What is Amira? The AI reading assessment and tutor explained

Amira is an AI-powered reading assessment and tutoring suite from Amira Learning that listens to students read aloud, scores what it hears, and uses that information for benchmark screening, progress monitoring, and one-to-one reading practice in kindergarten through grade 8. Teachers use it for instructional placement and to flag students who may need extra support. It is not an IQ test, and it is not a diagnostic instrument for dyslexia or any other disorder. In states that have approved it as a K-3 reading screener, a flag from Amira is a signal to gather more information, and any diagnosis belongs to a qualified clinician. This article covers where Amira came from, what it measures, what reviewers say about the evidence, and how it differs from an IQ test.

Amira Learning describes its technology as having started decades ago in research at Carnegie Mellon University. That research is Project LISTEN, which the university's Robotics Institute describes as an automated Reading Tutor that "uses speech recognition to listen to them read aloud" and responds with spoken and graphical feedback. Education press coverage credits Carnegie Mellon professor Jack Mostow with leading the underlying work.

The commercial company was built by former engineers and executives from Pearson, IBM, ACT, and Renaissance, according to EdWeek Market Brief, and is led by founder Mark Angel. In June 2024 Amira Learning merged with Istation, an assessment platform that EdWeek Market Brief reported had been adopted by more than 1,000 school districts. Terms were not disclosed. The combined business kept the Amira Learning name and Istation's chief executive Dick Collins became co-chairman of the board. As of this writing, istation.com redirects to Amira Learning's company page.

The merger explains a naming detail. Amira's assessment product is called "Amira ISIP Assess" because ISIP stands for Istation's Indicators of Progress, the name of Istation's long-running reading assessment.

The publisher describes Amira ISIP Assess as a benchmark assessment, universal screener, dyslexia screener, and progress monitor delivered in a single oral reading session of about 20 minutes. The student reads aloud into a microphone while the software listens, records "miscues" (misread words), pauses, and self-corrections, and scores the performance automatically. The publisher states that task difficulty adjusts in real time using "item response theory," a statistical framework in which each task has a known difficulty and the student's ability estimate updates after every response.

According to the publisher's product page, the assessment covers the following domains:

• Foundational and phonological skills: phonological awareness, phonological working memory, letter name identification, letter sound identification, and rapid automatized naming.

• Word-level reading and spelling: pseudoword identification, word identification, and spelling or encoding.

• Fluency and comprehension: oral reading fluency, reading comprehension, and listening comprehension or retell.

• Language and attention: expressive vocabulary, receptive vocabulary, structures and reasoning, and visual attention.

The National Center on Intensive Intervention (NCII), a federally funded technical assistance center, lists Amira ISIP Reading on its Academic Screening Tools Chart with an administration time of 20 minutes per student, delivery on a computer or tablet, Spanish versions of every task, and vendor-supplied pricing of $7.50 per student per year for the assessment alone. Teachers receive dashboards with scored transcripts, audio playback, and risk and growth summaries organized by the strands of Scarborough's Reading Rope, a widely used model of how word recognition and language comprehension combine into skilled reading.

A few of these domains, such as phonological working memory and visual attention, overlap with constructs on cognitive test batteries. That overlap does not make Amira a cognitive test; every task is built to describe reading development, and none produces a general ability score.

Amira Tutor is the practice component. The publisher describes it as a one-to-one reading tutor built on the "science of reading," the body of research on how children learn to decode print and comprehend text, for readers from pre-kindergarten through grade 8. A student reads a passage aloud, and when the software detects a struggle it interrupts with what the company calls a micro-intervention, such as a decoding hint or a vocabulary prompt. Sessions are available in English and Spanish, and each one feeds progress data back to the teacher. The publisher recommends about 30 minutes of weekly reading, and Amira's chief AI scientist Ran Liu told Education Week that the company sees a "ceiling effect" beyond sessions of about 25 to 30 minutes.

A third component, Amira Instruct, generates lesson plans from the assessment data, and the three are sold together as the Amira Reading Suite.

Amira's marketing states that it has earned "the highest possible ratings" from NCII. The published chart is more nuanced. For reliability and validity, NCII assigned its top rating of "convincing evidence" in every grade from kindergarten through grade 6. For "classification accuracy," the ability of a screener to correctly sort students who will and will not go on to score below the 20th percentile on the NWEA MAP Reading criterion, the ratings are "convincing evidence" for grades 1 through 3 at all three benchmark windows and "partially convincing evidence" for kindergarten in fall and, where data are reported, for grades 4 through 6. The reported "area under the curve" values, where 1.0 is perfect classification and 0.5 is chance, run from 0.84 to 0.86 in grades 1 through 3 and from 0.73 to 0.79 elsewhere. The profile is strong in the early grades and more modest in upper elementary.

Evidence for ESSA, the clearinghouse run by Johns Hopkins University's Center for Research and Reform in Education, gives Amira a "Moderate" rating for grades K through 4 based on two studies covering 15,602 students, with an average effect size of +0.15. A small 2013 randomized trial of 178 struggling readers found a large effect of +0.64 on the Woodcock Reading Mastery Test, while a 2025 matched comparison of 15,424 kindergarten and first-grade students in a large Texas district found effects of +0.26 and +0.06 on end-of-year DIBELS scores. The larger and more recent study shows the smaller effects.

