EdTech,
engineered as a product.
Multi-tenant learning platforms, content delivery, assessment, AI tutoring and the integration surface institutions demand, engineered for a company whose roadmap and runway are both finite.
What an EdTech platform is judged on.
An EdTech product is judged by its buyers on security review and integration depth, and by its users on whether learning actually happens. Both have to hold.
- Term-start
- Load profile
- LTI + SCORM
- Table stakes
- 6–10
- Weeks
- SOC 2 ready
- Engineered
Traffic that is near flat for months, then vertical for a week
Institutional integration standards that gate procurement
To first release on a scoped product or platform slice
Security posture built for the review that decides the deal
The EdTech models we build for.
Selling to a school district, to a university, to an employer or direct to a learner produces four different products. Architecture follows the motion you are actually running.
K-12 platforms
District and school sales, rostering, safeguarding duties and procurement cycles tied to the academic year.
Higher-education platforms
LMS adjacency, LTI depth, faculty adoption and the long institutional sales cycle that comes with them.
Consumer learning
Direct-to-learner products where acquisition, activation and completion economics govern the engineering priorities.
Corporate & workforce learning
Employer-bought platforms where the buyer, the learner and the beneficiary are three different people.
Assessment & credentialing
High-stakes assessment, proctoring, item banking and credential issuance, where integrity is the product.
Content & courseware
Rights-managed content platforms, adaptive courseware and distribution engineered as a product rather than a library.
Services we deliver into EdTech companies.
EdTech engagements are product engineering with an unusually demanding buyer. These are the practices we most often deploy, each linked to its dedicated page.
The EdTech platform we engineer end to end.
The EdTech companies that scale cleanly treat tenancy, integration and efficacy measurement as architecture decisions made early, not as things to be fixed once enterprise deals arrive.
- 01Multi-tenant platform core
- Tenancy, isolation, entitlement and seat management designed before the first enterprise contract rather than after it.
- 02Learning experience
- Course delivery, pathways, practice, feedback loops and the surfaces that decide whether a learner comes back tomorrow.
- 03Assessment & integrity
- Item banking, adaptive assessment, proctoring and academic-integrity controls proportionate to the stakes involved.
- 04Institutional integration
- LTI, rostering, SSO, gradebook write-back and the standards that quietly decide procurement outcomes.
- 05AI in the product
- Tutoring, feedback and authoring features with grounding, evaluation and cost control engineered as product concerns.
- 06Efficacy & analytics
- Engagement and outcome measurement your product team can act on and your customers can put in a board paper.
Programmes delivered in this sector.
What clients ask us first
01We are pre-Series A. Are we too early to work with you?
Not necessarily, but the engagement looks different. Early-stage work is about finding the smallest architecture that supports the next eighteen months, not the most complete one. We will push back on multi-region, microservices and anything else that spends runway buying optionality you do not yet need. What we do insist on early is tenancy and data modelling, because those are the two decisions that are genuinely expensive to revisit.
02Our enterprise deals keep stalling in security review. Can you fix that?
Usually, yes, and it is rarely one blocking issue. It is typically some mix of tenancy isolation that cannot be evidenced, missing audit trails, SSO gaps, unclear data retention and no coherent story about subprocessors. We assess against the questionnaires your buyers actually send, fix the engineering gaps, and produce the evidence rather than the assertions. The goal is clearing review without a six-week scramble each time.
03How much LTI and rostering support do we actually need?
More than feels reasonable, if you sell to institutions. LTI 1.3 with deep linking and gradebook write-back, plus rostering through OneRoster or a direct SIS integration, is the practical floor for most higher-ed and district buyers. The specifics depend on which LMS your buyers run. We will scope it against your actual pipeline rather than building the whole standard speculatively.
04How do you keep AI feature costs under control?
By treating inference as a unit-economics problem from the start. That means routing to the smallest model that holds quality for each task, caching aggressively where content is stable, grounding against your own material rather than paying for the model to reason from scratch, and instrumenting cost per active learner as a tracked metric. An AI tutoring feature that is wonderful in a demo and ruinous at ten thousand learners is a product failure, not a pricing problem.
05Can you prove efficacy, or just engagement?
We can engineer the measurement honestly, which is the part that is usually missing. That means instrumenting learning events properly, designing comparisons that are not self-selecting, and reporting outcomes the way a sceptical institutional buyer would want to see them. We will not help construct a metric that flatters the product, partly on principle and partly because education buyers are unusually good at spotting it.
Let's discuss your EdTech product.
Tell us where the product is today, who you sell to and what is in the way, whether that is scale, procurement, integration depth or the AI roadmap. We will arrange a conversation with the people who would build it.