AI-powered learning reinforcement platform for corporate training
A complete platform rebuild — from no-code prototype to scalable, AI-driven multi-tenant application — for a learning science company proving training ROI for the enterprise.
Challenge
elevator9 had validated a powerful concept: a learning reinforcement platform that deploys structured sequences of prompts to corporate learners before, during, and after live training events — measuring not just completion, but genuine cognitive progression from concept recognition through real-world application.
The problem? Their proof-of-concept was built on Bubble.io — a no-code tool that had served its purpose for early traction with enterprise clients like Franklin Covey and Galderma, but couldn't scale to meet the demands of a growing multi-tenant SaaS business.
The platform needed to handle significant complexity:
- A sophisticated multi-tenant architecture where each enterprise client operates in a secure, white-labeled environment with their own taxonomy of categories, learning journeys, users, and admin hierarchies
- A learning-science-based scheduling engine capable of orchestrating precisely-timed sequences of prompts around training events — managing overlapping journeys, multiple learning objectives per event, and dynamic rescheduling scenarios
- An AI-powered analytics pipeline that scores every learner response against the Community of Inquiry framework, generating real-time insights on cognitive presence, engagement, sentiment, and training ROI
- Six distinct user roles — from SuperAdmins managing the global platform, to Client Admins and Authors building learning journeys, to Learners and Sponsors interacting via a mobile-optimized experience
Solution
We partnered with elevator9 to design and build a custom multi-tenant web application from the ground up — replacing the Bubble.io prototype with a production-grade platform architected for scale, security, and AI-native analytics.
Platform Architecture & Multi-Tenancy
- A role-based access control system supporting six user roles with configurable permissions, allowing e9 to flexibly onboard enterprise clients into isolated, white-labeled tenant environments
- A hierarchical organizational model where each tenant manages their own Categories, Authors, Learners, and Sponsors — with SuperAdmins maintaining oversight across the entire platform
- Secure infrastructure on AWS with Auth0 authentication, ensuring data isolation between tenants while enabling cross-tenant reporting for leadership
LIFT Learning Journey Engine
- A full authoring experience for creating and managing LIFT Learning Journeys — structured sequences of touchpoints containing Retain, Apply, and Impact prompts designed around learning science principles like spaced repetition and elaboration
- A recommendation engine that generates suggested journey structures, schedules, and prompt copy based on author-entered information — reducing the expertise barrier for creating effective learning content
- Flexible response collection including text, video/audio capture, multiple choice, and file uploads — with a Sponsor accountability layer for tracking learner goals
- SMS and email notification workflows via Twilio/SendGrid integration, with configurable delivery schedules per journey
AI-Powered Analytics Pipeline
- A multi-layered GPT-powered analytics engine that scores every learner response on a Cognitive Presence scale — from Triggering Event through Exploration, Integration, and Resolution
- Three reporting tiers: executive-level journey summaries with cohort-wide engagement trends; per-LIFT drill-downs with quality narratives and completion stats; and individual learner deep dives showing personal progression versus cohort averages
- AI-generated quality summaries, sentiment scoring, and engagement scoring — with a chained architecture where later-stage calls synthesize prior analysis rather than re-processing raw data
- A confidence threshold system where AI self-reports scoring confidence, defaulting to "Unclear" when below 70% — ensuring stakeholders only see insights the system can back with evidence
Impact
The rebuild transformed elevator9 from a promising prototype into an enterprise-ready platform — giving their team and their clients the tools to demonstrate, for the first time at scale, that their training programs are driving real behavioral change.
- From no-code to production-grade: Replaced a Bubble.io application with a scalable, secure multi-tenant architecture on AWS — enabling elevator9 to onboard enterprise clients with confidence and operate white-labeled environments for each
- AI-native analytics at scale: Delivered a first-of-its-kind analytics pipeline that translates freeform learner responses into structured cognitive progression insights — giving L&D leaders evidence-based ROI data they couldn't get from any other tool
- Reduced authoring friction: The recommendation engine and structured prompt templates lowered the expertise barrier for creating learning journeys — addressing what had been the single biggest bottleneck in client onboarding
- Enterprise client traction: The platform now powers learning reinforcement programs for major enterprise clients, with a roadmap spanning multi-lingual support, cross-tenant reporting, AI-powered prompt generation, and advanced scheduler intelligence
Let's build something like this.
Tell us what you're working on. We'll let you know how we can help.
Start the conversation