WorkOS Empowers Non-Engineers to Build AI Tools

Garrett Galow explained that while AI has made building applications more affordable, the process of deploying these apps for colleagues still poses a considerable obstacle.

Many employees possess valuable ideas for solving internal problems but lack the necessary technical infrastructure to safely implement and share these solutions.

An account executive named Perona developed 'Perona GPT' to assist the sales team during her maternity leave, demonstrating the potential of employee-driven innovation.

Despite its effectiveness, the tool was inaccessible to CEO Michael because he did not have a Lovable account, exposing a common barrier to internal adoption.

The company was forced to assign two applied AI engineers to re-code and re-deploy the application, highlighting an unnecessary bottleneck in the process.

While tools such as Lovable, Opus 48, and Cloud Code have simplified application creation, the act of making these tools broadly available within a company has not seen similar advancements.

Non-engineers often struggle with fundamental infrastructure tasks, including setting up accounts for deployment platforms like Vercel or Cloudflare.

The core challenge lies in providing secure, authenticated access to internal applications without burdening engineers with manual deployment tasks or exposing sensitive API keys.

WorkOS established two key objectives: enabling any employee to deploy applications without requiring infrastructure or security approvals, and ensuring immediate, frictionless distribution.

The overarching vision is for employees to share a link in Slack that provides instant access to any colleague, eliminating bureaucratic hurdles and technical 'red tape.'

Building applications is inexpensive, but verifying their practical utility necessitates real-world feedback from users, according to the speaker.

WorkOS implemented company-wide 'Cloud Days,' single-day hackathons where teams develop applications, specifically designed to generate structured feedback loops.

During these events, non-technical employees are paired with engineers to lead application development, with engineers acting as consultants to remove technical obstacles and offer guidance.

WorkOS developed 'Wow CLI' to abstract away intricate technical details such as Git, GitHub authentication, and Doppler secrets management, simplifying internal app creation.

The tool is distributed to all employees via MDM, ensuring that their development environments are immediately ready for use.

A shared skills repository allows employees to contribute and retrieve reusable code snippets, facilitating common workflows and reducing repetitive development.

The CLI proactively addresses common roadblocks by performing automatic pre-configuration on developer machines, streamlining the setup process.

The Wow CLI automates the creation of GitHub repositories using pre-configured enterprise templates, streamlining the initial setup for new applications.

Deployment is optimized using Cloudflare, which provides rapid internal hosting for applications.

Users can transition directly from CLI initialization to AI-assisted coding, for example with Claude, immediately after deploying their application.

The internal registry currently tracks 278 total applications pushed, with 21 deployments recorded on the day of the demo alone, indicating active use and rapid development.

The internal application ecosystem at WorkOS includes a variety of tools such as APIs, Slack bots, and custom dashboard interfaces.

Sixty of these applications utilize custom-branded domain names for easier identification and access.

All these tools are rigorously protected by Okta SSO, ensuring that no external access is permitted and internal data remains secure.

Manual processes for formatting and distributing newsletters to Slack and email were identified as a significant pain point within WorkOS.

Garrett Galow developed an internal tool to centralize the drafting of change logs, simplifying the process for company-wide communications.

This tool enables non-technical team members to draft content, which is then reviewed and published directly via Slack, streamlining the update process.

The Atlas platform's design emphasizes invisible infrastructure, ensuring developers do not need to concern themselves with underlying deployment targets like Cloudflare, Vercel, or AWS.

Another core principle is security by default, which involves mandatory Okta SSO authentication to protect sensitive Personally Identifiable Information (PII) and internal data.

The platform also prioritizes discoverability, making internal tools easily accessible to any employee simply by sharing a link in Slack.

Each application automatically includes essential resources such as compute, storage, D1 databases, and KV values, providing a comprehensive environment for development.

Token-based authentication is implemented for all API access, enhancing the security posture of internal applications.

All network requests are logged and auditable, which helps in detecting potential data leaks or security vulnerabilities.

Secrets management is centralized through Doppler, preventing the accidental exposure of sensitive API keys or credentials in Git repositories.

WorkOS currently maintains a catalog of approximately 60 active applications that are regularly used and updated.

These internal applications collectively process over 3 million requests per week, a figure that includes significant machine-to-machine traffic, particularly via Slack.

Among the notable tools is one that enables users to upload arbitrary HTML to generate shareable links for various mocks and diagrams, showcasing practical utility.

WorkOS has successfully achieved its primary goal of empowering non-engineers to deploy applications.

Remarkably, 37% of the 60 deployed internal applications were developed entirely without any engineering commits.

These tools are created through direct collaboration between employees and AI models like Claude, highlighting the effectiveness of the platform.

WorkOS also intends to expand its data integrations to simplify connections with various data sources, such as Snowflake.

Automatic Slack integration is also in development to embed these tools directly into team workflows, enhancing accessibility and collaboration.