Team Avatar - Kane Hooper
Kane Hooper August 14, 2025

Introduction

G’day CTOs and product folks. If you’ve landed here, chances are you’re already thinking about how to build a mobile or web app that talks to your Salesforce data and does more than just display records. You don’t need me to convince you that AI and real‑time data matter; you’re looking for a practical way to bring them into your customer‑facing applications without reinventing the wheel.

In this article I’m going to walk you through how to marry Salesforce’s new AppLink add‑on and Agentforce to give your mobile app a conversational interface backed by live CRM data. AppLink essentially exposes your Mobile app as an API service in Salesforce and Agentforce. That means you can write your application in your favourite language, host it on Heroku and have Salesforce automatically create AI agent actions that can be invoked from your mobile app or a chat interface. Agentforce sits on top of that, providing the large‑language‑model smarts to understand natural language requests and orchestrate workflows across Salesforce, Data Cloud and your own services.

Over the next six sections we’ll look at how this works and what it looks like in practice. We’ll cover the basics of AppLink and Agentforce, explain why pairing them is the key to real‑time data updates and natural‑language interactions, explore a range of industry use cases, and sketch out an implementation roadmap that your engineering team can follow. By the end you’ll have a clear picture of how to build an AI‑enabled mobile app that lets your customers ask questions like “Where’s my order?” or “Can you update my address?” and have the answer come straight from Salesforce, via your own logic, in real time.

AppLink is a new add‑on that bridges the gap between your apps hosted on Heroku and Salesforce — including Data Cloud and Agentforce — without you having to write all the integration plumbing yourself.

At a high level, AppLink turns your app into an API service that Salesforce can discover, secure, and call directly. When you attach the add‑on to your app AppLink registers your endpoints inside Salesforce’s API catalog. From there, Salesforce automatically creates agent actions — reusable building blocks that can be invoked by Flows, Apex code, or Agentforce AI agents.

One of AppLink’s most important roles is managing authentication and user context. It makes sure that the link between your app and Salesforce remains secure and the data is protected. Your engineering team can focus on writing business logic instead of worrying about token management or API authentication.

Technically, AppLink is made up of several components:

  • The add‑on: connects your Heroku app to Salesforce.

  • Authentication and security: provides authentication, routing, and enforcement of Salesforce security.

  • Dashboard: to monitor connections, publications, and usage.

  • SDKs: make it easier to parse Salesforce request headers, perform DML operations, and interact with Salesforce data.

AppLink makes your app a first‑class citizen in the Salesforce ecosystem. That means any logic you host - whether it’s processing customer uploads, pulling in third‑party data, or crunching analytics - can be exposed to Salesforce and Agentforce in a secure, structured way. For a mobile app project, this allows your back‑end to participate in real‑time conversations and workflows right alongside Salesforce’s own features.

What is Agentforce?

Agentforce is Salesforce’s platform for building and running AI agents that can interact with your data, processes, and customers in natural language. It is my opinion that agents are going to be one of the most important economical changes in our history. As AI agents get better, they will be able to perform actions that were once relegated to a human user. Personally, I am already in the process of building out a number of local AI agents that are able to perform tasks on my behalf and improve my overall productivity.

Where traditional chatbots are limited to scripted responses, Agentforce agents can:

  • Understand a customer’s actual intent rather than respond to a few scripted key phrases

  • Turn complex requests into smaller steps and execute them in the right order.

  • Pull data from Salesforce and even third-party data.

  • Trigger actions in Salesforce; from updating a contact record to kicking off a multi-step approval flow.

Salesforce’s Agent Builder provides an easy to use low-code environment to build out your agent’s capabilities. You can combine pre-built actions (e.g., “Look up an order”) with custom actions exposed from your mobile app through AppLink. These agents aren’t static, they use LLMs to dynamically adapt responses and workflows to the user’s request.

Agentforce is designed to plug into multiple channels: web chat, mobile apps (via the Agentforce Mobile SDK), Slack, and other customer data source. This makes it possible for the same AI agent to work across your app, website, and messaging platforms while maintaining the same Salesforce-integrated intelligence.

