At Salesforce, the internal Slack Apps team is tasked with a singular, complex mission: enabling the company’s 76,000+ employees to work more efficiently by bringing external systems directly into Slack—the central hub of collaboration. From managing approval workflows to sophisticated internal communication, the team builds solutions that eliminate the need for employees to jump between multiple applications and systems.
Collaboration at Salesforce has evolved dramatically over the years, from Chatter to Google Chat to Yahoo, finally settling on Slack as the central hub where everyone works. For the Slack Apps team, this presented both an opportunity and a challenge: how do you build sophisticated internal applications that serve 76,000+ employees without getting bogged down in infrastructure management?
This is the story of how Salesforce’s own development team chose Heroku to build five major Slack applications, proving that when it comes to rapid development and seamless scaling, Salesforce trusts Heroku with their most critical internal tools.
Heroku definitely made us get to market faster. We could have built and hosted these apps on a laptop if we had to, but we would have needed a whole team to manage the infrastructure and all the data services.
Transformative workflows: Slack apps built on Heroku delivered 3,310 years in productivity savings
The Slack Apps team operates on a simple but powerful principle: bring external systems directly into where employees already work. Instead of forcing users to jump between multiple applications like Workday, Concur, or various approval systems, they build Slack applications that centralize these workflows.
A perfect example is their Approvals app, which revolutionized how decision-makers handle approvals across the organization. Previously, approvals meant switching between different systems, dealing with missed emails, and lengthy delays. Now, everything happens directly in Slack, cutting expense approval time from 2 days to just 2 hours for each expense report, and saving a cumulative 3,310 years in approval wait times.
The team’s user-research-driven approach ensures they’re solving real problems. They identify pain points, then build solutions that eliminate friction and make daily workflows more efficient for tens of thousands of employees.
Heroku for small teams: Solving key problems for lean development teams
Before adopting Heroku, the Slack Apps team’s development process was manual and labor intensive. It managed servers across three different data centers, each configured differently with unique access requirements and environment variable setups. Every deployment meant generating Java .jar files and manually copying them to each server using terminal commands.
The operations team relied on manual runlists to deploy artifacts, set environment variables, and restart servers—a process that was prone to human error and incredibly time consuming. With three servers needed for redundancy, every deployment became a complex orchestration that slowed down development cycles significantly.
Scalability was another major constraint. The team was locked into just three instances of its service, which became problematic as Slack Apps’ userbase grew from 4,000 to 12,000 to 76,000+ employees. The existing infrastructure simply couldn’t adapt to changing demands, and the developer experience suffered under the weight of manual processes and configuration complexity.
Why Salesforce’s Slack Apps team chose Heroku over AWS
Beginning with a lean team of just five or six developers, the Slack Apps team needed a solution that would let them focus on building great applications rather than managing infrastructure. While AWS was an option, its steep learning curve and numerous components would have required significant time investment that the small team couldn’t afford.
I think everybody knows there’s a steep learning curve to AWS—there’s a lot of components to it. We wanted something a lot simpler that can help us, as developers, get something running quickly and scale. And so that’s what led us to look into Heroku.
Heroku’s benefits were immediately apparent: simplicity and ease of use. The platform handled complex infrastructure while enabling developers to quickly build, deploy, and scale. When they needed to implement an event-driven architecture using Kafka, Heroku’s ability to provision complex data services (such as Postgres and Key-Value Store, as well as Kafka) without managing redundancies sealed the deal.
Modern CI/CD: Leveraging Heroku pipelines and Terraform for stack evolution
The Slack Apps team’s Heroku journey demonstrates how a platform can grow with a team’s sophistication. It started simple, deploying Java applications with Heroku Key-Value Store and Heroku Postgres through the Heroku UI. But as new features became available, it continuously evolved its approach.
Heroku Pipelines was a game-changer, eliminating the slow process of rebuilding slugs for each deployment stage. Instead of slow builds for every environment, it could build once in development and promote through to production seamlessly. Now, the team builds once in development and promotes seamlessly through to production.
Review Apps transformed the development process, allowing team members to test features by simply opening a pull request. Though it required some creative mapping as the team moved to microservice architecture, the ability to click through actual UIs for code reviews was invaluable.
The shift to infrastructure-as-code using Terraform eliminated manual configuration processes. Everything moved to source control, removing the human error factor that was a friction point in previous deployments. Heroku Scheduler optimized costs by replacing always-on cron jobs with on-demand dyno execution.
Perhaps most significantly, Heroku Connect revolutionized how they integrated with Salesforce, eliminating REST API complexity entirely.
We started leveraging Heroku Connect and just interfacing with Salesforce as a database. We got rid of our REST client and now we just use the DB client to do everything, which is really awesome for us.
Five mission-critical apps on Heroku: Powering 76,000 users with event-driven Kafka architecture
The Slack Apps team has created five major applications on Heroku, each serving specific needs across Salesforce’s 76,000+ employee base:
- Amplify enables persona-based communication, allowing users to send targeted messages to specific groups within the organization—from all developers to C-suite executives—scheduled for optimal timing.
- Approvals centralizes approval workflows from multiple systems, cutting expense approval time from 2 days to 2 hours per report, and saving 3,310 years in cumulative wait times.
- Task Hub manages employee tasks like benefits enrollment and license confirmations, helping the licensing team recover $2 million in unused software licenses through streamlined bidirectional communication.
- Ask Concierge has resolved 68,000 issues without requiring human intervention, demonstrating the power of automated support systems.
- Einstein achieved remarkable adoption with 121,000 users in its first month, answering 120,000 queries and scheduling 2,000 meetings, showcasing the platform’s capability to handle AI-powered applications at scale.
Each application uses event-driven architecture built on Kafka, proving Heroku’s ability to support sophisticated, interconnected systems that serve enterprise-scale user bases.
Achieving massive scale on Heroku without infrastructure management overhead
For the Slack Apps team, Heroku’s greatest strength is its invisible reliability, without infrastructure overhead. The platform allowed this small team to reach the market faster and focus on what matters most: building great user experiences for their colleagues. It enables teams to build powerful applications without anyone in the development team or business leadership needing to think about the underlying platform or incurring the additional cost of managing it.
Launch your next project with Heroku
How Slack Apps did it
- Heroku for building Java applications
- Heroku Key-Value Store and Heroku Postgres for data management
- Apache Kafka on Heroku for event-driven architecture
- Private Spaces for managing the inbound/outbound allow list and IP allow list
- Heroku Connect for replacing REST API
