{
  "id": "agentforce",
  "url": "https://oneninenordic.com/services/agentforce",
  "markdown": "https://oneninenordic.com/services/agentforce.md",
  "json": "https://oneninenordic.com/api/v1/services/agentforce.json",
  "title": "Agentforce",
  "question": "Where would an agent actually pay off first?",
  "summary": "AI agents built into sales, service and field work on Salesforce, with the permissions, approvals and review that make them safe to run.",
  "outcome": "efficiency",
  "unit": "core",
  "platform": [
    "Salesforce",
    "Agentforce",
    "Data 360"
  ],
  "people": [
    "morten-brandt",
    "jesper-andersen",
    "sinuhe-zamarron"
  ],
  "cases": [
    "universal-robots-mir-one-platform"
  ],
  "order": 1,
  "body": "We treat an agent as model + tools + rules working inside defined boundaries. The first agent is chosen for a bounded, repetitive process where a baseline exists, so the result can be measured. We design the workflow, the data access and the controls before the prompt.\n\nTypical engagement: a two-week discovery that names three candidate processes, then a 90-day path that takes one of them from bounded test to production with monitoring in place."
}