Everything an investor needs after the deck: what the research shows, what has been designed, how Bloom makes money, how the round is allocated, and where it grows next.
One in five pupils in England has special educational needs. Families are already spending heavily to fill the gap — on support that is scarce, expensive and inconsistent.
The people Bloom serves
The parent is the customer — they pay, they decide, and in the SEN community they recommend. The child, aged 6–16, is the user whose trust decides everything: if they feel safe, they come back. SENCOs and schools come next, under pressure from rising SEND numbers on flat budgets. And the Early Years parent — noticing something different in a two-to-five-year-old — is often the least supported family of all, and the front door to Bloom.
The closest alternatives
Built for dyslexia, with real evidence. Mainly literacy (Reading and Spelling is for ages 6–11), with no emotional or parent layer.
The closest consumer maths comparison. Genuinely good and adaptive to pace — but not designed around anxiety, regulation or neurodivergent profiles.
Free and global; its Khanmigo AI tutor costs parents $4 a month. Built for the mainstream, not for SEND, with no regulation tools.
Go deeper → Product story · Business case · Competitors · Sources
Every figure is checked and linked in the Bloom Sources & Evidence Register, October 2026.
A clickable iPad prototype covers the whole first experience: a parent sets Bloom up, the child meets their companion, chooses something to learn, stays regulated, and the parent sees how it went. Every design decision below is traced to the need it serves and the evidence behind it.
Four children, four journeys
The prototype ships with four illustrative children. They are fictional, but the combinations are common. Three are in the Early Years and one is in Key Stage 2. The same session reshapes itself for each of them.
ADHD (on a waiting list) and worries
⚽ football
Autistic, with a PDA profile
🚀 space
Motor and sensory differences, still being explored
🦕 dinosaurs
Dyslexia, dyscalculia and worries
🎵 music
The journey, step by step
Twenty decisions, from the parent’s first screen to the weekly summary. Each “Try it” link opens the prototype at that exact moment, with the right child loaded.
Data use is explained before anything is asked. Research sharing is off by default; safeguarding is always on.
Parents are handing over sensitive information about a child.
UK Data (Use and Access) Act 2025 — higher protection for children’s data; ICO Children’s Code — high privacy by default.
“Not sure yet” and “on a waiting list” are full answers. Bloom adapts to what the parent has noticed.
Families wait years for assessment and need help now.
National Autistic Society (Nov 2025): 227,813 open autism referrals, average wait over 16 months. NHS England ADHD Taskforce (2025): 40% report waits of two years or more.
The parent chooses up to three interests. Every example is built around them.
Children engage with what they already love.
Gunn & Delafield-Butt (2016), review: using autistic children’s special interests in teaching improves engagement and learning.
Every change Bloom makes is listed with the reason and its source, before the child sees anything.
Parents need to see why, not just what.
BDA Dyslexia Style Guide; EEF, SEND in Mainstream Schools (2020); PDA Society; NCETM; CAST Universal Design for Learning.
The same opening ritual every time. The companion grows when the child turns up and tries, and never shrinks. No points, streaks or scores.
A safe, steady relationship before any learning.
Self-Determination Theory (Ryan & Deci, 2000) — relatedness and competence. Deci, Koestner & Ryan (1999), meta-analysis: tangible rewards can undermine intrinsic motivation.
Subjects are offered, not assigned, along with the calm corner and the journal.
Agency lowers demand and anxiety.
Self-Determination Theory — autonomy. PDA Society — low-demand, collaborative approaches.
Before any learning, the child picks an animal that matches how they feel. A low answer means a shorter, gentler plan.
Many children can’t yet put feelings into words.
Kinnaird, Stewart & Tchanturia (2019), meta-analysis: around half of autistic people have alexithymia (difficulty identifying and describing feelings).
Today’s steps are shown before starting and can be changed without giving a reason. Music is chosen here too.
Surprises are a common trigger.
National Autistic Society guidance on predictability and visual structure.
4–6: counting by touching, stories read aloud, tracing. 7–11: fractions, reading for inference, spelling rules, science.
