Skip to content
Sun Software logo with the tagline Engineering Beyond Code
Education & E-learning

Education app development for the learner who gets stuck.

Learners sign up for your course, then stop after the first week. We're an education app development company, so we build the learning app or platform, the AI features that help a learner who is stuck, and the website that sells the course.

What the research says

What the research says about online learners.

Published studies and a US regulator's case show three things. Many learners never start. AI can help or harm, depending on how it is built. And learner data is a real responsibility.

  • of people who registered for a Harvard or MIT online course on edX, from 2012 to 2018, never opened the course material

    52%

    What we do

    We walk through your course as a new learner and build a first week that shows what to do next.

  • lower grades for high school maths students who practised with a plain AI chatbot and then lost it, compared with students who never had it

    17%

    What we do

    We build guided practice that gives hints before answers, with a tutor in the loop where a mistake would matter.

  • the median learning gain for Harvard physics students taught by a purpose-built AI tutor, compared with an active lesson in class

    Over 2x

    What we do

    We build AI tutoring that works from your own lessons and worked answers, and your tutors test it on real questions.

  • students had personal data exposed in one breach at a US education technology provider, according to the FTC's complaint

    10.1M

    What we do

    We agree with you before we build what learner data the product needs, who may see it and when it is deleted.

The problem

Where do your learners drop off?

Most course businesses do not lose learners because the teaching is bad. They lose them in the gaps around it. You may recognise some of these.

01

People pay or sign up, then never open the first lesson

02

Learners get stuck and nobody answers until the next day

03

Feedback on written work takes your tutors too long

04

Your course lives on a rented platform you can't change

05

An AI feature gives a wrong answer and you can't see why

06

The sales page does not explain who the course is for

One plan, several services

Who does what on your project?

A course business needs three things to work together: a page that sells the course, a product people learn in, and help for the learner who gets stuck. We plan them as one project, and each part of our team takes the piece it knows best.

  1. 1

    Web, Ecommerce & Digital Platforms

    Your course website, enrolment pages and payments.

  2. 2

    AI Agents & Business Automation

    AI feedback, chatbots, recommendations and the reminders that bring learners back.

  3. 3

    Custom Software & SaaS Development

    Your learning app, course platform and tutor or admin area.

  4. 4

    SEO, AI Visibility & Digital Growth

    Course pages and content built around what future learners search for.

First conversation

What we ask before we propose anything

Three questions show us what to build first.

01

Who are your learners, how old are they, and what do they do in the first week?

02

Where does a learner get stuck today, and who helps them?

03

What should the AI never do, such as hand over a full answer or mark work without a tutor seeing it?

How it works

How we build your education app or platform

We start with one learner and one lesson, not a feature list. Then we build outward from what that learner needs.

  1. 01

    Follow one learner

    We walk through your course as a new learner, from the sales page to the last lesson, and mark where people stop.

    What you get

    A map of the learner's path with the drop-off points marked.

  2. 02

    Agree what to build first

    We decide together which screens, lessons and AI features go into the first version, and what learner data each one needs.

    What you get

    A written scope and a list of the data the product will hold.

  3. 03

    Build the learning product

    We build the app or platform in stages. You see working lessons, progress and payments on a private test copy as we go.

    What you get

    A working product you can click through before launch.

  4. 04

    Test the AI on your material

    You and your tutors try the AI feedback, chatbot or recommender on real course questions. We fix what it gets wrong before learners see it.

    What you get

    AI features checked against your own lessons and answers.

  5. 05

    Launch and watch the numbers

    We launch, show your team how to run it, and set up the figures to watch: who starts, who continues and where they stop.

    What you get

    A live product and a clear view of learner progress.

What you get

What changes for your course business?

Here is what you and your tutors should notice. Exactly what you get depends on what we agree to build.

What we show as proof

We've built apps with an AI recommender, AI essay feedback, an AI chatbot and AI-made language courses. Those clients are private, so we do not name them or quote their numbers. Where a benefit has not been measured, we call it intended, not proven.

  • A clearer first week. New learners see what to do next from the moment they sign up.

  • Help when a learner is stuck. Hints, feedback and answers drawn from your own course material.

  • Tutor time spent on teaching. AI drafts the first round of feedback, and your tutors decide what goes to the learner.

  • A product you own and can change. Your lessons, your rules and your learner records, in a product built for you.

  • Progress you can see. You see who starts, who continues and where people stop, so you know what to fix.

Public cases

Three public cases worth learning from

One went right and two went wrong. None of them is our client. Each is reported by the source we link to.

  • Published study · 2025

    Harvard gave its AI tutor the worked answers first

    What happened

    Researchers at Harvard built an AI tutor for a physics course and tested it against an active lesson in class. They did not leave the AI to work out the answers alone. They wrote step-by-step answers into its instructions and made it take students through each problem one part at a time. The students learned more, in less time.

    What we do about it

    When we build AI tutoring or feedback, it works from your own lessons and worked answers. You and your tutors test it on real questions before learners see it.

    Source: Scientific Reports
  • News report · 2024

    A school district's AI chatbot went dark with its supplier

    What happened

    Los Angeles Unified School District launched Ed, an AI chatbot for students and parents, in March 2024 under a $6 million contract with a company called AllHere. In June, AllHere furloughed most of its staff and the district unplugged the chatbot. A former AllHere engineer also said the way it handled student records put them at risk.

    What we do about it

    We agree at the start who holds the code, the accounts and the learner data, and we show your team how to run the product. Your course should not stop because a supplier does.

    Source: The 74
  • Regulator action · 2022

    Chegg exposed the details of about 40 million customers

    What happened

    The US Federal Trade Commission took action against the study-help company Chegg after four data breaches between 2017 and 2019. In one, a former contractor used a login shared among staff to reach a database with the personal information of about 40 million customers. The FTC said Chegg had stored personal data in plain text and had no written security policy until 2021.

    What we do about it

    Before we build, we agree with you what learner data the product needs, who may see it and when it is deleted. We collect only what the course needs.

    Source: US Federal Trade Commission
FAQ

Questions from course and learning businesses

Will AI features help my learners finish the course?

They can help, but we don't promise a completion rate. Published studies show AI built to guide learners can help, and a plain chatbot that hands over answers can hurt. What works for you depends on your subject, your learners and how the AI is set up.

What stops the AI from giving a wrong answer?

Nothing removes the risk completely. We reduce it by having the AI answer from your approved lessons and worked answers, testing it on real course questions with you, and keeping a tutor in the loop where a mistake would matter.

Should we build a custom e-learning platform or use an existing course tool?

It depends on how unusual your teaching is. If a ready-made course tool does what you need, we'll say so. Your own platform makes sense when you need features, pricing or AI the tool can't give you.

How do you handle learner data, especially for children?

We agree with you before we build which rules apply to your learners and what data the product really needs. We then build to what the contract sets out. We do not give legal advice, so the decision on which rules apply should come from you or your adviser.

We already built a prototype with an AI tool. Can you use it?

Usually, yes. We review the code first and tell you what can stay, what needs fixing and what should be rebuilt before real learners use it.

  1. 1Talk Through My Course
  2. 2We reply within one business day
  3. 3Discovery call
  4. 4Step one: Follow one learner

Tell us what you teach and where learners stop.

If learners sign up and then drift away, tell us how your course runs today. We'll talk through what to build first and what it should do for the learner.

  • The first conversation is free
  • NDA before you share anything
  • A reply within one business day