Salesforce SIforce Consulting

Your teams now use AI tools beyond Salesforce, from Claude to Slack. SIforce, formerly AIforce, connects them to your Salesforce data and workflows, with your permissions enforced every time. Systep designs the architecture, security and governance, so every AI tool gets one safe, controlled route into your CRM.

  • Architecture first — which AI tools connect, how, and with what access
  • Built on the Headless Toolkit — MCP servers, skills, APIs and the Salesforce CLI
  • Permissions enforced — every request runs under each user's Salesforce access
  • Governed write-back — clear rules on what AI can change, with audit logging
  • Ready for what's next — new skills and tools added as Salesforce releases them
 
 

Discuss Your Salesforce AI Strategy

Speak with a certified Salesforce expert about your Salesforce AI strategy. We'll be in touch within 1 business day.

What you'll get

  • A session with a certified Salesforce consultant
  • A view of which AI tools should connect to Salesforce
  • An outline SIforce architecture for your org
  • Clear next steps and an estimated scope

Name
How can we help?


Prefer to talk? enquiries@systep.com | (212) 347-5505

Certified Salesforce consultants on demand
Salesforce Partner USA - Outsource Salesforce Administration
Your outsourced Salesforce team
Salesforce Managed Services - US Partner
Salesforce Partner USA - New York

What SIforce is, and what it isn't

SIforce, formerly AIforce, connects Salesforce data and workflows to AI tools outside Salesforce. Salesforce launched it as AIforce at Dreamforce on September 15, 2026, and renamed it in October 2026. Some Salesforce documentation still uses the old name.

SIforce is not a single product or license. It is an architecture: a layer that separates Salesforce data, logic and workflows from Salesforce's own screens. AI tools outside Salesforce can then query records, update data and run processes directly. Salesforce's help documentation itself describes it as not a single product or tool.

That matters because your people will use more than one AI tool. Without a governed layer, every tool gets its own export, its own integration and its own security gaps.

Salesforce SIforce Consulting

The Headless Toolkit

SIforce is built on Salesforce's Headless Toolkit, which has four building blocks:

MCP servers

Model Context Protocol servers connect AI agents to Salesforce through an open standard. Salesforce hosts standard and product-specific servers, and you can register external or custom ones.

Plugins and AI skills

Plugins bundle skills, MCP servers and tools into one install. Skills teach an AI tool how to do specific Salesforce tasks correctly.

Salesforce CLI

Build, deploy and manage Salesforce components from the command line.

APIs

Programmatic access to Salesforce data, business logic, workflows and Agentforce for agents, apps and external systems.

MCP and skills

MCP gives AI tools a standard way to discover and use Salesforce actions, such as finding an account, updating an opportunity or launching a Flow. Skills package the know-how for doing those tasks well. We design which actions each AI tool may use, build custom skills for your processes, and combine tools into custom MCP servers where a workflow spans systems.

Guidance

Skills

Define how Salesforce tasks should be completed, in line with your processes and standards.

Execution

MCP

A standard protocol that enables AI tools to discover and use Salesforce actions.

  • Find an account
  • Update an opportunity
  • Launch a Flow

Cross-system workflows

  1. We build custom skills around your business processes.
  2. We define which actions each AI tool is permitted to use.
  3. We combine tools into custom MCP servers for workflows that span multiple systems.

Permissions

Hosted MCP servers authenticate through your Salesforce org. Your org's permissions decide what any external AI tool can see and do. That is a strength, but it also exposes existing weaknesses: if sharing rules are too broad today, an AI tool will inherit that.

Before anything connects, we review:

  • Profiles, permission sets and sharing rules

  • Field-level security on sensitive data

  • Integration users and OAuth configuration

  • Which actions are read-only and which can write back

Connected interfaces

SIforce is the foundation for several end-user experiences:

  • Claudeforce

    Salesforce inside Anthropic's Claude.

  • Slackforce and Slackbot

    Salesforce data, views and actions inside Slack.

  • Other AI platforms

    Salesforce has announced integrations including Google Gemini Enterprise.

  • Your own agents and apps

    Built on the APIs and MCP servers.

The foundation

Salesforce SIforce

Salesforce SIforce brings business data and workflows to any AI agent.

  • MCP servers
  • APIs
  • Command-line tools

What we deliver

  1. Architecture review: Which AI tools should connect, how, and with what scope.

  2. Security and access design: Permissions, integration users, authentication and audit.

  3. Build: MCP configuration, custom skills, Flows and APIs.

  4. Governance: An approved-tool list, usage monitoring and a write-access policy.

  5. Ongoing support: Upgrades and new skills as Salesforce releases them.

Design it once, connect every AI tool safely

Frequently Asked Questions

  • SIforce is Salesforce's layer for connecting your Salesforce data, logic and workflows to AI tools outside Salesforce. It lets tools such as Claude, Slack and your own agents query records, update data and run processes directly, under your Salesforce permissions. It is an architecture rather than a single product.

  • Yes. Salesforce launched AIforce at Dreamforce in September 2026 and renamed it SIforce in October 2026. Some Salesforce documentation still uses the original name.

  • Agentforce is Salesforce's platform for building AI agents that work inside Salesforce and on your customer channels. SIforce opens Salesforce up to AI tools outside it. The two work together: many organizations use Agentforce for customer-facing agents and SIforce to give employees Salesforce data in the AI tools they already use.

  • MCP (Model Context Protocol) is an open standard for connecting AI tools to business systems. Instead of building a custom integration for every AI tool, you expose Salesforce actions once through MCP servers, and any compatible tool can use them. Salesforce hosts standard MCP servers, and you can add custom ones for your own workflows.

  • Salesforce has announced connections for Claude through Claudeforce (Salesforce in Claude), for Slack through Slackforce, and for Google Gemini Enterprise. Any agent or app that supports MCP or Salesforce's APIs can connect too.