Content must fit development, not just diagnosis.
EYFS statutory framework — Early Learning Goals; National Curriculum KS2; DfE Reading Framework (2023).
Ruby colours parts of a bar before naming fractions, and pictures stay available at the number stage.
Dyscalculia makes abstract symbols hard to hold.
NCETM Teaching for Mastery — representation and structure (concrete and pictorial before abstract).
No red crosses and no “wrong”. Hints come step by step. After two tricky answers in a row, a break is offered.
Maths anxiety and low performance feed each other.
Cambridge / Nuffield Foundation (2019), 1,700 pupils: the anxiety–performance cycle.
Lexend font, wider spacing, an off-white background and “Read to me” on every screen.
Dense text blocks access to everything else.
BDA Dyslexia Style Guide; CAST UDL — multiple means of representation.
Large tap areas, a wide and forgiving tracing path, and nothing to type.
Fine-motor demand shouldn’t gate learning.
EYFS — Physical Development: fine motor control as a foundation for handwriting.
The break opens with the child’s own calming strategy first. “Stop for today” is always allowed. ADHD profiles are offered a movement break halfway.
Learning pauses for the human.
EEF, Improving Behaviour in Schools (2019) — teach and support self-regulation.
Calm, focus or rain sounds, generated on the device. Controls stay on the learning screen; a parent can suggest a default; it never starts by itself.
Some children focus better with background sound; others don’t.
Söderlund, Sikström & Smart (2007): background noise helped inattentive children and hindered attentive ones. The evidence is mixed, so it is always the child’s choice.
Balloon, five-finger and (7+) square breathing, paced on screen with counts and an optional voice.
A concrete tool to come back down.
Zaccaro et al. (2018), systematic review: slow breathing (under ten breaths a minute) is linked with calmer physiology and mood.
Praise names what the child did, such as looking again or taking a break. The child chooses a face and a word for their journal. No typing.
Praise for effort protects motivation after setbacks.
Mueller & Dweck (1998): praising intelligence undermined persistence after failure; praising effort did not.
Every session ends the same way. The child decides what happens next: show a grown-up, explore more, or finish with a goodbye.
Predictable endings make starting again easier.
National Autistic Society guidance on transitions.
How the session went, one thing to celebrate, one pattern, a tea-time conversation starter, and a switch to flag a hard morning.
Parents carry the load between appointments.
Sense / Censuswide (2026), 1,000 parents of children with complex needs: average £8,500 a year of their own money on support; 35% have left work.
Hosted on Bloom’s own UK/EU servers, pseudonymised, with no third-party AI. Download or delete everything at any time.
Trust is the product.
Data (Use and Access) Act 2025; 5Rights Foundation — EdTech code of practice (2025).
It shows how Bloom behaves. It is not evidence that Bloom works. Adaptation is rules-based, as planned for the v1 beta. The lesson content is example material written for the prototype; the full curriculum still needs authoring and review by specialists, starting with the first 40 maths concepts. The research supports the design principles, not Bloom’s outcomes. Where the evidence is mixed, as with background music, the choice stays with the child. Testing these journeys with families is the closed beta’s first job.
The wider suite → User flows · AI & curriculum · Specialists · Sources
Families of children with complex needs already spend an average of £8,500 a year on support. At £14 a month, Bloom is not a stretch — it's a bargain if it works.
Three revenue streams · parent-direct first
Seed round · technical co-founder + closed beta
The vision, product flows, AI architecture and business case are mapped out. This round turns them into a working product: the first maths curriculum, a tested prototype, the build, and a safe closed beta with 20 families.
Revenue at each family milestone
Annual recurring revenue from parent subscriptions only — schools and NHS sit on top:
Via community and word of mouth
Family plan, 18-month average
Payback under 3 months
Target at scale
Pricing, unit economics and family targets are pre-launch assumptions from the Bloom business case — CAC and retention will be measured in the Phase 1 closed beta. The round size is still to be finalised.
