Last updated October 3, 2026. Every price, directory figure and source below was re-checked on that date. This article also covers the EHCP and SEND material that previously sat on a separate page.
If you teach special education and want AI to give you time back, start with the job, not the tool. For IEP goal data and progress monitoring, FrenalyticsEDU is the closest thing here to a purpose-built system. For visual schedules, social stories and communication boards, Glint does in a minute what used to cost a prep period. For English local authority teams drafting EHCPs, Sendix is the only tool in our directory built for statutory drafting. For reading material at several levels at once, Diffit. For delivering accommodations on the documents students already open, Kami. For students revising alone, Knowt. For reading aloud in UK primary schools, Readling.
The thing that decides most of these adoptions is not a feature. It is where the data goes. Of the 29 tools in this category, five carry an independent security assessment and four do not name the country their data sits in, which is the question your district privacy officer asks first.
The seven tools compared
| Tool | Best for | Pricing (checked October 3, 2026) | Min age | Data hosted | Independent assessment |
|---|---|---|---|---|---|
| FrenalyticsEDU | IEP goal data, progress monitoring | Free plan; EDU Essentials and EDU Extras published, annual billing | No minimum | United States | No |
| Glint | Visual schedules, social stories, boards | $18/mo or $150/yr | 18+ (educator account) | United States | No |
| Sendix | Statutory EHCP drafting (England) | Quote only, no public pricing | No minimum | United Kingdom | No |
| Diffit | Texts at several reading levels | Free Basic; schools flat-rate annual by enrollment | Unknown on record | Unknown on record | No |
| Kami | Accommodations on documents | Free Basic; Teacher plan $149/yr | No minimum | United States | Yes |
| Knowt | Student study and recall | Free teacher tier; Ultra $199.99/yr | No minimum | United States | Yes |
| Readling | Reading aloud in UK primary schools | No public pricing | No minimum | United Kingdom | No |
The min age, hosting and assessment columns come from the directory record, which reflects what each vendor declares. The assessment column is the independent one. Pricing comes from each vendor's own published page, linked in that tool's section below.
Which special education math software do teachers use?
There is no AI tool in this category built for math intervention, and you should be wary of any list that implies otherwise.
Twelve of the 29 tools list mathematics among their subjects. Nine of the twelve list it inside a blanket set of eleven to sixteen subjects, which is a claim to cover everything rather than a claim to be good at math. Only two list mathematics among four subjects or fewer: FrenalyticsEDU, where math practice is tied to IEP goals and the ready-made content is aligned to K-3 Common Core standards and ABAS-3 domains (Frenalytics), and Alder Branch, which coaches the adult rather than teaching the student.
So if you are searching for special education math software, what you probably want is a structured intervention curriculum, and the established names in that space are not AI products. Use AI for the wrapper around it: practice sets at the right reading level, visual number lines and step-by-step problem guides from Glint, progress data from FrenalyticsEDU. Do not expect it to supply the intervention design, the diagnostic sequence or the error analysis, because none of these tools claims to.
For students working several years below grade level, our guide to special education math software goes through the six tools that come closest and where each one stops.
That gap is worth naming rather than papering over. It is also, for anyone building in this space, the clearest opening in the category.
How to choose, in four questions
Four questions, in this order. The first two rule out more than the last two.
Does a student log in? If yes, you are in a procurement conversation, not a trial. Age floor, hosting location and data agreement all have to be answered before the first login. If no, as with Glint and Diffit used on published material, you can start on Monday.
Which part of this job is the doing, and which part is the thinking? Typing the goal in statutory language is doing. Deciding the goal is thinking. Writing the visual schedule is doing. Deciding the student needs one is thinking. A tool that only touches the first column is one you can adopt quickly. A tool that offers to do the second is one to slow down on.
Can the vendor answer your privacy officer's questions from its own website? Four of these 29 cannot, and four more leave the age floor blank. That is not proof of anything except that your adoption will stall.
Does it produce evidence, or only documents? After a year, a plan built on a document is a plan you have to defend from memory. A plan built on a record defends itself. Given the complaint numbers below, that distinction has stopped being administrative.
The complaint numbers that should shape how you use any of this
Written state complaints about special education rose 56 percent between the 2023-24 and 2025-26 school years across the 35 states surveyed, and due process complaints rose 7 percent, on a survey by the Council for Exceptional Children and the National Association of State Directors of Special Education (K-12 Dive, September 9, 2026). State dispute resolution agencies told that reporting that filings now arrive running to hundreds of pages, carrying misreadings of IDEA and case law citations that do not exist, because a chatbot invented them.
