Governance & Trust
Compliant by design, ready for what's next. Honors the regulatory floor your org already lives under (SOC 2, HIPAA, GDPR, FERPA where applicable) and prepares for what's coming: EU AI Act, NIST AI RMF alignment, emerging US state AI rules, and Salesforce trust and shield requirements for Agentforce. Audit trails, scoped permissions, data residency, and humans-at-the-helm wherever the model takes consequential action.
Agentforce 360 + Headless 360
Subagents, actions, and instructions, with Agent Builder, Agent Script (ASL), Agentforce Voice, and Intelligent Context grounded in Data 360. Headless 360 (announced at TDX 2026) makes the same surface invokable from any client through the API, MCP, and Agent Script. Slack-native, no browser required.
Multi-cloud integrations
Connectivity between Salesforce, AWS, GCP, Azure, Cloudflare, Oracle, and AI providers. Identity, data movement, and event flow across cloud boundaries.
Real-world AI workflows
Agent Workflow Design and Context Engineering applied to actual business processes. Humans-at-the-helm at every consequential decision point.
Evals
Quality measurement for agent and AI outputs using established benchmarks and task-specific eval suites, so you ship with evidence instead of vibes.
Harness engineering
The scaffolding around the model: tool wiring, retrieval, guardrails, observability. The harness is what determines whether an agent is useful or a toy.
Context & memory management
Designing what an agent knows, what it remembers across sessions, and what it should forget. Context windows, retrieval, scoped memory, and the policies that keep agents accurate as the conversation grows.
Salesforce Systems Thinking
The four systems wired together as one surface:
- Context: Data 360
- Work: Customer 360
- Agency: Agentforce 360
- Engagement: Slack
The agent gets the data, the workflow, and the action. Humans-at-the-helm across all surfaces and touchpoints.