The outcome
Seven figures a year — invisible, because stopped waste doesn't appear on any dashboard.
The money you save is the money nobody can point to. Measured on a real enterprise data-platform program — figures rounded, not exact, by agreement.
13 sessions · 22.5 hours · 1.5 CSTU semester credit hours
Seven figures saved. The metric everyone tracked had nothing to do with it.
The business bottom line doesn't care how many tokens you burned, how many messages went out, or how many people replied. It cares about one thing: did the number move — and how fast.
The evidence
The number nobody could see moving — until it did.
The speed
What used to move in months now moves in days. Sometimes hours.
The gap wasn't willingness. It was the channel — and how fast a decision could travel through it.
The leading indicator underneath it
Before — email notification
10%
response rate. Buried in an inbox, no urgency signal, easy to defer — and defer becomes never.
After — automated Slack outreach
70%
engagement within 24 hours. Direct, personalized, and specific enough that a stakeholder can act on it in under 30 seconds.
How this is built
One workflow. Two layers.
You're not learning a tool. You're building a workflow that remembers your people and reaches them without you in the loop. The tool is where you build. The workflow is what you own.
Tier 1 · Build layer
Where you construct the workflow
An agentic coding tool — Claude Code, OpenAI Codex, or OpenCode. Terminal-native, roughly interchangeable for this job. You build hands-on in Claude Code; what you build transfers to the others.
Tier 2 · Operating layer
Where it lives after this course
An agent operating layer such as Hermes Agent (Nous Research): persistent memory of each stakeholder, cross-session context, self-scheduled follow-ups, running in Slack without depending on your laptop. You don't build in Hermes during this course — it's where the loop you build graduates to next.
What participants ship
One killer skill, built on your own program.
A working outreach agent
An agent-driven Slack workflow that replaces batch email notifications with personalized, direct messages to individual stakeholders. Built hands-on in Claude Code.
A measured response-rate lift
Your own before/after numbers — not a vendor's case study. Track the gap between email and direct outreach on your actual program.
A reusable Skill
The workflow packaged so your team can run it after the course ends — not a one-off automation only you understand.
Curriculum
Thirteen sessions. One workflow that outlives the course.
22.5 hands-on hours over 6.5 weeks, twice a week — weekday evenings run 2 hours, Saturdays run 1.5 hours. Starts from zero: environment setup is folded into Session 1, so no separate prerequisite course is needed. Built directly on your own stakeholder list, not a generic demo dataset.
Set up your agent, then diagnose the response gap
Weekday evening · 2h
Environment setupStakeholder auditRoot-cause diagnosis
You leave with: Your agentic coding tool installed and working, plus a documented baseline of your own communication failure, in numbers
Tools · Environment setup (Claude Code, shell basics, git) followed by a review of your current notification history with your agent
Set up the direct channel
Saturday · 1.5h
Direct channelMessage design
You leave with: A working Slack integration ready to send messages
Tools · Slack API and webhook basics, workspace setup
Build the outreach agent
Weekday evening · 2h
Agent workflowPersonalization at scale
You leave with: An agent that sends one round of personalized outreach
Tools · Your agent, scaffolded to read a list and draft a message per recipient
Wire it to real program data
Saturday · 1.5h
Data-grounded outreachSpecific, not generic
You leave with: Messages that reference live status and blockers, not generic templates
Tools · Connecting your agent to the data source that already tracks your program
Handle responses
Weekday evening · 2h
Response handlingTwo-way loop
You leave with: A two-way loop — the agent sends, captures, and logs the response
Tools · Structured decision capture: approve / block / escalate. This is the first seed of persistent memory — the operating layer (Hermes) is where it fully lives, later.
Design escalation and cadence
Saturday · 1.5h
Escalation logicCadence design
You leave with: A cadence and escalation policy your agent enforces automatically
Tools · Follow-up timing logic that stays urgent without becoming noise — the self-scheduling behavior an operating layer like Hermes runs continuously once this graduates past a script
Segment your audience
Weekday evening · 2h
SegmentationAudience-specific messaging
You leave with: Segmented messaging logic — executives, ICs, and partners get different shapes
Tools · Audience-type branching inside your agent
Measure what the P&L sees
Saturday · 1.5h
Response-rate measurementTime-to-decision
You leave with: A dashboard showing your own before/after numbers — response rate and time-to-decision
Tools · Response-rate and time-to-decision tracking, built by hand
Pressure-test with real results
Weekday evening · 2h
Live results reviewReal-world validation
You leave with: Instructor and peer feedback on your live, measured numbers — not just your build
Tools · A working demo of your agent against your own real data
Package it as a reusable Skill
Saturday · 1.5h
Reusable SkillHandoff-ready
You leave with: A reusable Skill your team can run without you in the room
Tools · `skill-creator` — package the workflow so it outlives this course. This is what you own; an operating layer like Hermes is where it can later run without your laptop.
Scale across programs
Weekday evening · 2h
Multi-program scaleData privacy
You leave with: A scaling plan for using this beyond your original pilot
Tools · Permissions, rate limits, and privacy handling for outreach at scale — the same concerns that govern a persistent operating layer once you graduate to one
Build the business case
Saturday · 1.5h
ROI argumentBusiness case
You leave with: A one-page business case with your own numbers
Tools · Translating the response-rate lift into dollars saved and time-to-decision
Ship the case study
Saturday · 1.5h
Capstone demoPortfolio-ready
You leave with: A live demo and a portfolio-ready before/after case study
Tools · The full arc — diagnosis, build, measured lift, and business case — presented end to end
After Session 12
You don't have to present the business case alone.
Session 12 ends with a one-page business case in your own numbers. Want Ping to present it with you — or come into your organization as your Forward Deployed Engineer and implement it?
Who this is for
Anyone accountable for a number they can't move by themselves.
You're measured on an outcome, but you can't command it directly — you have to change other people's behavior at scale to move it. The mechanism isn't specific to program management — it's specific to broadcast versus direct:
- Program & Technical Program Managers — chasing decisions from stakeholders one calendar invite at a time
- Transformation & Change Leads — whose rollout succeeds or fails on whether people actually change what they do, not on whether a plan exists
- Chiefs of Staff — who own the follow-through on decisions they didn't make and can't force, only chase
You already know who you need to reach and what you need from them. What's missing isn't access — it's a channel that gets a response. This course assumes you can describe your audience in plain language and have (or can get) a workspace on the platform they already live in.
Get started
Everything you need, in one place.
Enroll now
Register for the Builder With Agentic AI Cohort by 8/31/2026.
Open link →Ask about the course
Questions about fit, timing, or your specific stakeholder setup.
Open link →Prefer to go slower first?
Optional: the standalone $980 Foundation Lab, if you'd rather ease into the toolchain before this course's pace. Not required — this course is fully self-contained.
Open link →