That is the context every tool here now sits in. Parents are bringing AI to the table. So is the district. The Center for Appropriate Dispute Resolution in Special Education publishes a parents' guide to using AI in advocacy, in English, Spanish, Chinese and Korean, with a section on protection, privacy and the problems that come with it (CADRE).
The Center for Democracy and Technology reported that 57 percent of teachers used AI to develop an IEP or 504 plan during the 2024-25 school year, an 18 point rise on the year before, and told schools to consult their compliance offices about the legal risks of AI in IEP development, naming IDEA and FERPA (CDT, October 28, 2025). Read that as a design rule rather than a scolding. Outsource the typing. Keep the judgment about the child.
How these seven were chosen
The pool is the 29 tools in the directory's special education and accessibility category, checked October 3, 2026. Nobody pays to be included and no vendor saw this before publication. Selection ran on fit for a named job, what the vendor documents rather than what it markets, published pricing, and whether the tool leaves the professional decision with the teacher.
Three kinds of tool were left out on purpose. AI detection tools are excluded from every list on this blog because they misfire, and they misfire hardest on the students whose writing is already atypical. Tools whose only special education claim was a "differentiate this" button were excluded, because that is a general classroom feature, not special education. And tools that name no data hosting location were not recommended for anything involving a named student.
What this article is not: a hands-on trial. Every claim below comes from vendor documentation, the directory record, or published research, and each one carries its source. Where a figure could not be confirmed, it says so.
What the directory data shows, including the uncomfortable part
Three numbers from the 29, all checked on October 3, 2026.
Five of the 29 carry an independent security assessment, against 36 of 138 across the whole directory. That is 17 percent here against 26 percent everywhere else, and those five are TeachAid, Knowt, Conker, Kami and Read&Write. Look at that list. Four of the five are general classroom tools that also serve special education. Not one of the purpose-built special education tools in the category holds an independent assessment. The category handling the most sensitive student records in the building is the least independently checked. You can browse the assessed tools on the security assessments page.
Nine of the 29 record no minimum user age, and four more leave it unknown. For a tool a teacher uses alone that is unremarkable. For anything a student signs into, an absent age floor is the question to put to the vendor before a single child logs in, because COPPA makes under-13 accounts a procurement decision rather than a teacher decision.
Four of the 29 record their data hosting as unknown. A blank is not evidence of bad practice. It does mean you cannot answer your district privacy officer's first question from the vendor's own website, and that is enough to stall an adoption.
FrenalyticsEDU is the best fit for IEP goal data and progress monitoring
FrenalyticsEDU is the best fit for turning student practice into IEP progress data, because the practice session and the data collection are the same act rather than two jobs on your evening.
A student works through a personalized session on the platform. The vendor states that built-in assessments "create and collect accurate & quantifiable data during sessions which directly correlate to each individual's IEP goals", so the evidence for the annual review accumulates as the student works rather than being reconstructed from memory in April (Frenalytics). Content is aligned to K-3 Common Core standards and ABAS-3 domains across ELA, math and life skills; speech pathologists, occupational therapists and other service providers can be given access; and families can start sessions and see progress data from home.
What sets it apart: most tools in this category generate a document. This one generates a record. That difference is what makes it useful when a plan is questioned.
Watch out for: the ready-made content skews early elementary, which is a poor fit if you run a high school resource room. The free plan is genuinely free but small: 10 student profiles, one staff account, one daily learning session per student.
Pricing: a free plan at the limits above, then EDU Essentials at 15 student profiles and two staff accounts and EDU Extras at 20 profiles and five staff accounts, with extra students at $1 each on Extras. Paid plans run on annual billing after a two-week free trial (pricing page). The headline monthly figures on that page render without a decimal separator, printing as "$1499/month" and "$1699/month", so read them on the page before you budget. Worth telling us about: our own directory record still lists this tool as quote-only, which is now out of date.
Safety: the vendor states FERPA and COPPA alignment and the record shows US hosting with no minimum age. Because students log in and their goal data lives in the platform, this is a district-level decision with a data agreement, not a tool to trial quietly with one class.
Glint is the best fit for visual supports, and it is the one to start with this week
Glint is the best fit for visual schedules, social stories, behavior expectation charts, task analyses and communication boards, because it produces them from a prompt in the time it takes to find the laminator.