When combined with AppLink, Agentforce becomes even more powerful: it can extend beyond Salesforce’s native data to interact with your apps complex logic and even third party integrated services. That’s the point where AI in your mobile app moves from “nice-to-have” chat responses to a real operational tool that can update records, process transactions, and guide customers through personalised journeys in real time.

Why Pair Mobile Apps with Agentforce?

The real strength of combining your mobile apps with Agentforce lies in exposing Salesforce data to your customer-facing applications, and doing so in a way that’s both secure and real time.

Real-time Salesforce updates from your app

When your mobile or web app is hosted on Heroku connected through AppLink, every customer interaction can be reflected in Salesforce near instantly. If a user updates their contact details, books a service, or changes an order in your app, that change flows straight into Salesforce without waiting for batch jobs or manual syncs. This allows your CRM to remain accurate and your teams always have the latest information.

Natural language as a front-end to your app and Salesforce

Agentforce gives customers an LLM-powered chat inside your app. So instead of navigating menus or filling forms, your users can type or speak a request like:

“Can you update my shipping address to 42 Top Street?” “What is the tracking number for my last order?” “Is my package going to arrive today?”

Agentforce interprets the intent, uses your app to perform any custom logic or external API calls, updates Salesforce through AppLink, and replies in natural language. From the user’s perspective, it feels like talking to a highly capable assistant that understands both their needs and your business processes.

Extend beyond native Salesforce capabilities

While Salesforce has robust tools, some workloads are better suited to a custom application - heavy computation, niche API integrations, or proprietary algorithms. Your custom code on Heroku is ideal for hosting that logic in the language and framework your developers prefer. AppLink makes that external capability look native to Salesforce and Agentforce.

For a CTO, this pairing means you can deliver more dynamic user experiences while keeping Salesforce as the single source of truth for customer data.

AI-Agent Use Cases for Customer-Facing Mobile Apps

When I think about where Heroku AppLink and Agentforce really shine, I picture the conversations I’ve had with clients over the years - the real situations that made them lean forward and say, “Yes, that’s exactly what we need.” Let me walk you through a set of scenarios in that same spirit.

  • Order tracking: The anxious customer waiting on a package

A regular rings you up, frustrated because they can’t find their tracking link. In an AppLink + Agentforce world, they just open your app and type “Where’s my order?” Behind the scenes, Salesforce pulls up their record, your Heroku service checks the courier’s API, and the agent replies with the exact location and expected delivery. No call queue. No copy‑pasting tracking numbers.


  • Menu questions: The dinner guest with dietary needs

I remember a restaurant chain that got constant calls about vegan options. Now, diners simply ask the in‑app chat. Agentforce pulls the latest menu from Salesforce, checks with a Heroku microservice that knows about last‑minute specials, and responds in seconds. Staff focus on service, not phone duty.


  • Billing queries: Getting the altest invoice

For a publisher, the end of the quarter used to mean inboxes overflowing with invoice requests. An agent now handles it: “Send my latest invoice” triggers a Salesforce fetch, a Heroku PDF generator formats it, and the customer has it in their email before they can even open their accounting software.


  • Technical support: Self‑serve tech support

A medical device provider needed a way to help patients without long hold times. Their app’s agent can now walk a new user through setting up their glucose monitor, step‑by‑step, referencing Salesforce knowledge articles and a Heroku‑hosted interactive guide.


  • Knowledge base: All your answers in one place

One health brand had hundreds of articles scattered across systems. We pulled them into a single searchable index on Heroku, wired it into Agentforce, and now customers can ask “How do I change my subscription?” and get a clear, immediate answer.


  • Customer complaints: When the bill looks wrong

Telco customers don’t like surprises. If someone asks “Why is my bill higher?” Agentforce reviews Salesforce usage data, Heroku analytics spots an overage, and the agent can apply a credit on the spot.


  • Financial matters: High‑stakes financial moments

From flagging fraudulent transactions in a banking app to renewing a policy in an insurance app, agents guide the user through each step while updating Salesforce in real time.


  • Placing orders: Personal shopping without the pushiness

In a luxury retail app, a customer says “Find a $100 gift for a 10‑year‑old.” The agent queries Salesforce for the catalogue, uses Heroku to check stock, and presents a curated list.


