🤖 AI Summary
Salesforce unveiled Agentforce 360, the latest iteration of its enterprise AI agent platform, ahead of Dreamforce. The update adds Agent Script — a prompting tool (beta in November) that programs agents with if/then-style logic to handle nuanced customer queries — plus Agentforce Builder (beta in November) for building, testing and deploying agents from one console, and Agentforce Vibes, a no-code/low-code “app vibe” coding feature. The platform also supports “reasoning” models from Anthropic, OpenAI and Google Gemini (models that claim to deliberate before answering), and deepens integration with Slack: core Agentforce apps will surface in Slack starting this month and expand through early 2026, while Slack pilots a personalized Slackbot agent. Salesforce says Agentforce has 12,000 customers, with early pilots for the 360 upgrades at Lennar, Adecco and Pearson.
The announcement is significant because it signals Salesforce’s bid to own enterprise agent workflows through tighter tooling, multi-model support and workplace embedding — a direct play against recent moves from Google (Gemini Enterprise) and Anthropic (Claude Enterprise). Technically, the emphasis on conditional prompting, unified build/deploy tooling, and model-agnostic “reasoning” agents aims to improve predictability and operationalization. But broader industry headwinds remain: enterprises still struggle to realize ROI (an MIT study found ~95% of pilots fail), so Salesforce’s challenge is turning richer capabilities and Slack integration into reliable, production-ready value.
Loading comments...
login to comment
loading comments...
no comments yet