About Litmos
Litmos is a corporate learning platform that develops eLearning solutions for top-performing companies. Established in 2007, they offer easy-to-use Learning Management Systems (LMS) and content libraries that 4,000 companies trust to create, curate, and connect learning to 30 million users in 150 countries across 35 languages.
In 2022, Litmos was carved out from SAP and acquired by Francisco Partners.
Litmos’ Challenge
With new ownership came a new leadership team at Litmos, including Chief Technology Officer Tommy Richardson. His mandate was clear: transform the company’s learning platform to drive better outcomes for learners.
“Our aim was to innovate the industry in a way that would produce better outcomes for our learners,” says Richardson.
While corporate learning platforms have remained largely unchanged for decades, LLMs are poised to disrupt that stagnation. Litmos, encumbered by a legacy codebase and hundreds of thousands of rigidly formatted courses, saw the risk of falling behind—and the opportunity to leap ahead.
Rather than settle for incremental feature updates, Richardson and his team viewed AI as the key to meaningful innovation. After evaluating several paths—from content creation to validation—they chose to focus on the learner experience. Their goal was to make the LMS more intuitive, enabling human-like interactions that help learners engage more deeply.
“AI is God’s gift to EdTech companies,” Richardson says.
One persistent challenge for learners is content discoverability. LMSs house vast libraries of materials—video, audio, PDFs, and especially SCORM-packaged courses. But current search systems rely on rigid keyword matching, which doesn't reflect how people naturally ask questions. As a result, learners often spend more time navigating the platform than actually learning.
Richardson believed that with a more conversational, human-like interface, learners could access the information they need more quickly—and come to rely on the LMS as a go-to resource.
“If we can—even at a minimum—give learners easier, more intuitive access to content, we can improve learning and retention. That’s a home run,” he says.
Why Tribe AI?
Litmos already had a talented team of engineers on staff, but the team agreed that partnering with an experienced third party would help accelerate the project.
Richardson discovered Tribe AI through a recommendation from a trusted former colleague at Francisco Partners, Litmos’ private equity investor. During an introductory meeting, Tribe shared their references and past projects that demonstrated a very technical business perspective on AI.
“Many companies ‘do AI’ but we wanted to partner with one that specialized in it. After meeting Tribe, I knew they were the one.” said Richardson.
Solutions
AI Assistant
Because the Litmos LMS was fundamentally content—which is, at its core, language—the Litmos team knew that LLMs were a likely path to reaching their goals.
The Tribe team proposed building an AI Assistant alongside various LLM transformations to improve content discoverability and deliver a human-like LMS interface.
Tribe agreed that building a tool that helped learners find the content they need quicker would not only improve learner outcomes but would also increase LMS adoption. Additionally, Tribe proposed building in AI Assistant functionality that would both answer and pose questions for learners based on LMS content. This added functionality would aid in the reinforcement and retention of information. It would also allow learners to use natural language to ask questions like “I’m looking for courses that will help me X.” or “Can you remind me of the key functionality of Y?”
The project was broken into two phases. Phase 1 was focused on delivering a proof of concept (POC) that showcased a more desirable state of LMS functionality than what currently existed. The POC prioritized a subset of courses (100) and two key pieces of functionality: “assign courses” & “see my courses that are due.”
Phase 2 was dedicated to delivering the minimum viable product (MVP) for all courses (300,000+) and all functionality including multi-language content, secure access control, plus integration with the client’s system and database while maintaining business and AI logic as a separate modern system.
AI-Powered Content Generation
Beyond just enhancing content discoverability with the AI Assistant, the Tribe and Litmos teams identified a powerful opportunity to streamline how course materials themselves are created. Traditional course development workflows are time-intensive and often require coordination between instructional designers, subject matter experts, and content teams. To accelerate this process without compromising quality, Tribe helped Litmos develop a two-part AI content generation system embedded directly in the LMS.
