Topics:

• Workflow patterns: linear, branching, looping, human-in-the-loop
• Designing decision trees for where agents act vs where humans approve
• n8n deep dive: nodes, expressions, error handling, retries
• Prompt engineering for agents (not chat): system vs task prompts, role + goal + constraints + output format
• Memory & retrieval: short-term, persistent, vector stores (no code)
• Knowledge base construction in Lindy and Relevance AI
• Source citation & grounding — making agents say "I don't know"
• Why grounded agents beat generative agents for any decision-grade output

Hands-on:

• Build-along #2 — Branching Decision Agent (lead qualifier pattern)
• Build-along #3 — Your Research Agent (customer feedback / competitor watch / vendor research / talent market — learner picks)

Topics:

• The universal intake pattern: capture → classify → enrich → score → route
• Zapier deep dive + Zapier Agents: when Zapier wins over n8n / Lindy
• Webhooks, APIs, MCP servers explained without code
• Enrichment patterns: pulling context from LinkedIn, Clearbit, web, internal KBs
• Scoring rubrics an agent can apply consistently
• Routing logic: round-robin, owner-based, priority-based, SLA-based
• When you'll hit walls and need a developer — how to scope that ask

Hands-on:

• Build-along #4 — Intake & Triage Agent (inbound leads / feature requests / expenses / tickets / resumes — learner picks)
• Build-along #5 — Score & Route Agent extending Build #4 with enrichment + scoring + smart routing

Topics:

• Bidirectional sync patterns: push, pull, mirror, merge
• Conflict resolution: who wins when systems disagree
• Tool spotlight: Lindy + Zapier + n8n for cross-system sync
• The transcript-to-tracker pattern that kills 30% of admin work
• Tool spotlight: Fireflies / Otter / Granola + Jira / Asana / Notion / HubSpot
• Auto-update from voice, email, chat, and forms

Hands-on:

• Build-along #6 — Bidirectional Sync Agent (Jira ↔ Notion / HubSpot ↔ Sheets / Books ↔ ERP / ATS ↔ HRIS)
• Build-along #7 — Meeting-to-Tracker Agent built on a real meeting from the learner's week

Topics:

• The watch → detect → nudge → escalate pattern
• Designing escalation ladders that don't annoy people
• State tracking, deduplication, quiet hours, snooze logic, manual override
• Multi-agent patterns: manager-worker, sequential pipeline, debate/critique, parallel research
• Why splitting agents beats one mega-agent
• Relevance AI deep dive: building an "AI workforce" of role-based agents
• Conversational agent layer: Voiceflow for internal copilots and helpdesk bots

Hands-on:

• Build-along #8 — Your Watchdog Agent (stale tickets / silent deals / late invoices / pending approvals / SLA breaches — learner picks)
• Build-along #9 — Multi-Agent Coordination: connect Research + Intake + Sync + Watchdog into one orchestrated flow — first view of the full capstone stack

Topics:

• The reporting layer: composing narrative weekly briefs from multiple agent outputs
• Reliability engineering for non-engineers: failure modes, eval mindset, shadow-mode deployment, alert-on-failure
• Cost & token economics: monthly cost math, cheap vs frontier model selection, total cost of ownership
• IT, security, compliance for the non-technical: data residency, PII, SOC 2, DPA reading, avoiding the shadow-IT trap
• Selling AI internally: 1-page business case, ROI framing, scoping a safe pilot, manager pitch mechanics
• Portfolio, positioning & career: LinkedIn rewrite, resume bullets, demo video, Notion portfolio template
• Three post-program paths: in-company AI champion · freelance/consulting · founding

Hands-on:

• Build-along #10 — Weekly Brief Agent that pulls from all prior agents and composes one narrative report for stakeholders
• Stress-test workshop — pair up, break each other's agents, document failure modes, fix