Aligned Impact built the research foundation and the demos. Now we need the educators who work with multilingual learners every day to explore what we built, tell us what lands, and shape what comes next. Your voice builds the real platform.
Brilla removes the barriers. It creates a belonging space where students engage with rigorous grade-level content using the language and academic supports that fit their needs. Teachers get real-time knowledge to act on in the classroom. Leaders have a clear pathway to support their teachers. Families are engaged throughout.
And language development progress is visible — not just to one teacher in one classroom, but across the campus, in the framework educators already use to understand their students.
Brilla was designed from the ground up for multilingual learners. The research, the scaffolds, the language supports, the agency tools — every design decision starts there. And because Brilla is built on Universal Design for Learning principles, the supports that serve MLs well strengthen the experience for every student who uses it. What is essential for some is beneficial for all.
Every audience has a dedicated experience. Every experience is connected to the same underlying data. No siloes. No afterthoughts.
For curriculum directors and district leaders who want to understand ML language development and math engagement not just in one school but across their entire system. Built on the same data architecture as the school-level tools, aggregated to give district leaders the visibility to make decisions about professional development, resource allocation, and school support.
What did Brilla get right? What is missing? What does AI actually need to do in your classroom? Your feedback shapes what gets built next.
Share Your Feedback →The demos are real. The research is real. What we need now are the educators who work with multilingual learners every day to tell us what Brilla gets right and what it still needs to become.
Six research pillars shaped every decision in this vision before a single line of code was written.
WIDA establishes that multilingual learners develop content knowledge and academic language together, not in sequence. When instruction honors both at once, mathematical thinking deepens and language grows. Brilla is designed around that simultaneity.
Conversational fluency and academic language proficiency develop on different timelines, 1 to 2 years and 5 to 7 years respectively. Brilla supports MLs through both stages with scaffolds that grow as they grow.
Differentiated, linguistically responsive pedagogy built on students’ individual assets produces stronger outcomes for MLs in both math and language development. Brilla operationalizes that pedagogy in every interaction.
When students are positioned as intellectually capable, neural pathways for deeper learning activate. Brilla’s asset-based design is a cognitive architecture decision grounded in how learning actually works.
A meta-analysis of 2023 to 2025 studies found that generative AI interventions meaningfully enhance K–12 students’ mathematics learning outcomes when implementation prioritizes pedagogical quality over tool sophistication.
Multiple means of representation are not accommodations, they are the architecture of excellent instruction for every learner. Brilla builds UDL into every interaction as a design principle, not an add-on.
“Brilla did not start with technology. It started with a question: What would math class look like if every multilingual learner walked in knowing the room was built for them?”
Select any topic below to go deeper. Each section is built for educators who want to understand the research, the design, and the thinking behind Brilla.
The four design pillars that shape every Brilla interaction, grounded in asset-based intelligence, productive struggle, community, and mathematical identity.
Explore Vision →Every design decision tested against evidence before anything was built. Twelve frameworks including WIDA, Hammond, Cummins, Vygotsky, and the 6 Cs.
Explore Research →Brilla thinks differently for a second grader than a seventh grader. Every scaffold, question, and prompt is calibrated to the student's developmental stage.
Explore Grade Bands →Brilliance flows in every direction. How Brilla supports students, teachers, and leaders as an interconnected ecosystem, not three separate tools.
Explore Capacity →Every tool built for how multilingual learners actually learn math. Translanguaging, the six Math Language Routines, UDL, and bilingual support as a right, not an accommodation.
Explore Language Supports →Student names never enter the system. Districts own their data. No selling, no sharing, no exceptions. Built for communities that have been most harmed by data misuse.
Explore Privacy →Six commitments that govern every AI output in Brilla. No answer-giving, bias auditing, community consent, and the human always in control.
Explore Ethical AI →The Biweekly Brilliance Bulletin delivered to families in their home language. When families understand what their child is building, the learning does not stop at 3pm.
Explore Community →How Brilla tracks language growth using the proficiency framework your state already uses. Includes the Language Growth Tracker and Brilliance Heatmap explained.
Explore State Alignment →What each person experiences and why it matters. Students, teachers, leaders, and families, with honest framing about what the demo shows versus what the live platform will do.
Explore the Demos →See all three portals side by side. Student Agency Portal, Teacher Sidekick, and Leader Coaching Lens in one view. Switch between roles and explore what Brilla envisions for each.
Explore the Platform →The best way to understand Brilla is to experience it. Each demo takes about 5 minutes.
Brilla is designed so the data teachers and leaders see matches the proficiency language their state already uses for assessment. A teacher in California sees Emerging, Expanding, and Bridging. A teacher in Texas sees Beginning through Advanced High. The student experience is identical. Only the data language adapts.
