James Parham - Brand and Land: Agentic AI

Presentation Title

James Parham - Brand and Land: Agentic AI

Presentation Speaker

James Parham

Presentation Abstract

Most project managers meet AI as a chat box. That confuses two different things: the large language model (LLM), which generates language, and retrieval-augmented generation (RAG), which pulls your own documents into that generation so answers are grounded in your project, not in whatever the model remembers. You do not need to build or train an LLM. You need governance first, then an agent that retrieves from knowledge you control.

In this one-hour session, James Parham walks an at-risk portfolio-and-below scenario (Northbridge Care Group / Project Harbor) and shows how to use PMI Inputs, Tools and Techniques, and Outputs (ITTOs) and the Exam Content Outline as reference models to define recovery governance before any AI runs. The room sees why a project can look green on throughput while Demand-to-Staff cannot name a staffable gate. He uses Tonic Fabricate to generate synthetic cost-of-nonconformance and delay data so the cost of dual capacity truth, open priority decisions, and an unstaffed release gate is visible in numbers, not only in narrative. Reference models give the structure; the synthetic ledger makes the price of waiting concrete; the team selects and adapts what applies. He then shows his own agentic system, built around his own RAG over his standards and project materials, generating a recovery plan and information radiators (including teaching freehands and an A3 problem card) that stakeholders can read at a glance. The demo is not Claude or ChatGPT as the product. The point is owning the agent and the retrieval layer while using known good models under the hood. Participants leave able to separate LLM from RAG, use PMI reference models plus quantified nonconformance and delay signals to shape governance, and sketch a governed starter pattern for their next at-risk initiative.

Learning objectives:

By the end of this session, participants will be able to:

1. Use ITTOs and the Exam Content Outline as reference models to define governance for an at-risk portfolio-and-below scenario before any AI generation starts.
2. Distinguish an LLM (the model that generates language) from RAG (retrieval of your own knowledge that grounds that generation), and state why owning RAG matters more than picking a chat brand.
3. Explain why using known good models is enough for project work, and why building or training an LLM is not required for this pattern.
4. Read synthetic cost-of-nonconformance and delay data (Fabricate teaching set) as evidence for freeze-or-continue decisions, and connect that evidence to an A3 / recovery radiator before drafting with an owned RAG agent.
5. Describe how a project-scoped agentic system (own RAG, skills, human review) can produce a recovery plan and information radiators, and draft a one-page starter pattern for their own project.

Speaker Bio

James Parham is a project & portfolio practitioner, I/O psychology graduate (M.S., Arizona State University), and retired U.S. Army Chief Warrant Officer Four (915E). He holds nine active PMI credentials: PfMP, PgMP, PMP, PMI-PMOCP, PMO-CP, PMI-SP, PMI-PBA, PMI-ACP, and PMI-RMP. He is a Certified Business Architect (CBA) Level 1 with the Business Architecture Guild and holds The Open Group TOGAF credentials (TOGAF 10 Foundation; TOGAF Enterprise Architecture Foundation; TOGAF Business Architecture Foundation). He leads Army Strong PM Consultants LLC. He instructs & supports learners in St. Michael’s Learning Academy’s ServiceNow program across career-transition tooling, architecture literacy, and platform training, and he speaks on the Brand & Land TAP team with Signofied, LLC. His current practice includes building project-scoped agentic systems that use known good language models with his own retrieval (RAG) to produce governed plans &teaching freehands used as information radiators.