An AI-powered learning service for neurodivergent children — and the business built around getting it to the families who need it most.
The Pitch
What Bloom is, why it exists, and why it matters — before the full story.
The problem in one sentence
Neurodivergent children are failed systematically — by a school system built for neurotypical learners and by EdTech that treats accessibility as an afterthought. Their families are left to fill the gap with expensive, scarce, and inconsistent specialist support.
The solution in one sentence
An AI learning service that adapts not just to how a child learns, but to how they feel — built from the ground up for neurodivergent minds, with their safety and dignity as non-negotiable constraints.
What makes it different
The Problem
The scale of unmet need is not marginal. It is structural. Neurodivergent children are not failing — they are being failed by systems not designed with them in mind.
The early years matter: EPI (2025) links low social and emotional scores at age five to later EHCPs, which makes early identification and support worth getting right. Yet few EdTech products serve this group — parents of 2–5 year olds with suspected neurodivergence often have the least information and support.
Bloom is uniquely positioned to serve them — not with AI tutoring (the child isn't school-age yet), but with parent guidance, diagnosis navigation, play-based learning support, and school transition toolkits. This creates a powerful acquisition funnel: parents find Bloom when their child is 3. When that child turns 6, Bloom is already trusted.
Illustrative voices, composited from what parents and teachers commonly describe. Not quotes from Bloom research interviews.
The Solution
Not an accessibility overlay on existing EdTech. A ground-up system built around how neurodivergent children actually learn, feel, and thrive.
Every child who uses Bloom has a continuously-updated model of how they learn. Not a diagnosis label. Not a learner type assigned at onboarding and never updated. A living portrait — attention window, processing speed, emotional baseline, modality preferences, frustration threshold, preferred break type — updated after every session.
The AI tutor reads this model before every interaction. It knows that this child does better with visual diagrams than text. That 4–6pm is their best window. That three consecutive errors is the threshold before intervention. That on Mondays their regulation score is typically 18 points lower than average.
The child doesn't experience a generic product personalised with their name. They experience a service that has learned them — and keeps learning.
| Capability | bloom· | Khan Academy / Khanmigo | Lexia / Reading Plus | Eedi / Sparx | SENECA |
|---|---|---|---|---|---|
| Built for neurodivergent learners | ✓ Primary target | ✗ | ~ Dyslexia only | ✗ | ✗ |
| Emotion / regulation detection | ✓ Every 30s | ✗ | ✗ | ✗ | ✗ |
| Child model (persistent, per-child) | ✓ Full model | ~ Progress only | ~ Reading level | ~ Mastery only | ~ Topic tracking |
| Safeguarding infrastructure | ✓ Multi-layer + human | ✗ | ✗ | ✗ | ✗ |
| Parent digest + EHCP reporting | ✓ Weekly + on demand | ✗ | ~ Progress report | ~ Basic report | ✗ |
| No streaks / no leaderboards | ✓ By design | ✗ Streaks present | ✓ | ✗ XP and leaderboards | ✗ |
| Child data sovereignty | ✓ Self-hosted AI | ✗ OpenAI backend | ~ Limited | ~ Limited | ✗ |
Bloom's own assessment of publicly described features, October 2026. Competitor products change; Bloom's column describes its design, not features already built.
Why Now
Four structural forces have converged to make this the right moment to build Bloom. None of them existed at this scale five years ago.
ADHD diagnoses in the UK have risen significantly over two decades (UCL, 2023). 227,813 autism assessment referrals are open in England, with an average wait of over 16 months. The population Bloom serves is growing rapidly — and the supply of specialist support is not keeping pace. This is a structural imbalance, not a trend.
The LLM capabilities that make Bloom's tutor AI possible — contextual understanding, warm consistent persona, adaptive explanation, emotional sensitivity — only became available as purpose-built tutors in 2023. The window to build a genuinely differentiated AI learning service is open now. It won't stay open.
Families of children with complex needs already spend an average of £8,500 a year of their own money on support (Sense, 2026). Spending is proven; willingness to pay for Bloom is what the beta will test. What is missing is a product worth paying for — one that actually understands their child. Bloom at £14/month is a fraction of what they are already spending.