The teacher types what the student needs and Glint returns the material. Its education pages list routine visual schedules and choice boards, task analyses, safety procedures and independence checklists, behavior expectation charts and classroom job systems, graphic organizers, story maps, concept maps, timelines, cause-and-effect diagrams, comprehension questions with visual supports, number lines and step-by-step problem-solving guides, and data tracking sheets for progress monitoring (Glint for education).
What sets it apart: compared with building the same supports in Canva or PowerPoint, the saving is not the drawing. It is not having to think about layout at all when a student needs a new schedule at 8:40.
Watch out for: the generated symbol set is not a clinical AAC system. If a student has a prescribed communication system, Glint makes supporting materials around it; it does not replace it, and your speech-language pathologist should see anything that touches communication. Our guide for speech-language pathologists covers that line in more detail, and the same caution applies to anything touching motor skills or sensory regulation, which is where the AI tools for school-based occupational therapists belong.
Pricing: $18 a month, or $150 a year. An enterprise tier for schools and districts is priced on request and adds an organization-wide library synced across schools, usage analytics and compliance reporting, and bulk licensing with admin controls (pricing page). A free trial exists; the published pages do not state its limits.
Safety: the vendor says it is FERPA and COPPA aligned, and the directory records the minimum age as 18+, which reflects that this is an educator account. Students do not log in, and that is the reason this is the lowest-risk tool on the list to try: you can generate a visual schedule without a single child's name leaving the building, provided you do not type one in.
Sendix is the best fit for statutory EHCP drafting in England
Sendix is the best fit for local authority SEND teams drafting Education, Health and Care Plans, because it is the only tool in our directory built for statutory drafting rather than for classroom paperwork.
An EHCP is the English equivalent of an IEP, and it is a legal document: a local authority writes it, a tribunal can test it. The company has described Sendix as drafting the plan and then verifying each clause against the Children and Families Act 2014 before saving. That claim is unverified as of October 3, 2026: the vendor's site returned no readable product or compliance content on this run, so treat the clause-checking as something to confirm in a demo rather than as an established feature (Sendix).
What sets it apart: every other tool here writes prose and leaves the legal check to you.
Watch out for: it serves one jurisdiction. US readers on an IEP should not adopt this; the framework it works to is English law. There is also no independent security assessment on record, which for a tool handling named children's statutory records is the gap to raise in procurement. Our directory notes its minimum age as "No minimum", which for this product is a record that needs tightening rather than a claim about children using it.
Pricing: quote only. No public pricing at the time of checking.
Safety: UK hosting, GDPR stated, and FERPA and COPPA recorded as not applicable, which is right for an English local authority product. Ask for the data processing agreement and where subprocessing happens before a single plan goes in.
Diffit is the best fit for one text at several reading levels
Diffit is the best fit for putting the same content in front of a class reading across five grade levels, because it rewrites a source text to different levels and generates the question sets to match.
You give it an article, a topic or a link. It returns the passage adapted to the levels you choose, with vocabulary support and comprehension questions attached. For a co-taught class where the content standard is fixed and the reading level is not, that is the daily problem.
What sets it apart: compared with asking a general chatbot to "simplify this", Diffit keeps the same structure across levels, so students are visibly working on the same thing. That matters in an inclusion setting more than the quality of any individual rewrite.
Watch out for: the free plan only keeps access to content created in the last 90 days, and the paid tier is where standards alignment, MTSS tier and DOK tagging, Google and Microsoft export, and complete resource history live (pricing page). The directory record has hosting, minimum age and privacy status all recorded as unknown, and the vendor's public pages did not close those gaps on this check. For adapting a published article that is fine. For anything with a student's name in it, get the answers first.
Pricing: a free Basic plan with the 90-day access limit, and Diffit for Schools as a flat-rate annual subscription tiered by student enrollment, quoted through the vendor's pricing portal or a district quote (pricing page). No individual teacher subscription is listed on that page as of this check.
Safety: unknown on the record for hosting, minimum age, FERPA, COPPA and GDPR, and not independently assessed. Treat it as a teacher-facing content tool and keep student data out of it until the vendor answers.
Kami is the best fit for delivering accommodations on the documents students already use
Kami is the best fit for making a worksheet, PDF or assignment accessible in place, because the accommodations attach to the document rather than requiring a separate one.
