering leaders, and business leaders who are responsible for turning “go build AI” into a plan that actually delivers value.
Guest + credibility:
Tim Harting and Gavin Franklin lead AI implementation work at Cirrius Solutions. The seven-deliverable framework in this episode is what they have coalesced on after iterating across multiple agent engagements and rollouts.
Primary CTA:
Planning an agent rollout around Anthropic, OpenAI, or Salesforce? Learn more at https://cirriussolutions.com/podcast/ (Optional: Follow the show for more Agentic AI episodes.)
Recommended Links & Resources
Any Questions: Reach out to us at CirriusTalk@cirriussolutions.com
We cover
- Synthesizing Key Elements for AI Success
- Establishing a Use Case Value Model
- Conducting a Platform Fit Assessment
- Developing Initial Architecture for AI Projects
- Creating an Agent Evaluation Framework
- Chartering Initial AI Pilots
Chapters
- 0:00Introduction to AI Implementation Challenges
- 3:24Synthesizing Key Elements for AI Success
- 6:05Establishing a Use Case Value Model
- 11:15Conducting a Platform Fit Assessment
- 15:03Developing Initial Architecture for AI Projects
- 18:55Creating an Agent Evaluation Framework
- 22:11Chartering Initial AI Pilots
- 26:01Executive Proposal and Engagement Roadmap
Resources and downloads
The artifacts Greg’s team built for this episode, free to download.
AI Agent Strategy: 7 Deliverables to Lock In Before Development
Lays out a three-phase, seven-deliverable path for taking an AI initiative from idea to funded pilot: Assess (use case value model, platform fit scorecard), Architect (target-state architecture, agent evaluation framework), and Execute (pilot charter, executive proposal, engagement roadmap). Each phase ends at a gate before the next begins.
Related episodes
Agentic AIAI Dev Tool Security: 4 Risks to Act On, 4 to Ignore
Agentic AIAgentic OS Explained: 4 Pillars That Keep Your AI Agent Sharp
Agentic AIClaude Code vs Wireframes: Building Interactive POCs in Hours, Not Weeks
The Military MinutePat Hubbell (Candorful) – Why 94% of Vets Who Practice Land a Job Within 6 MonthsPat Hubbell, Candorful
Salesforce at ScaleTaylor Stewart: Is Custom Code Still Scary? AI’s Impact on Salesforce DevelopmentTaylor StewartTalk to us about AI agents
Keep going: See how we build AI agents for real businesses, Talk to us about a custom AI agent, More from the Cirrius blog