Claudeforce News

While I build out my Salesforce Field Service + Claude (AI) model I thought it would be a good time to start tracking my skills and how they are matching up with this new relationship. So far it looks pretty good. I will be providing a run through of my new project: Salesforce Field Service Case Resolution AI Architectural model. It will include the following items: Management Summary showing the highlights of the build and an end to end Demo of the entire process. Below is a quick notice on a very interesting concept that will begin to take over meetings and mindsets.

News: Anthropic brings Claude into Google Meet to tackle tasks live

11/16/2026 -

Anthropic has announced Claude Meetings, a beta feature for Team and Enterprise customers.

Rather than simply recording or summarising a meeting, Claude can appear as a visible participant, follow the conversation, answer questions in chat and help develop plans and timelines while the meeting is taking place.

Anthropic is also expanding its workplace tools with Claude Whiteboards, Claude Sheets and Claude Data.

There is an important connection to the Salesforce partnership announced in August.

Salesforce and Anthropic introduced Claudeforce, combining Claude's reasoning with Salesforce data, workflows, business rules and governed actions. The initial Salesforce-in-Claude offering included 37 prebuilt sales skills. Read Salesforce's original announcement.

Salesforce Investor Relations

Why it's worth your time: This is a shift in how enterprise applications are likely to be architected.

Consider a meeting about an important customer issue:

  1. Claude listens to the conversation.

  2. It identifies the underlying business problem.

  3. It retrieves relevant customer and service information.

  4. It proposes actions.

  5. Approved actions are executed through governed enterprise systems such as Salesforce.

The architecture challenge is no longer simply connecting a model to an API. It is managing the entire chain: identity, permissions, data access, business rules, action approval, auditability and human accountability.

A particularly important design question is whether an AI meeting participant should be able to execute an action merely because someone discussed it. My answer would be no: discussion, recommendation and authorization should remain distinct.

This is exactly the sort of architectural thinking that separates someone who can configure AI tools from someone who can design enterprise AI systems.

Next
Next

The Book of Five Rings: Five Timeless Lessons for AI Architecture