The image features a stat on the left that reads, “12 days to develop their first AI agent.” On the right is an interface with a calendar icon to the top left and sparkles above a human hand and a white robotic hand bumping fists to the bottom right.

Engine accelerates AI agent development on the Headless 360 platform

Agentforce combines developer tools like Builder, Apex, and Vibes so teams can design, test, and deploy in one system — cutting delivery from months to days.

The results

days to develop their first AI agent

1,000 +

automated tests per agentic deployment

Developer bottlenecks slowing work

Engine is a modern business and group travel platform built for small and medium businesses and trusted by over one million travelers. As the company grew from a startup into a $2.1 billion unicorn with over 1,000 employees, their team faced a growing need to move faster — delivering new capabilities quickly without losing control over quality, security, and governance.

Engine wanted to support employees and customers with agentic experiences that could handle refunds, complete transactions, and update accounts. That required complex integrations, strict testing, and enterprise-grade security.

Traditional options didn’t fit. Low-code tools were fast but limited, while custom development offered flexibility but often took months and depended heavily on a small group of developers. They needed a new approach — and a platform built to handle complex work.

Production-ready agents, no compromises

Engine no longer has to choose between moving fast and building robust, production-ready AI agents. With the Headless 360 platform teams work in a hybrid development environment where no-code, pro-code, and AI-assisted tools come together in a single, unified process.

Before, delivering an innovation like Eva, their customer self-service agent, would have required multiple tools and long release cycles. Now, everything happens seamlessly within one platform — from defining how Agentforce should behave to deploying it into production. This is how Engine was able to deliver Eva in less than two weeks.

And new capabilities build on what Engine has already created. Instead of starting from scratch, their teams reuse and adapt existing actions and Apex code — turning Eva into a foundation for future agents.

“Each Apex action we built could be called by any agent, any flow, any process,” explained Joshua Stern, Engine’s Director of Revenue Operations. “That modularity meant our second agent took days, not weeks.” The result is a platform where Engine can deliver transaction-ready agents quickly while maintaining control and trust.

No breaks between steps

When building a new AI agent, Engine’s team starts by defining how Agentforce should handle real scenarios — like canceling a hotel reservation. From there, they build or validate the logic and integrations essential for execution. For example, when a customer interacts with Eva, their request moves through a coordinated series of decisions and actions designed to complete real tasks — not just return answers:

Agentforce Builder: Eva starts by identifying the customer’s intent through defined topics like “Cancel Trip.” Each topic is tied to specific actions and rules, shaping how Agentforce responds.
Salesforce Flow: Once the request is understood, Flow determines the path forward using decision logic. For a cancellation, Eva first checks whether the customer is authenticated — prompting for credentials if not, or retrieving booking details if they are.
Custom Apex: With context confirmed, Apex executes the requested transaction. Eva queries Engine’s booking platform, applies cancellation policies, calls payment APIs to issue refunds, and updates booking records across systems.
Agentforce Vibes: When Engine needed Eva to handle financial transactions, developers described the requirement in plain language. Vibes generated the Apex logic and API integrations needed to connect Eva directly to payment systems, aligned with existing data and security rules.
Agentforce Testing Center: Before going live, every step of Eva’s cancellation flow was tested in a sandbox. Engine validated permissions, transaction logic, routing accuracy, and edge cases like failed payments or incomplete data.

12-day build, safer release

In the case of Eva, what would have taken months using traditional approaches was delivered in 12 days, with design, development, and testing happening in the same environment. Because everything runs through native application lifecycle management — the process that manages how software is built, tested, and released — teams can ship updates more frequently without introducing risk.

Testing, in particular, has become much more efficient. With Agentforce Vibes, writing tests that once took days now takes a few hours. Each deployment runs hundreds to thousands of automated tests before any changes are recommended.

Vibes also changes how developers approach backend work. “Using Agentforce Vibes means the AI is right there beside our developers, supercharging our Apex code,” said Stern. Instead of spending weeks building and refining complex integrations, teams can generate and iterate on it quickly — turning what used to be multi-week efforts into focused sprints of a few days.

Taken together, these changes mean that Engine can release faster with fewer risks. Teams spend less time on manual work, catch issues earlier, and ensure each update performs as expected.

Why Engine chose Salesforce

Salesforce Professional Services

In partnership with the Engine team, Salesforce Professional Services designed Eva to operate safely at scale, building in validation for identity, authorization, and eligibility before any action. By leveraging existing APIs and logic, they helped Engine move faster without rebuilding core systems.

Salesforce Trust Layer

The Salesforce Trust Layer ensures Agentforce Vibes runs in a secure environment, so Engine’s proprietary data stays protected. Teams can generate code and build agents quickly without exposing sensitive financial or customer data to external or unmanaged systems.

Real-time optimization

Engine uses Agentforce Observability to monitor usage, performance, and feedback across all agents — including Eva. When performance dips, teams can quickly update agent logic, test changes, and deploy fixes the same day without waiting for a sprint cycle.

Hybrid agent development

“I can design a conversation flow in Builder, immediately write supporting Apex in the same environment,” said Sarah Morton, Senior Salesforce Administrator. With vibe coding and built-in tools to test, deploy, and monitor Salesforce Analytics, Engine achieves both speed and technical depth.