How an Education Portal Built Personalized Training Plans for Every Student Using Unfold

The Problem: Personalization That Does Not Scale

Every educator knows the dilemma. You want every student to have a training plan tailored to their skill level, their career track, and their learning pace. But when you have 200 students across five different tracks, "personalized" quickly becomes "impossible."

We tried the obvious approaches. Spreadsheets. Templated PDFs. LMS platforms with custom fields. Each one hit the same wall: the manual effort of creating, distributing, and tracking individual plans consumed more time than the training itself.

We needed a system that could generate structured, multi-step training plans automatically, customize each one based on a student's track, distribute them individually, and let us monitor progress from a single dashboard.

What Unfold Brings to the Table

Unfold is not a traditional LMS. It is a goal decomposition platform. You describe what someone needs to achieve, and the AI breaks it down into clear, actionable steps with substeps, resources, and milestones.

How We Set It Up

Step 1: Create the Organization Workspace

We started by creating an Unfold organization for our academy. This gave us a shared workspace with admin visibility, org-scoped API keys for programmatic access, and the ability to generate claim links -- unique URLs that assign a pre-built goal to whoever clicks them.

Step 2: Design the Training Templates

For each learning track (frontend, backend, data science, DevOps, product management), we created a template goal in Unfold. Each template described the full training arc:

Unfold's AI decomposed each template into 8-12 structured steps, each with a clear objective, substeps breaking it into daily or weekly actions, suggested resources, and an estimated time commitment. We reviewed, refined, and marked them as templates.

Step 3: Generate Claim Links via the API

This is where the automation kicked in. Using Unfold's REST API, we wrote a script that reads our student roster, creates a goal from the matching template for each student, generates a unique claim link, and emails it.

POST /api/v1/org/goals
{
  "title": "Frontend Engineering Training - Sarah M.",
  "template_id": "tmpl_frontend_2026",
  "claim_link": true,
  "metadata": { "cohort": "spring-2026", "track": "frontend" }
}

The response includes a claim_url that the student clicks to accept the goal into their Unfold workspace. No account setup friction. Just click and start.

Step 4: The MCP Integration

Our platform already had a conversational AI assistant for student support. Using Unfold's MCP (Model Context Protocol) server, we connected the assistant directly to Unfold. Now when a student asks "What should I work on this week?", the assistant can look up their Unfold goal, check which step they are on, and give a contextual, plan-aware answer.

Step 5: Connect the Resource Ecosystem

The biggest upgrade came when we connected our licensed learning platforms to Unfold's resource routing. For each track, we configured category-aware resource discovery:

• Our Udemy Business catalog surfaced relevant courses alongside free resources
• Our curated YouTube playlists (50+ instructor-recorded walkthroughs) appeared as priority resources in each step
• Official documentation (React docs, AWS docs, Python docs) was auto-prioritized over random blog posts

When a student opens a step on "React Component Architecture," they see the official React docs, our instructor's walkthrough video, and a GitHub starter repo with exercises -- not a random Medium article from 2019.

What the Admin Dashboard Shows

This is the part that changed how we run the program. Before Unfold, we had a spreadsheet with checkboxes. Now we have real-time visibility into every dimension that matters.

Cohort Overview

Students at risk are flagged automatically when they have not made progress in 7+ days. We used to discover stalled students only when they missed a deadline. Now we intervene early, often with a single message from the AI assistant that says "You have been on Step 3 for 8 days. Want to break it into smaller pieces?"

Completion Rate by Track

Not every track performs equally, and knowing where students struggle tells us where to invest.

Backend Development sits at 65%. When we drilled into the step-level data, we found that students were stalling on "Database Design and Optimization" -- the step had no hands-on exercises, just documentation links. We added a GitHub starter repo with practice schemas and the completion rate for that step jumped from 48% to 74% in two weeks.

Step-Level Drop-off Funnel

The most actionable metric. For any track, you can see exactly where students lose momentum.

The steep drop between "React Components" and "State Management" was invisible before. Now it is obvious, and we know exactly where to add more scaffolding, more resources, or more instructor check-ins.

Resource Engagement

This is where Unfold's resource routing pays off. We can see what students actually use -- not just what we assign them.

The video preference signal was strong enough that we shifted our content strategy. Instead of writing more documentation, we recorded 20-minute walkthrough videos for the highest-friction steps. Student velocity on those steps improved by 28%.

Additional KPIs We Track

Beyond the dashboards above, Unfold gives us granular telemetry:

• Time-to-first-action: how quickly students start after claiming their plan (median: 14 minutes)
• Claim link activation rate: 94% of sent links are claimed within 48 hours
• MCP assistant usage: students who use the AI assistant complete plans 1.6x faster
• Substep completion granularity: which specific substeps are skipped vs completed, revealing curriculum gaps
• Resource type effectiveness: GitHub exercises correlate with 23% higher step completion vs documentation-only steps
• Cohort comparison: Spring 2026 cohort is tracking 18% ahead of Winter 2025 at the same point in the program

What Changed

For students

• No more vague syllabi. Every student sees exactly what to do next.
• AI substep decomposition matches realistic work sessions (30-60 minutes each).
• Resources come from their org's licensed platforms and curated libraries, not random search results.
• Progress is visible. They mark steps complete and see their trajectory.
• The AI assistant gives plan-aware guidance, not generic answers.

For administrators

• Onboarding a new cohort went from 2 weeks of manual work to a single API script.
• Stalled students are flagged in real time, not discovered at deadline.
• Step-level analytics reveal exactly where curriculum needs work.
• Resource engagement data drives content strategy decisions.
• Template updates propagate to future cohorts automatically.

Lessons Learned

Let the AI do the first draft, but review it. Unfold's AI-generated plans are solid starting points, but domain experts should review step order, resource recommendations, and time estimates. The AI does not know your specific curriculum constraints.

Claim links reduce friction dramatically. The simpler the student's first interaction, the higher the adoption. A single click beats a signup form plus onboarding wizard every time.

The MCP integration is worth the setup. If you already have a conversational AI, connecting it to Unfold via MCP turns generic chatbot responses into plan-aware coaching. The difference in student engagement is noticeable.

Let the data reshape your content. The biggest insight was not the completion rates -- it was the resource engagement data. Knowing that students prefer video over text, that GitHub exercises drive higher completion, and that certain steps are bottlenecks changed how we build curriculum. The analytics are not just reporting; they are the feedback loop.

What if You Could Assess Before You Plan?

Everything above assumes you already know what each student needs. But what if you could measure it first?

Unfold now supports AI-powered skill assessments. Before creating a learning path, you can generate MCQs for any skill, score the results, and use the gap data to create plans that focus on exactly the right areas -- skipping what the learner already knows.

Read the full walkthrough: From Assessment to Action: How Skill Assessments Create Targeted Learning Paths.

Build This for Your Academy

If you run a training program, bootcamp, or corporate learning platform, you can set this up with Unfold's existing API and MCP server.

• The Developers page has the MCP demo and REST API reference
• The Org API supports claim links, goal templates, and progress tracking
• The MCP server is open-source: github.com/Unfold-it/unfoldit-mcp-server