Education Week's November 2025 report on AI reading tutors, which described Amira as one of the most widely adopted products on the market with statewide contracts in 10 states, called the research picture inconclusive. Most Amira studies compare students by dosage, and University of Florida professor Matt Burns told the publication that the rigorous studies needed to be confident in AI tutors do not yet exist.

Two state reviews are verifiable at government sources. The Michigan Department of Education's July 2026 list of approved K-3 screening and progress monitoring assessments, issued under the state's 2024 K-12 Literacy and Dyslexia Laws, includes Amira ISIP Assess as one of four approved tools. In Georgia, the Department of Education's May 2026 superintendent digest states that it is partnering with Amira as the state's free universal screener provider for K-3 reporting under the Georgia Early Literacy and Dyslexia Act. The publisher also claims approvals in 39 states; that figure is the company's own and was not independently verified for this article.

A screener's job under these laws is to identify children who may be at risk of reading difficulty so that schools can intervene and collect more data. A risk flag from Amira, or from any screener, is a starting point for follow-up. It is not a diagnosis of dyslexia, and it cannot rule dyslexia in or out; that determination requires a comprehensive evaluation by a qualified professional.

Amira belongs to the same family of tools as DIBELS, the i-Ready Diagnostic, and FastBridge: screeners and benchmark measures that schools give to whole classrooms several times a year. What sets Amira apart is its input method. It scores spoken reading rather than clicks, without a teacher administering the test one child at a time, which is why speech recognition accuracy across children's accents, dialects, and noisy classrooms matters so much for its validity.

An IQ test measures something different. It samples reasoning, spatial ability, working memory, and related abilities across a broad range of content, and it reports how a person's general cognitive ability compares with a norm group of the same age. Its questions are chosen to be as independent of specific instruction as possible. A reading screener does the opposite on purpose: it measures skills that instruction is supposed to change, so the score moves when teaching works. A child can have an average or high IQ and still be flagged on Amira because of a specific weakness in decoding, and a low screener score says nothing about IQ. Amira also produces no normed standard score on the mean-100, standard-deviation-15 scale; its outputs are skill estimates, risk categories, and fluency rates meant for a teacher planning instruction.

Amira answers the question "how is this child's reading developing, and what should the teacher do next?" It does not tell anyone how a person's reasoning ability compares with the population, and it was never built to. For adults who want that comparison, the right instrument is a psychometrically constructed IQ test with published norms, and the questions Dr. Warne raises about online cognitive tests apply. Our guide to whether online IQ tests are legitimate walks through what to look for. The Reasoning and Intelligence Online Test, developed by RIOT IQ with psychometrician Dr. Russell T. Warne, is an online option for adults 18 and older. It includes 15 subtests across six cognitive indices (verbal reasoning, fluid reasoning, spatial ability, working memory, processing speed, and reaction time), takes about 52 minutes, and reports scores on the familiar mean-100, standard-deviation-15 scale. Like any online test, it does not replace an individually administered diagnostic evaluation. You can learn more or take the test at riotiq.com.

No. Amira is a reading assessment and tutoring suite. It measures skills such as phonological awareness, decoding, and oral reading fluency, and it produces no IQ score.

No. The publisher describes Amira ISIP Assess as a dyslexia screener, and some states have approved it for K-3 screening under early literacy laws. A screener flags children who may be at risk so that schools can intervene and gather more information. Diagnosing dyslexia requires a comprehensive evaluation by a qualified professional.

The publisher markets the Amira Reading Suite for kindergarten through grade 8, with Amira Tutor supporting readers from pre-kindergarten. The NCII screening chart reports technical data for kindergarten through grade 6.

The publisher and the NCII tools chart both list about 20 minutes per student, delivered as an oral reading session on a computer or tablet with a microphone.

Evidence for ESSA rates Amira "Moderate" based on two studies with an average effect size of +0.15. Education Week's 2025 review of AI reading tutors described the overall evidence as not yet conclusive.

1. Amira Learning. (n.d.). Amira ISIP Assess: Reading assessment for elementary students and middle school learners.. amiralearning.com

2. Amira Learning. (n.d.). Amira Tutor: AI reading tutor and program.. amiralearning.com

3. Ng, A. (2024, June 14). K-12 dealmaking: Amira Learning merges with Istation; Weatherford Capital invests in BusPatrol. EdWeek Market Brief. marketbrief.edweek.org

4. National Center on Intensive Intervention. (n.d.). Amira ISIP Reading: Academic screening tools chart.. charts.intensiveintervention.org

5. Michigan Department of Education. (2026, July). K-3 valid and reliable screening and progress monitoring assessment list (MCL 380.1280f).. michigan.gov

6. Georgia Department of Education. (2026, May 15). May 15, 2026: Superintendents. GaDOE Insights. georgiainsights.gadoe.org

7. Schwartz, S. (2025, November 20). AI tutors are now common in early reading instruction. Do they actually work? Education Week. edweek.org

8. Evidence for ESSA. (n.d.). Amira. Center for Research and Reform in Education, Johns Hopkins University. evidenceforessa.org

Hero photo: A child pointing at a laptop screen. Photo by Shixart1985, via Wikimedia Commons, licensed under CC BY 2.0 (cropped).