Talk to Professional Services Book a Demo

How Pulse Uses AI

AI at Pulse

AI that does the legwork and asks before it acts.

For home care, behavioral health and skilled nursing teams, Pulse puts AI to work in two places: Pulse AI Agents, built on Agentforce, and the workflow agent in Pulse Core, built on Claude from Anthropic. Both are designed so your team stays in charge.

Built on Salesforce. Built with Claude.

Two kinds of AI, and the team that builds with them.

Pulse AI Agents, built on Agentforce.

Ask about a referral, patient, applicant or schedule in plain language and get the answer from your Pulse records, within each user's Salesforce permissions. Listed on AppExchange after passing Salesforce's security review. See Pulse Agents

The Pulse Core workflow agent, built on Claude.

It proposes the next step, asks your team for a missing fact, or moves work forward when its requirements are met, all inside your approval rules. The design is model-flexible, so the AI provider can change without rebuilding your workflows. See Pulse Core

People who build agents.

Pulse Professional Services (formerly Madison Ave Consulting) designs, builds and supports Agentforce agents for intake, recruiting and field teams, and helps teams choose between Agentforce and Claude for each use case. Start with a free org health check that we produce with Claude. See the AI practice

Salesforce approved the security review for Pulse AI Agents in August 2026, and the package is listed on Salesforce AppExchange. See the AppExchange listing

You decide how much the agent does on its own.

These settings apply to the Pulse Core workflow agent. Pulse AI Agents always wait for the user to confirm a change.

  • Suggest only. The agent proposes the next step, and your team carries it out.
  • Act with approval, the default. The agent prepares the step, and it waits in a decision queue until a person approves or rejects it.
  • Handle low-risk steps automatically. Only if you turn it on, and only for low-risk record updates. Every other step still waits for a person to approve it.
  • Logged with a reason. Every proposal and decision is logged with the reason behind it.
  • Suggested facts wait for review. Pulse reads the emails tied to a record and suggests the facts to capture. Your team approves each one before it touches the record.

Where your data goes.

The two kinds of AI in Pulse take different paths. Here is each one, in plain terms.

Pulse AI Agents.

Pulse AI Agents run on Agentforce inside Salesforce, so Salesforce's Einstein Trust Layer protections apply. Every answer respects the user's Salesforce permissions, and nothing changes on a record until a person confirms it. How Salesforce describes its Trust Layer

The Pulse Core workflow agent.

The workflow agent calls Claude through Anthropic's API from your Salesforce org, using a Salesforce Named Credential, so no keys sit in code. This path is separate from the Einstein Trust Layer. We review this path with your compliance team before any patient data is used.

Claude through the Salesforce trust boundary (Coming).

We are building Pulse so customers can run Claude through Salesforce's trust boundary. Salesforce now offers Claude inside its trust boundary. Read the Salesforce announcement of August 26, 2026

Access controls, agreements and AI data handling are covered on our security page. Security and your data

Coming next.

Work in progress is labeled Coming.

  • Document agent (Coming). An agent that reads the referral packets your intake team receives.
  • Voice agent (Coming). An agent for caregiver call-outs and shift confirmations.
  • Voice-note activity logging (Coming). Field reps dictate an end-of-day note, and the agent logs their visits, notes and referrals.
  • AI case summaries (Coming). AI summaries of each case for the team working it.

Help choosing and building with Agentforce and Claude.

Pulse Professional Services has been a Salesforce consulting partner since 2018.

  • Agentforce or Claude, use case by use case. We help teams choose between Agentforce and Claude for each use case.
  • Salesforce in Claude setup. We set up Salesforce in Claude for your team.
  • Agentforce agents. We design, build and support Agentforce agents for intake, recruiting and field teams, on data and processes we map first.
  • Tools we build with Claude. Our free, read-only Salesforce org health check is produced with Claude. It shows where an AI agent would help and where your data needs work first.

Questions about AI at Pulse.

Short answers for operators and their compliance teams.

Which AI models does Pulse use?

Pulse AI Agents run on Agentforce. The Pulse Core workflow agent is built on Claude from Anthropic, and its design lets the provider change without rebuilding workflows.

Can the AI change records on its own?

Pulse AI Agents never do. They propose an update and wait for the user to confirm. The Pulse Core workflow agent acts with approval by default, and you choose whether it may handle low-risk steps on its own.

How do the AI features handle patient data?

Pulse AI Agents run inside Salesforce, where the Einstein Trust Layer applies. For the Pulse Core workflow agent, we review data access with your compliance team before any patient data is used.

What does AI cost?

AI usage is priced as a custom add-on to the Pulse license. The license is $89 per user per month, and organizations with 500 or more users move to custom Enterprise pricing.

Keep exploring: Pulse Agents, Pulse Core, AI practice, Free org health check, Security and your data, Pricing

Planning an Agentforce or Claude project of your own? Talk to Professional Services

See AI in your workflow.

Book a demo and see Pulse AI Agents answer from your records and wait for your team to confirm each change.

Book a Demo