A bottom-up budget built from the roles and phases in the Bloom team plan. The founder’s own salary is shown on its own line, separate from the cost of building Bloom. Everyone on a below-market salary is also offered equity, so cash and ownership are shown together.
The founder
Building Bloom £420k · 84%
Equity, alongside the cash
Below-market salaries only work if the people taking them own part of what they build. Each group’s cash and equity are shown together, with ranges taken from the Bloom team plan.
If one funder fronts the full £500k
The cash split stays as above. What changes is who owns Bloom afterwards. That depends on the valuation agreed before the money goes in (the pre-money). Three cases are shown. The middle one is the UK median for seed-stage deals in the first half of 2026, which Beauhurst puts at £2.0m. AI companies at seed were valued about 25% higher.
The technical co-founder takes 35% of the founding shares (the midpoint of 30–40%). A 10% option pool and about 1% for advisers are created before the money goes in, as most funders ask. A 30% or 40% co-founder share moves the founder about 3 points either way.
If the option pool is created after the investment, the funder shares the dilution. At £2.0m their stake would fall from 20% to 18%, and the founders’ stake would rise by the same amount. This can matter almost as much as the headline valuation.
If the funder raises its money from individuals, £250k can go in under SEIS (50% income-tax relief) and the rest under EIS (30%). The SEIS shares must be issued first. A company investing its own money gets neither relief. If the funder offers a loan instead of equity, nobody’s ownership changes, but repayments start before Bloom has revenue, which is rarely workable at this stage.
Salaries: founder £45k a year for 18 months, technical co-founder £55k a year from month 3, and an engineer at £55k from month 8, all plus about 18% employer costs. Contract specialists are priced per deliverable from the Bloom specialists plan. Self-hosted AI costs about £1.8k a month once the beta is live. Equity percentages are proposals, as a share of the company before the seed round; the final split, the option pool and the investors’ share are set when the round is agreed. Ownership figures are illustrations, not advice; take legal and tax advice before agreeing terms. Valuation benchmark: Beauhurst, The Deal H1 2026. A proposal to pressure-test, not a fixed plan.
Phase gates
Working web app, child UI, parent hub and tutor AI integrated. Safeguarding live. Maths only. Gate: 20 families, NPS above 70, week-4 retention above 80%.
Educator dashboard, Early Years companion, Literacy added, EHCP reports. First ML hire.
Bloom's own fine-tuned model, full curriculum, first five school partnerships — and the evidence for a Series A of £3–5m.
The UK is the right market to prove the model, build the evidence base, and establish the trust that opens every other market. The AI architecture, pedagogy and experience design transfer entirely.
718,000 EHC plans, the SEND Code of Practice and active, well-connected SEN parent communities. The SEND White Paper (Feb 2026) reshapes support plans, with implementation around 2030: Bloom's reporting is designed to fit both systems.
IDEA gives children with disabilities a federal right to a free appropriate public education and serves 8.2m children and young people. Expansion means district licensing, curriculum localisation to each state's standards, IEP-format reporting and COPPA parental consent — bounded work, not a rebuild.
Revenue as the phases land
Illustrative, driven by the phase gates in the business case. Consumer ARR at each milestone, with school revenue as upside:
A directional view built from the family targets and pricing in the Bloom business case. NHS commissioning and US revenue are not included.
From first concern to GCSE
Bloom is designed to serve neurodivergent families from first concern through to age 16. Each new segment would build on the same child data, curriculum and trust — not a new product.
Recognition guides, diagnosis navigation, play-based activities and a school transition toolkit. Parents find Bloom when their child is 3; at 6, Bloom is already trusted.
EHCP-aligned reporting, evidence of the impact of SEND support and a SENCO morning briefing. SENCO champions become buyers.
Long NHS waits create institutional demand. Every session today builds the RCT dataset that unlocks commissioning.
Bloom is at the point of its first technical hire. Three things would move it fastest.
In the next 90 days: complete the 10 parent interviews, finish the Figma prototype of the first child session, engage a Designated Safeguarding Lead before any child uses the product, and commit the technical co-founder.