Train your engineers on AI. See who's learning.
Give every engineer a personalized AI learning path. Track who's making progress, who's stuck, and what's working -- across your entire team, in real time.
Every engineer gets their own AI-powered path.
Unfold creates a personalized learning plan for each person based on their role, stack, and timeline. Resources are automatically curated from official AI docs, GitHub, and curated video content -- no manual content curation.
Your entire cohort, one view.
Every metric that matters -- who's completing, who's stuck, which steps are failing, which tracks are thriving.
No spreadsheets. No surveys. No manual tracking. Every number here updates as your engineers learn.
The system notices things before you have to.
You don't need to analyze the data. Unfold surfaces the things that require your attention -- and tells you exactly what to do about them.
Step 4 is where engineers are giving up.
79% of your team completed Tool Use. Only 55% made it through Building Agents. That 24-point drop is the sharpest in the funnel -- and median time on that step is 2.1x higher than any other.
3 engineers are about to go quiet.
Alex, Sam, and Jordan haven't made progress in 7+ days. Based on their velocity on earlier steps, they won't recover without outreach. The window to re-engage is closing.
You can see who will complete -- on day 3.
Engineers who engage with resources in the first two steps are 3x more likely to complete the program. Right now, 8 of your 24 engineers haven't opened a single resource. Intervening now changes the outcome.
65% of your team is below target on RAG.
Pre-project assessments show that only 7 of 20 engineers meet medium proficiency in RAG & Retrieval. The weakest facets -- reranking, evaluation pipelines, and hallucination mitigation -- are exactly the skills the knowledge base agent project needs. Targeted plans are ready to generate.
Video resources cut step time by 40%.
Steps where engineers watch a video before starting take 40% less time than article-only steps. Steps 4, 5, and 6 -- your hardest steps -- have no video resources. That is fixable today.
From zero to a running AI learning program.
The full ecosystem -- from personalized plan creation to cohort analytics -- runs continuously, without manual input from your side.
Assess skills before building a single plan.
Before anyone starts learning, your AI agent generates skill assessments tailored to the work your team is about to do. Eight questions, validated by a semantic AI judge, anchored to the actual project context. In under 10 seconds you know exactly where each engineer stands -- and where the gaps are.
- MCQ assessments for any skill -- RAG, agents, prompt engineering, and more
- Proficiency bands (beginner, low, medium, high) with configurable thresholds
- Weak sub-skills identified per engineer -- plans focus only on gaps
Targeted plans, built from real data.
Assessment results flow directly into plan generation. Engineers who scored below target get a learning path that prioritizes their weak sub-skills and skips what they already know. Plans are anchored to the actual project, not a generic syllabus.
- Plans focus on weak facets from the assessment, skip strong ones
- Each plan is anchored to the work item context, not generic curriculum
- Metadata tags group engineers into cohorts and tracks for analytics
Engineers learn at their own pace, deeply supported.
Each engineer receives their own learning path. As they work through steps, the system surfaces curated resources -- official AI documentation, GitHub repos with exercises, vetted video tutorials -- matched to exactly where they are.
- Resources from Anthropic, Google AI, HuggingFace, and GitHub
- Your org can add its own licensed platforms and content
- Each step includes acceptance criteria to prevent checkbox completion
You see everything, in real time.
As your team learns, a complete picture builds. Completion rates by track, individual progress, stall points in the content, who hasn't started, who needs a nudge, which resources are actually being used -- all surfaced without any manual work.
- Cohort health score updates continuously
- At-risk learners flagged before they disengage
- Step-level funnel shows exactly where content is failing
The program gets smarter every cohort.
Compare cohort over cohort. See which content changes improved completion. Identify your fastest learners -- they become your internal AI mentors. Benchmark tracks against each other and against industry-standard timelines.
- Cohort-over-cohort improvement tracking
- Identify top performers for internal mentorship
- AI recommends curriculum adjustments based on funnel data
Drop Unfold into your stack
in an afternoon
Your portal stays yours. Unfold handles plans, learning, and analytics. One MCP call to enroll an engineer. Live progress data back on both sides.
Your portal calls Unfold MCP to assess and enroll
First generate_skill_assessment to measure gaps, then create_goal with the results. Assessment data drives plan personalization automatically.
Unfold AI generates a plan and returns a claim link
A structured, step-by-step learning path is built for that engineer. You get back a one-time claim link.
You send the link to your engineer however you already communicate
Email, Slack, your onboarding portal -- the link works anywhere. Engineer clicks, lands on Unfold with their plan ready.
Engineer learns on Unfold progress flows back to both sides
Your portal can query MCP for live progress. Your L&D team watches the cohort on the Unfold analytics dashboard.
generate_skill_assessment({ skill, target_proficiency, work_item_context }) → score → create_goal({ additional_context: assessment_results })→ returns claimLink with a targeted planYour engineers. AI-ready.
Create your organization, set up your first AI learning cohort, and have your entire team on personalized learning paths today. The analytics start flowing immediately.