The English Language Proficiency Assessments for California measures Emerging, Expanding, and Bridging levels across Listening, Speaking, Reading, and Writing. Brilla aligns teacher and leader data views to these levels and surfaces Reclassification Criterion 3 signals for students approaching reclassification readiness.
The WIDA ACCESS assessment uses a 6-level framework across three dimensions: Discourse, Sentence, and Word/Phrase. Brilla surfaces student progress across all three dimensions so teachers can see not just where a student is overall but which dimension needs the most instructional attention.
The Texas English Language Proficiency Assessment System uses four proficiency levels across Listening, Speaking, Reading, and Writing. Brilla aligns teacher-facing language progress data to these four levels and flags reclassification readiness signals for students approaching Advanced High.
The New York State English as a Second Language Achievement Test uses a 5-level continuum from Pre-Emerging through Commanding. Brilla surfaces student progress in these terms so teachers can connect what they see in the portal to what they see on annual ELP assessments.
The Language Growth Tracker is a teacher and leader-facing tool that shows how each multilingual learner is developing across three language dimensions over time. It uses the proficiency framework your state already uses so there is no translation layer between what Brilla shows and what educators already know.
Each student card shows their current proficiency level, their growth across Discourse, Sentence, and Word/Phrase dimensions, a six-week trend visualization, and a reclassification signal flag when the data suggests they may be approaching readiness.
In the live platform, the tracker updates after every session. In the demo, it shows six weeks of simulated data so educators can experience what the tool would look like in practice.
Each student by first name, nickname, and Scholar code. Current level, three dimension bars, a six-week trend chart, and a reclassification flag. Framework switches between ELPAC, WIDA, TELPAS, and NYSESLAT.
The same student-level data teachers see, plus school-level language engagement trends and reclassification signals across classrooms.
Real student names. The name-to-Scholar-code connection lives only in your district's own system. Brilla only ever processes the Scholar code.
The Brilliance Heatmap gives teachers a one-screen view of every student's engagement level in that day's session. Each cell represents one student and is color-coded by what the data suggests they need right now.
Teachers can click on any student cell to see a detailed view including their current proficiency level, this week's pattern, and one specific suggested move for today's lesson.
The heatmap only shows what Brilla can observe from portal interactions: which support path a student selected, how often they engaged, and whether they used bilingual support. It does not claim to know what happened in the physical classroom.
A student who selects "needs processing time" every session is not struggling. They may be doing sophisticated reasoning in their home language before they are ready to produce in English. The heatmap surfaces that pattern so the teacher can respond with more entry time, not more pressure.
The demos are interactive illustrations of the Brilla vision, not a finished product. The conversations, data, and tools you see show what the experience is designed to feel like. A live platform would respond dynamically to each person's actual input, adapt to real classroom data, and grow with your students over time. The goal of the demos is for you to understand the experience and tell us whether we are designing it right.
Each demo takes about 5 minutes. The feedback button on every page is how you tell us whether we got it right.
Grade 4 demo built around standard 4.NBT.B.5 (multiplying two-digit numbers) using a school fundraiser context. Features a full Three Reads entry sequence with numbers hidden on the first read, an interactive area model visual, three-phase entry (context activation, language access with cognates, Three Reads), and three session modes: Independent, Partner Explore, and Small Group. Includes Socratic sense-making support, bilingual toggle with full mathematical warmth in Spanish, sketchpad, sentence frames, hint ladder, mood check, metacognitive end-of-session reflection, and demo guidance throughout. Designed to support equitable access to grade-level math for multilingual learners without reducing rigor.
Full Grade 4 teacher dashboard with a class mode toggle (Independent, Partner Explore, Small Group) that controls what students see in their portals. Equitable Lesson Roadmap for 4.NBT.B.5 with explicit language routines, ML moves, What NOT to Do guidance, and Tier 2 and Tier 3 notes per phase. Brilliance Heatmap showing all 18 students by first name with clickable three-dimensional trend analysis (language, conceptual, content-specific) and grouping suggestions that name the specific barrier, who to group with, and when to revisit. Language Growth Tracker across four state frameworks with reclassification signals. Lesson Setup with flexible pacing, curriculum-aware newsletter deployment, and 48-hour prompts. Family message inbox with red dot notifications and outreach history.
School-level leadership dashboard with focus area selection from five high-leverage ML instructional moves, walkthrough brief generator, observation log, four-stage coaching cycle tracker (Introduce, Develop, Deepen, Sustain), AI-powered coaching action recommendations including Professional Learning, Teach Back, Peer Observation, and Collaborative Planning with a focus.