  • Automated support requests: Keeping machines running

For a manufacturer, IoT sensors feed data into Heroku. Agentforce spots an anomaly, books a service appointment in Salesforce, and tells the customer when the technician will arrive.


  • Booking appointments: Healthcare that feels personal

A patient can type “Book my next check‑up in September” into the app, and the agent checks the doctor’s availability in Salesforce, books it, and confirms instantly. The thread through all of these? Real conversations driving real actions — Salesforce keeping the record straight, your app doing the heavy lifting, and Agentforce making it all feel effortless to the user holding the phone.


Implementation Overview for a CTO

If you’re looking at this from the CTO chair, your job isn’t to write the API spec yourself — it’s to make sure the architecture is solid, the security is watertight, and your team knows exactly what they’re building. Here’s how I’d map it out:

  • Start with your back-end on Heroku

Pick the language and framework your crew knows best - Node.js, Python, Java, Ruby, whatever gets you moving fastest - and deploy to Heroku. This is where your custom business logic lives: crunching data, talking to third-party APIs, or running processes Salesforce isn’t built for.


  • Bring in Heroku AppLink

Add the heroku-applink add-on to your app. That’s the handshake between your service and Salesforce - the bit that makes your endpoints appear inside Salesforce as native, secure services.


  • Connect the dots to Salesforce

Using the CLI or dashboard, hook your Heroku app to your Salesforce org. Decide who the app will “act as”: the logged-in user, a user-plus with extra reach, or a dedicated integration account.


  • Describe what you’ve built

Write up your endpoints in an OpenAPI 3.0 spec, add the x-sfdc extensions, and publish through AppLink. This tells Salesforce exactly how to generate the external service and agent actions.


  • Wire it into Agentforce

In Salesforce, point your Agentforce agent at those new actions. Now when a customer asks in the app, Agentforce can call your Heroku service like it’s part of the family.


  • Put the agent in the app

Use the Agentforce Mobile SDK to drop the chat or voice interface into your iOS, Android, or cross-platform app. The SDK handles secure connections and passes customer requests straight to the AI agent.


  • Keep an eye on it

Use the AppLink dashboard to watch API usage, latency, and security logs. Scale Heroku dynos up or down as traffic changes.


A few things I always flag for CTOs:

  • Mirror your data access rules between Salesforce and Heroku.

  • Keep your API design consistent so new devs can onboard quickly.

  • Bake security reviews into your release process.

  • Measure the right things - whether that’s faster ticket resolution, higher NPS, or reduced call volumes.

Done right, Salesforce remains your source of truth, Heroku stays your flexible workbench, and the two together deliver an AI-powered experience your customers will actually use - and enjoy.

Conclusion

Pairing Heroku AppLink with Salesforce Agentforce gives you a straightforward path to building customer-facing mobile and web apps that can interact with Salesforce in real time - all while offering a natural language interface powered by large language models.

Heroku provides the flexibility to run your custom logic, integrate third-party services, and handle workloads that don’t fit neatly inside Salesforce. AppLink makes that logic visible and secure within Salesforce, automatically creating agent actions your AI agents can call. Agentforce then delivers the conversational front end, interpreting user requests, orchestrating workflows, and updating Salesforce and connected systems on the fly.

Across industries - from retail and banking to healthcare and manufacturing — the combination opens the door to:

  • Instant, AI-driven responses for customers.

  • Seamless updates to Salesforce as users interact.

  • The ability to blend CRM data with external sources in real time.

For a CTO, this architecture means you can:

  • Keep Salesforce as the single source of truth for customer data.

  • Maintain freedom in your technology stack.

  • Roll out new AI-powered capabilities to multiple channels without rebuilding core logic.

The practical takeaway: if your roadmap includes AI-driven customer experiences that are deeply tied to your Salesforce data, Heroku AppLink and Agentforce together offer a proven, scalable approach. The examples and patterns outlined here can serve as inspiration for your own use cases — and a blueprint for your technical team to deliver them.

 

Ps. if you have any questions

Ask here