The DfE SEND review, the Ofsted SEND framework update, and the ICB digital health agenda are all creating institutional demand for technology-enabled SEN support. Schools, local authorities, and NHS commissioners are actively seeking solutions. The regulatory tailwind is real.
The pandemic EdTech wave created both opportunity and scepticism. Parents have now seen enough bad EdTech to know what it looks like — and enough good technology to know what's possible. The bar is higher. Products that genuinely work will win disproportionate trust and recommendation.
Academic understanding of neurodivergent learning — polyvagal theory, UDL frameworks, mastery-based progression, strength-based approaches — has reached a point where it can be systematically implemented in software. Bloom is not ahead of the science. The science is ready to be built.
Market
A large, growing, underserved market with three distinct customer segments — each with real willingness to pay and a clear value proposition.
UK first — the right starting point
Bloom launches exclusively in the UK. This is a strategic decision, not a limitation. The SEND Code of Practice, 718,000 EHC plans and the 2026 SEND reforms create clear demand for rigorous, evidenced support — exactly what Bloom is designed to provide.
UK SEN parent communities are active and well connected. Early traction here creates a replicable playbook and a clinical evidence base that opens every other market.
US expansion — built in from the start
The US has its own compelling case: IDEA serves 8.2m children and young people. IDEA (Individuals with Disabilities Education Act) gives children with disabilities a legal right to a free appropriate public education
The AI architecture, pedagogical frameworks, and experience design transfer entirely. US expansion requires curriculum localisation (to each state's standards), IEP-format reporting and COPPA parental consent — meaningful but bounded work, planned from Phase 3.
Competitive Moats
These are not features. They are structural advantages that compound over time — and that a well-funded competitor cannot simply replicate by throwing engineers at the problem.
The child model becomes more accurate with every session. After 10,000 sessions, Bloom will understand neurodivergent learning in a way that no new entrant — however well-funded — can match without years of data. The moat compounds continuously and silently.
The curriculum is not neurotypical content with accessibility overlays. It is rebuilt from learning objectives outward, for neurodivergent cognition as the primary design target. Replicating this requires deep pedagogical expertise, not just engineering effort.
Trust in the SEN parent community is earned slowly and lost instantly. A product that genuinely works — that parents recommend to each other — creates word-of-mouth growth that no paid acquisition can replicate. Emotional switching cost is very high once established.
Most EdTech is not willing to build robust safeguarding infrastructure — the liability, the complexity, the human review requirement. Bloom builds it properly from day one. This is a genuine barrier to entry for any competitor building quickly.
A Bloom-original subject that exists in no national curriculum anywhere in the world. Children learning about their own neurodivergent minds is not a feature — it is a category. It cannot be quickly copied because the content must be developed with specialist clinical input.
The SEN parent community is exceptionally connected and vocal. A recommendation from a trusted parent carries more weight than any marketing campaign. Every delighted family is a distribution channel. Every EHCP report shared with a SENCO is a school sales conversation starting.
The Founder
The combination of skills that makes Bloom possible — and honest about what is still needed.
Design Director-level expertise across UX design, design thinking, product strategy, and digital design. The ability to hold a complex product vision and translate it into precise, buildable specifications — the skill most technical founding teams lack entirely.
Before any build spend, Bloom has its user flows mapped, an AI architecture written up, a six-subject curriculum framework, a safeguarding policy framework and a business case. The next step is turning that plan into a tested prototype.
A technical co-founder who owns the engineering vision the way I own the product vision. Someone who has shipped a real consumer product, understands why child data privacy is non-negotiable, and is personally connected to the mission — not just the opportunity.
When the right person joins, they are not starting from a blank page. The vision, user flows and AI architecture are mapped — together we turn them into a working system.
Full Documents
Five documents that together tell the full story — from consumer-facing concept through to the founder strategy that gets this built.
If you're a technical founder who has read this far and feels the pull of the problem — this is the conversation to have. The vision is mapped and the plan is written. What's missing is you.