Engine Delivers Autonomous Travel Support with Hybrid Development on the Agentforce 360 Platform

Engine’s engineering culture is built on a simple mantra: “ship fast, learn faster.” But as the $2.1B travel tech unicorn scaled from startup to over 1,000 employees, the company faced a critical paradox. They knew AI agents were the key to scaling their operations, but their existing enterprise development tools made building those agents a bottleneck. Traditional enterprise development cycles — taking months to spec, develop, test, and deploy new capabilities — were fundamentally incompatible with hypergrowth demands. Yet cutting corners on quality, security, or testing wasn’t an option when building agents that would handle customer transactions, process refunds, and access sensitive financial data.

“We needed enterprise-grade AI agents that could execute real business processes,” explains Joshua Stern, Engine’s Director of Revenue Operations. “That meant complex integrations with our CRM, proprietary booking systems, payment processors, and data warehouses. But we couldn’t spend six months on each agent. We needed a development methodology that delivered both speed and technical rigor.”

Engine’s previous attempts with disparate development tools created bottlenecks at every stage. Building required switching between separate low-code platforms and coding environments. Testing requires external frameworks. Deployment required custom integration work. Data access required building connectors for each application. Every integration point introduced delays, complexity, and potential failure.

Engine’s leadership set a clear goal: deploy enterprise-grade applications in days, not months, using a unified development platform. They needed a platform supporting both rapid prototyping and production deployment, tools enabling both citizen developers and experienced engineers, and comprehensive ALM capabilities building quality in from the start. Most critically, they needed a unified platform where applications could share modular, reusable code—enabling compounding velocity where each new capability accelerated future development.

How Salesforce Helps Engine

Engine chose the Agentforce 360 Platform not only because of their deep integration with the Salesforce’s ecosystem, but also because it offered a true hybrid development environment where low-code, pro-code, and vibe code tools worked seamlessly together within a unified ALM process.

The 12-Day Development Cycle

For their first production agent, Eva (Engine Virtual Assistant), the team established a rhythm that balanced velocity with rigor:

Days 1-3: Low-code foundation

Using Agentforce Builder, the team rapidly prototyped conversation flows, defined topics, and mapped user journeys. Salesforce Flow provides orchestration capabilities without writing code. This visual development allowed both technical and non-technical stakeholders to collaborate on design in real-time.

Days 4-8: AI-accelerated pro-code development

Engine didn’t just write code; they used Agentforce Vibes to generate the sophisticated, multi-system logic required for financial transactions. By describing intent in natural language, Agentforce Vibes produced the ‘invocable actions’ that allowed Eva to talk to payment processors, turning weeks of manual backend engineering into a 4-day sprint.

Days 9-11: Rigorous testing and refinement

The team used Agentforce Testing Center within their Salesforce Sandbox to validate changes without risk to live business operations. This was critical for Engine: they needed to test complex financial transactions, multi-system integrations, and permission models without any possibility of disrupting real customer bookings or payment processing. 

Agentforce Testing Center ran hundreds of discrete tests against their Apex code—validating business logic, permission models, and integration points. Agentforce Vibes also assisted with creating comprehensive test classes, historically one of the most time-consuming aspects of Apex development. What might have taken days of manual test writing now happened in hours, with AI assistance ensuring edge cases were covered and test quality remained high. Running these tests in their Sandbox gave them the freedom to build and test aggressively, knowing they could validate every change before it touched production.

“We learned from past ad-hoc testing issues,” Stern explains.” Agentforce Testing Center allowed us to run hundreds or even thousands of tests at once, drastically improving code quality and confidence before anything reached production.”

Day 12: Production deployment

With comprehensive test coverage ensuring quality, Eva went live—autonomously handling customer cancellation requests end-to-end, processing refunds, and updating systems without human intervention.

But the key to Engine’s velocity wasn’t just the hybrid tools—it was architectural discipline. “We prioritized building recyclable, modular, invocable Apex,” Stern emphasizes. “Each Apex action we built could be called by any agent, any flow, any process. That modularity meant our second agent took days, not weeks, because we could reuse existing building blocks.”

This approach transformed Engine’s development model from building isolated solutions to building an interconnected ecosystem where each new capability accelerated the next.

The Salesforce Difference

The Agentforce 360 Platform gives developers the tools to design, code, test, deploy, and iterate without ever switching systems or losing context.

This integrated approach transformed how Engine scales beyond initial use cases. After deploying Eva, Engine faced a new challenge: internal employees needed the same high-velocity automation for IT, HR, and Finance support. To solve this, they deployed Mae (Multi-purpose Admin Expert).

Mae operates across Slack and Salesforce, acting as a single interface for employee needs—from routing password resets to automating expense reporting. Because Engine used Agentforce Builder and Agentforce Vibes, they didn’t have to start from scratch. They simply “re-vibed” existing logic. By describing new requirements in plain language Agentforce Vibes acted as an AI-powered coding partner, generating necessary Apex modifications in seconds. Developers then refined the code, and Agentforce Builder allowed the team to map out employee-specific flows in hours.

This transition from external support to internal operations proved that a unified platform doesn’t just launch agents; it scales an entire enterprise. This integrated loop has transformed Engine’s speed of iteration into a three-part competitive advantage:

• Real-Time Optimization: When Agentforce Analytics flags a drop in resolution, Morton doesn’t wait for a sprint cycle. She updates the configuration in Agentforce Builder or uses Agentforce Vibes to generate updated logic, tests in a Sandbox, and deploys—often within a single afternoon.
• Compounding Velocity: Every action built with Agentforce Vibes is a modular and “invocable” building block. Mae was deployed in a fraction of the time by literally plugging in the same logic that powered Eva.
• Confidence at Scale: By utilizing the platform’s native ALM framework and Agentforce Vibes’ ability to assist with complex test classes, the team can “ship fast” without the fear of breaking critical financial integrations.

By removing the friction between “the idea” and “the code,” Engine has turned its development process into a continuous flywheel of innovation. With the Agentforce 360 Platform, Engine no longer views each project as a climb from zero; instead, every subsequent agent is smarter, faster, and more robust than the last.