Build an Agentforce-orchestrated multi-step Journey Builder flow with personalization grounded in Data Cloud. Includes the segmentation eval and the lift-measurement harness.
Copy-paste patterns for the agents we ship in production — ServiceNow Now Assist, Salesforce Agentforce, Workday Illuminate, LangGraph, MCP. Code, prompts, evals, gotchas. No theory.
Each recipe is a complete, runnable repo: source code, prompts, eval suite, deployment guide, and a "what we tried that didn't work" section.
Classify incoming incidents by category, urgency, and routing assignment using Now Assist + AI Agent Studio. Includes a 200-row golden-set eval and the cost-per-classification benchmark.
Build an Agentforce topic that handles a 4-step service workflow against a Service Cloud org, grounded in Data Cloud. Atlas reasoning, deflection metrics, and the "when not to deflect" playbook.
Field-tested pattern for handling leave/payroll exceptions through Illuminate. The conversation-design rules, the approval escalation paths, and the audit trail your compliance team will actually accept.
Reference architecture for a multi-agent system using LangGraph orchestration over MCP-exposed tools. Includes the eval harness, the cost-tracking middleware, and the LangSmith integration.
The eval suite we ship with every agent: golden-set accuracy, drift detection, cost ceiling, latency budget, escape-hatch coverage. Wired into CI for ServiceNow and Salesforce tenants.
Every recipe is a standalone repo. Clone, set your tenant credentials, run the smoke test, and you've got a working agent in 10 minutes.
The prompts and eval cases are templated — swap your data, your taxonomy, your golden-set, and you've got a domain-specific agent in an afternoon.
Each recipe ships with a tenant-deployment guide for ServiceNow / Salesforce / Workday. Or hire our COE to do it with you — same code, more hands.
Subscribe below to get them the day they ship. We add roughly one recipe per sprint.
Build an Agentforce-orchestrated multi-step Journey Builder flow with personalization grounded in Data Cloud. Includes the segmentation eval and the lift-measurement harness.
End-to-end onboarding flow — equipment, accounts, day-1 readiness, manager intros — orchestrated through Now Assist. The conversation-design playbook and the SLA harness.
A reference MCP server pattern that exposes Confluence + Notion + a private GitHub org as a single agent-grounding source. Auth, rate-limiting, and access-control patterns included.
How we set per-tenant cost ceilings and per-agent latency budgets, and the alerting that fires the moment a drift breaks them. Includes the dashboard JSON and the alerting rules.
The moment a new recipe ships, you get a link to the repo, a 90-second video walkthrough, and the gotchas section. Unsubscribe in one click.