Module Generation
The first piece of the solution is the Module Generator: a tool that creates complete, multipage learning modules in minutes. More than just stringing together paragraphs of text, the Module Generator applies instructional design principles like Bloom’s taxonomy, ensuring the content follows pedagogically sound learning objectives.
It grounds its output in Litmos’ vast content library, drawing from over 600,000 existing courses, and can generate up to 30-page modules—with as many as 90 modules created simultaneously. Each module includes titles, content blocks, and relevant images, enabling teams to scale high-quality learning materials with unprecedented speed.
AI-Enhanced Content Authoring Tool (CAT)
The second layer of the solution enhances Litmos’ existing Content Authoring Tool (CAT). Tribe worked with the Litmos team to embed 12 different AI-enabled components into the CAT interface, making it easy for instructional designers and admins to create and customize individual course elements on the fly.
These components range from simple text blocks to more complex interactive elements like Flip Cards. The more advanced components are capable of:
- Generating multiple text and image pairings simultaneously
- Automatically pulling relevant imagery through integrated image queries via Unsplash
- Maintaining groundedness, whether the source is Litmos’ course library or content uploaded by a customer
This blend of full module generation and granular content block creation gives users both scale and control. Instructional designers can quickly (in 3-4 minutes) spin up draft modules using AI, then dive into specific areas to refine, augment, or personalize as needed.
Together, these tools act as a specialized AI-powered CMS built directly into the Litmos LMS—one that reflects a deep understanding of learning science, leverages existing knowledge assets, and empowers teams to dramatically speed up content creation without compromising on instructional quality.
Conquered Project Challenges
As with all big projects, obstacles arise. Litmos leaned on Tribe to address several challenges that emerged during this project.
Litmos has a large legacy codebase
Oftentimes companies believe applying AI initiatives requires a completely new, modernized tech stack. In actuality, a legacy system can create extra steps when implementing AI, but this is an obstacle that can definitely be overcome. Tribe’s team was able to identify and implement modern technology patterns that played well with Litmos’ existing system.
Much of Litmos’ LMS content is owned by their customers
In order to create the transcripts necessary for utilizing AI on Litmos’ 300,000+ LMS courses, Tribe worked with Litmos to develop a legal approval process for customers. Additionally, given their customer base, Tribe worked closely with Litmos to implement strict access controls for content, ensuring that licensed content is not accessed by incorrect parties.
Litmos’ content was in a combination of formats that required prep work for LLM use
As mentioned earlier, Litmos’ LMS contained content in a variety of formats. Audio and video files needed to be automatically transcribed in order to provide content for the LLMs. Much of Litmos’ content was housed in SCORM files which needed to be broken down into its core content to improve discoverability and make them play well with LLMs. Content in PDF, Word, native Litmos objects, and other formats needed to be ingested to provide a seamless experience to learners.
Thanks to Tribe’s extensive experience in AI and their existing toolkit, obstacles like these were more quickly and easily managed.
Tech Stack Details
Full stack cloud-based application working alongside the existing legacy Litmos environment:
- Cloud: Primarily Azure with trace amounts of Amazon Web Services (AWS)
- Large language models: Meta's Llama 3.1
- Other models: Whisper for speech-to-text
- Model serving: vLLM, BentoML on K8s
- Languages used: FastAPI backend in Python, React frontend in JavaScript
- Search: OpenSearch
Litmos’ Experience Working with Tribe
“Working with Tribe was one of the best decisions of my career,” said Richardson.
Tribe’s goal is to be a valued business partner helping clients reach their goals. When asked, Richardson expressed that Tribe had—to date—over delivered on this goal. Richardson also revealed three key areas the Tribe team provided an above-average experience for the Litmos team.
Rigorous Tech Comparisons & Cost Modeling
When Richardson’s team first began down the road of AI innovation they were open to considering a variety of approaches. Tribe’s team provided Litmos with the means to compare LLMs, evaluating the user experience and cost implications of each in order to select the right model for their needs. After considerable experimentation, Litmos decided to run the AI Assistant models directly in their own Azure environment to deal with their legal and compliance needs and to be able to keep a firm handle on costs.
