Skip to content
Agents tracked: 501 Releases (7d): 777 Agent status: no reported issuesSince last limit reset: Claude 10d · Codex 8d For AI agents: Markdown Updated Nov 21, 2026
Case study OpenTableResults as reported by Salesforce

How OpenTable resolved restaurant and diner questions with Agentforce

OpenTable launched Agentforce agents on its website for restaurants and diners, grounded in its existing knowledge articles. Salesforce reports that most restaurant cases were resolved within weeks of launch.

The problem

Simple questions about reservations, reward points and account updates came in at high volume. The FAQ chatbot couldn't handle nuanced or multi-part questions, and knowledge-base searches returned too many results. Reps switched between several disconnected systems.

How they did it

  1. Grounded the agent in 1,500 Service Cloud knowledge articles, using Data 360 for context.
  2. Launched a restaurant-facing agent first (about three weeks to launch), then a diner-facing agent.
  3. Set the agent to hand complex cases to human reps along with the full conversation.

Results as reported by Salesforce

  • Salesforce reports 73% case resolution within 3 weeks of launch for the restaurant agent.
  • The two agents handle about 11k conversations a week.
  • The source reports a 40% improvement over the previous chatbot.

These figures are as reported by Salesforce in Salesforce customer story ↗ (Aug 7, 2025). Salesforce sells the product used here. AgentGid did not measure or independently verify them.

Takeaway. Launch one audience-specific agent on a well-maintained knowledge base first, then reuse that setup for the next audience.
AgentGid's take

Best suited to a team already on Service Cloud with a maintained knowledge base, since the setup relied on 1,500 articles and Data 360. Watch the cost: at 11k conversations a week, per-conversation ($2) or per-action ($0.10) pricing adds up, and the 73% and 40% figures are Salesforce-reported. Rasa (Free) is a cheaper open-source option if you have developers, though its free tier covers only 1,000 external conversations a month.

The agent used here

Similar use cases

Guides