Skip to main content
GPTfy - Salesforce Native AI Platform

Open Source AI. Your Infrastructure. Salesforce.

GPTfy connects Meta Llama running on AWS to Salesforce declaratively — with native PII masking, full audit trail, and automated bulk processing across your entire database.

For organizations that want to use open source models like Meta Llama without buying additional platform components, this demo shows how GPTfy connects Salesforce to Llama on AWS in minutes — with native PII masking, a complete security audit trail, and the ability to automate AI across hundreds of records without user intervention.

Capabilities demonstrated

Declarative Open Source AI Connection

  • Configure a Llama connection on AWS via Salesforce's secure credential management — entirely point-and-click.
  • The same connection pattern works for any open source or proprietary AI model on any cloud infrastructure.

Native PII Masking and Audit Trail

  • Account names and sensitive identifiers are automatically substituted with generic tokens before data leaves Salesforce.
  • A full audit record inside Salesforce shows the exact substitutions made, the Llama response, and what was shown to the user.

Multi-Object Data Context

  • Prompts pull from the primary Salesforce object plus any related objects — tasks, opportunities, custom objects, external data.
  • Field mapping is configured declaratively; adding a new related object or field requires no code changes.

Automated Mass Processing

  • GPTfy prompts integrate as Salesforce Flows, callable from Apex or REST endpoints for fully automated workflows.
  • Run account summaries or case analysis across hundreds of records automatically when records are created or updated.

Use this video when

A sales ops team wants account summaries generated by an open source model on their own AWS infrastructure instead of a third-party SaaS AI

An IT architect needs to confirm that sensitive account names are masked before any data reaches an external AI provider

A service team wants case sentiment analysis run automatically on every closed case without requiring an agent to trigger it manually

A Salesforce admin wants to extend the AI data context to include custom objects and managed package data without writing Apex

An operations team needs to process AI-generated insights across thousands of account records overnight in bulk without user interaction

A compliance team wants a complete audit trail inside Salesforce showing which fields were sent to Llama and how PII was protected

Frequently asked questions

GPTfy uses Salesforce's secure credential management — Named Credentials or External Credentials — to establish a declarative connection to your Llama instance running on AWS. No code is required for the connection setup. The configuration is entirely point-and-click inside GPTfy's admin settings, and all API key handling follows Salesforce security best practices.

Yes. GPTfy's native data masking runs on every prompt payload regardless of the connected AI model. In the demo, account names are replaced with generic identifiers such as Company One before the data reaches Llama on AWS. After the response is received, GPTfy re-identifies those tokens before showing results to the user. The complete masking behavior is logged in the security audit record.

Yes. GPTfy's data context configuration lets admins select fields from the primary Salesforce object and any related objects — including custom objects, managed packages, and even external data via Lightning Connect. In this demo, the account prompt pulls from the account record, related tasks, and related opportunities. Additional related objects can be added at any time without code changes.

Yes. GPTfy prompts integrate as Salesforce Flows and can be called from Apex code or REST endpoints, enabling fully automated mass processing. For example, you can trigger account summaries automatically across hundreds of records when they are updated, or run bulk case sentiment analysis when cases close. This automation runs without any user interaction required.

For every Llama interaction, GPTfy logs a complete audit record inside Salesforce: the exact data fields pulled, which values were masked (showing the substitution, such as an account name replaced with 'Company One'), the full AI response from Llama, and the data ultimately presented to the user. This record is stored natively in Salesforce and subject to your org's access controls and retention policies.

No. GPTfy is an AppExchange-approved app and works within your existing Salesforce org. The demo explicitly shows that you can connect to open-source models like Llama on your own infrastructure without requiring additional platform products. Your Llama instance runs on infrastructure your organization controls, and GPTfy manages the secure connection from Salesforce to that instance.

Ready to see this in your Salesforce org?

Book a 45-minute session and we'll walk through this use case using your own data.

Book a Demo
Video transcript
Organizations want to use powerful open source AI models such as Meta Llama for their Salesforce workflows. However, connecting external AI infrastructure to CRM data creates complex security, integration, and cost challenges. Most solutions force you to buy expensive platform add-ons or compromise on data security when bringing AI capabilities to your sales and service teams. GPTfy transforms your Salesforce into an intelligent workspace by bringing Meta Llama directly into your CRM — securely connecting open source AI power to your business data without additional platform costs. Let me show you how to bring Meta Llama open source AI directly into your Salesforce. GPTfy is an AppExchange-approved app that lets you connect to your own AI models on any infrastructure, and you can use it without buying additional platform components. I want to show you three things: how you connect to open source models like Llama, what business scenarios this enables, and how your data stays secure. First, here's the connection setup in GPTfy's configuration. I can see my Llama connection running on AWS. This is completely declarative — just a simple connection using Salesforce's secure credential management. With this setup, you can bring any open source AI into Salesforce. Now watch what you can do with it. I'm on an account record with opportunities and related data. GPTfy is embedded directly on this page with predefined prompts for accounts. I'm going to run Account Summary. Watch what happens. GPTfy pulls information from Salesforce and sends it directly to Llama on AWS, then displays an intelligent response. You're seeing CRM data enhanced by Llama's AI capabilities — summaries, sentiment analysis, and business insights that would take your team a significant amount of time to compile manually. Here's where it gets really powerful. Let me show you what happened behind the scenes. GPTfy automatically creates an audit record for every AI interaction. Looking at this, you can see how the security layer protected sensitive information by replacing account names with generic identifiers like Company One before sending to Llama. Let me show you how the prompt builder makes this work. You can see the prompt setup and how it grounds AI responses with specific Salesforce data, like annual revenue. The connection shows it's linked to Llama on AWS, with options for field updates based on AI responses. Now we're looking at the data mapping — pulling from the account record, specific fields, related tasks, and related opportunities. But here's the powerful part: if I want more information, I can add any related object through standard Salesforce relationships. This includes custom objects, managed packages, and even external data via Lightning Connect. The field mapping shows exactly which data gets sent and how masking protects sensitive information — account names become Company One or Company Two, which is exactly what we saw in the security audit. These prompts integrate as Salesforce Flows, so you can use them in Apex code or REST endpoints. But here's the real game changer: you can automate them to run without any user intervention in mass processing mode. Imagine automatically generating account summaries for hundreds of records when they are updated, or processing bulk case analysis when cases are closed. AI can handle mass processing scenarios across your entire database. Your sales team gets intelligent account insights and next best actions instead of manually analyzing scattered CRM data. Your service teams receive automated summaries and sentiment analysis that accelerates resolutions. Your operations teams can process data with AI-driven insights without manual work. The result: open source AI integrated directly into your existing Salesforce workflows, running on your chosen infrastructure with complete data control. Check out Salesforce and AI real-world use cases in our demo section, or book a demo to see your Salesforce and AI use cases in action.

Explore More

Demo Videos

Browse all Salesforce and AI demo videos by category.

Bring Your Own Model in Salesforce

Connect any AI model to Salesforce without vendor lock-in using GPTfy's BYOM architecture.

API Data Sources

Connect external data sources and APIs to enrich Salesforce AI context.

Salesforce AI Use Cases

Explore real-world AI use cases across Sales Cloud and Service Cloud workflows.

Last updated: February 2026