Software Engineers with a Business Perspective
Unlike other tech partners Litmos had worked with, Tribe team members served as more than engineers. In addition to providing cost modeling and tech expertise, Tribe’s team members were core business partners, offering a sophisticated level of consulting on technical approach and product design in strategy team meetings.
“Tribe’s team members weren’t just order-takers aimed at pleasing us. They offered conflicting and compelling insights and really pushed the project forward with a critical perspective. That confidence to ‘argue’ for improved outcomes is invaluable,” says Richardson.
Functionality Demos for Less-Technical Stakeholders
Thanks to Tribe’s extensive experience in AI, engineers can demo functionality early on in the project so clients can quickly see the project's future state. This allows less-technical stakeholders to more easily engage with the project. It also makes gaining user perspectives simpler. For the Litmos project, Tribe was able to provide their first demo within two days of the start of the engagement. Richardson remarked how the Tribe team blew through deliverables and exceeded expectations regularly.
Tribe Team Members
- Nick: Engineering Lead
- Geoff: Product Lead
- Max: Product & Data Science
- Aditya: Engineer
- Justin: Engagement Manager
- Alex: Technical Lead
- Nik: Technical Lead
- Dana: Product & UX Design
- Caio: KIS Services & Engineer
- Gabriel: KIS Engineer
The Launch
Before the launch of these new solutions, the Litmos and Tribe teams administered a deep internal team testing. These tests focused primarily on functionality, performance, and security. As with any tech launch, the team made plans to closely monitor and administer small engineering tweaks to optimize functionality along the way.
Upon passing the internal test, the AI Assistant was rolled out to a small subset of customers where the teams had the opportunity to better understand how—and the frequency with which—customers used the tool. Although the goal was for customers to value and use the tool more than anticipated, Litmos knew there was a cost associated with that over-usage that they would need to plan for. Data collected from this customer subset also informed the go-to-market strategy to the thousands of remaining customers of Litmos.
“We want our customers to think ‘This tool is awesome. My people are more productive now.’,” says Richardson.
Project Impact
Litmos’ AI Assistant launched in May 2024 and AI-Powered Content Generation launched in January 2025. Richardson’s team has witnessed positive impacts from the work.
Increased Learner Productivity
With a more human-like interaction, the Litmos AI Assistant and the AI-Powered Content Generation allow users—from novice admins to expert instructional designers—to not only find the material they need, but also rapidly generate high-quality course modules using both uploaded content and Litmos’ existing library. Learners who can access training, validate their learnings, and ask questions more easily are able to perform their jobs with increased confidence and efficiency.
AI Value Creation for Litmos
With the development of robust AI products and a close partnership with the Tribe team, Litmos has not only successfully positioned itself as an AI company, but also turned a sales liability into a growth engine—adding $4M+ in ARR (and counting) by launching paid AI features adopted by 70% of new customers. AI-powered content generation also slashed course creation time from hours to minutes, enabling scalable training and reinforcing Litmos’ position as a leader in AI-powered corporate learning.
Further, their decision to house all of their content in their own RAG system, which they own and control, alleviates potential client concerns regarding privacy and solidifies their AI asset. Litmos has witnessed their AI solutions’ ability to create significant value for their customers and the company, as a whole, by driving higher valuations for any exit down the road.
The Future
These projects have laid the foundation for future AI innovation at Litmos. With both the AI Assistant and AI-Powered Content Generation tools now live, Richardson sees these advancements as key steps toward realizing a vision of “learning built for me.”
Looking ahead, Richardson believes AI will unlock even more personalized learning experiences—moving beyond one-size-fits-all courses to dynamically targeting learners based on their individual needs and skill gaps. Rather than assigning the same preset content to everyone, AI could help surface learning paths based on weak subject areas, role requirements, or work performance KPIs.
Future projects under consideration include leveraging AI to skill-tag LMS content at scale—reducing the manual burden on administrators—as well as adding command/response functionality and generating assessments on the fly to further drive adaptive learning.