Text to speech, annotation, voice and video comments, and drawing tools sit on top of whatever the class is already opening, inside Google Classroom, Schoology, Canvas or Teams. A student who needs text read aloud and a student who needs to answer by voice both work on the same file as everyone else, which is a smaller thing than it sounds until you have watched a student receive a visibly different sheet.
What sets it apart: it holds one of the five independent security assessments in this category, which for a tool students sign into is the relevant difference from most of the field.
Watch out for: the free Basic plan covers "10+ essential tools" and excludes the LMS integrations, which are most of the reason to use it in a special education setting; the paid tiers open all 40+ tools and the Google Classroom, Schoology, Canvas and Teams connections (pricing page). Budget for the paid tier or do not start.
Pricing: free Basic plan; Teacher plan at $149 a year, billed annually; school and district pricing by quote (pricing page).
Safety: the vendor states FERPA, COPPA and GDPR alignment, US hosting, and it carries an independent assessment. The record shows no minimum age, so student accounts still need your district's sign-off.
Knowt is the best fit for students who need to study without a study partner
Knowt is the best fit for a student who can learn the content but cannot revise it alone, because it turns notes and materials into flashcards and practice without an adult having to build them.
For students with working memory difficulties or executive function needs, the barrier at home is rarely the concept. It is that nobody is there to quiz them. Knowt generates the practice from what the class already produced.
What sets it apart: it holds an independent security assessment, and the teacher free tier is unusually generous: unlimited AI use for teachers and students on teacher-shared files (teacher plans).
Watch out for: the free tier caps file creation, live collaboration and storage at 20, which a single caseload will pass quickly. Nothing here is designed for special education specifically; it is a general tool that happens to remove a real barrier.
Pricing: free teacher tier with the limits above; Ultra at $199.99 billed upfront for the year, shown on the page as $16.99 a month, which adds unlimited files, Google Classroom integration and analytics; school site licensing by volume quote (teacher plans).
Safety: US hosting, FERPA and COPPA stated, independently assessed, no minimum age on record. Students create accounts, so this is a district conversation.
Readling is the best fit for reading aloud practice, if you are in the UK
Readling is the best fit for hearing every child read when you cannot, because pupils read aloud from a book and the feedback comes back immediately rather than waiting for one-to-one time you do not have.
The vendor describes it as an app where "pupils read aloud, get friendly AI feedback, and teachers and parents track progress together", built for UK primary schools (Readling). It carries the highest review count in this category on our directory, eight reviews at a five star average, which is a small sample and the only rating in the category with more than five reviews behind it.
What sets it apart: it listens. Nothing else in the category assesses a student's spoken reading.
Watch out for: it is built for UK primary schools, so the book levels and terminology will not map onto a US elementary scope and sequence. There is also no published pricing.
Pricing: no public pricing.
Safety: UK hosting, GDPR stated, FERPA and COPPA recorded as not applicable, no minimum age on record, and no independent assessment. Students read into it, so voice data is in scope and belongs in the data agreement.
The big names, briefly
Search assistants get asked about these constantly, so here is where each sits for special education work.
MagicSchool and Brisk Teaching both ship IEP goal drafters and are widely used; treat their output as a first draft to argue with, and check what your district has agreed with them before a student's details go in. Khanmigo is a tutoring and teacher assistant layer, strong for practice, not built for special education documentation. NotebookLM is useful for a different job: put the evaluation reports and prior plans in and ask questions across them, but only if your district has cleared it for that data. ChatGPT and Claude will draft goals well and will also invent a citation with complete confidence, which is precisely the failure mode showing up in the complaint filings. Quizlet is a study tool, fine for recall, with no special education design in it. None of these five keeps a record of what a student did.
What I would do
Pick by job. Glint this week if you want something working by Friday, because no student logs in. FrenalyticsEDU if the thing drowning you is progress data. Sendix if you draft EHCPs in England. Kami if accommodations are what keep slipping. Then check the two things almost nobody checks: where the data sits, and whether anyone independent has looked. In this category, five times out of six, nobody has.
If you write IEPs specifically, our guide to AI tools for IEP writing goes deeper on that one job, and there are separate guides for students with dyslexia and for intervention teachers. To compare the rest of the field yourself, the special education and accessibility category lists all 29, and the wider tools directory covers 138.
The paperwork is the part to hand over. The child is not.
Dan Fitzpatrick is a Forbes contributor, four-time number one bestselling author and founder of The AI Educator, and has trained more than 150,000 educators across 30+ countries. More about Dan.