Phone-first family experience with no login required. Weekly Brilliance Bulletin showing student breakthroughs, real-world home activity connected to the lesson, newsletter archive organized by module and lesson, mood signal for families to send to the teacher, and weekly text message delivery in the family's home language. Available in English, Spanish, and French.
Six-week language development tracking across three dimensions for every ML student. Supports CA ELD/ELPAC, WIDA ACCESS, TELPAS, and NYSESLAT frameworks. Shows reclassification signals, dimension-level growth bars, and six-week trend visualizations. Accessible to both teachers and leaders with full student name and nickname visibility.
Every update on this page reflects feedback from educators who explored the demos and told us what was missing. If you want to influence what gets built next, join the pilot.
Brilla is an AI-powered math platform designed from the ground up for multilingual learners in grades K–8. It was built by Aligned Impact, a consulting firm that works at the intersection of instructional leadership, equity-centered design, and AI strategy.
The name Brilla is Spanish for shines. It is a reminder that every student who walks into a math classroom carries brilliance. The question is whether the tools, the language, and the support structures are designed to let that brilliance show.
Brilla was designed from the ground up for multilingual learners. The research, the scaffolds, the language supports, the agency tools — every design decision starts there. And because Brilla is built on Universal Design for Learning principles, the supports that serve MLs well strengthen the experience for every student who uses it. What is essential for some is beneficial for all.
A belonging space where students engage with grade-level math using Socratic AI tutoring, bilingual support, visual scaffolds, and agency tools. Students choose a nickname. Their thinking in any language is evidence of brilliance.
Real-time knowledge that teachers can act on in the classroom. Live engagement heatmap, equitable lesson roadmaps, language growth tracking across state frameworks, lesson setup with curriculum-aware newsletter deployment, and a direct line to family messages.
A school-wide view that helps principals support their teachers with precision. Focus areas, walkthrough briefs, observation logs, a four-stage coaching cycle, and AI-powered recommendations that tell leaders exactly what to do next.
A phone-first, no-login portal that reaches families where they are. Weekly Brilliance Bulletin, real-world home activities, curriculum newsletter archive, and a mood signal that goes directly to the teacher. Available in multiple languages.
Aligned Impact is an education consulting firm with deep roots in instructional leadership, culturally responsive teaching, and equity-centered design. The team has worked across California, Texas, Tennessee, and Georgia in schools serving high proportions of multilingual learners.
Brilla is Aligned Impact's answer to a question that kept coming up in schools: what would AI in math look like if it were designed first for the students most AI tools leave behind?
Brilla’s vision is a K–8 math experience where every multilingual learner is positioned as a capable mathematical thinker from the moment they enter the room. Not after they have proven themselves. Not after completing a program. Right now, in every lesson, with every teacher, for every ML in the classroom.
Most edtech starts from what students cannot do yet. Brilla starts from what they already bring. Every feature begins from what a student demonstrated, not what they failed to demonstrate. The question is never what is wrong with this student. It is always what did this student show me, and what does that make possible next.
This is grounded in Funds of Knowledge theory (Moll, Amanti, Neff, and Gonzalez, 1992), which establishes that every household and community holds deep knowledge built over generations. Multilingual learners bring those funds into every math classroom. Brilla is designed to surface and build from them.
The moment Brilla gives a student an answer, a learning opportunity dies. Every scaffold is a thinking prompt. Every response is a question. Students do the mathematical work. Brilla makes sure they have what they need to do it.
Vygotsky’s Zone of Proximal Development is not just a citation in Brilla. It is the architecture of the scaffold system. Every prompt is calibrated precisely to the edge of a student’s current capacity. Not below it. Not beyond it. At the exact place where supported effort produces growth.
Learning does not happen in isolation. Belonging and mathematical confidence are not separate things. When a student feels unseen in a room, their brain does not take the risk that reasoning out loud requires. Brilla centers peer discourse, family connection, and home language as core instructional infrastructure. Not enrichment. Not supplemental. The community is the curriculum.
Brilla’s bilingual support is grounded in translanguaging theory (Garcia and Wei, 2014), which establishes that multilingual learners draw on a unified linguistic repertoire. When a student begins explaining in Spanish and moves into English mid-thought, that is a multilingual thinker using every tool they have. Brilla receives it as complete mathematical thinking.
Golinkoff and Hirsh-Pasek’s Becoming Brilliant (2016) establishes six conditions through which deep learning happens. These are not outcomes that follow instruction. They are the conditions through which content knowledge, language, and identity develop simultaneously. Every Brilla feature maps to at least one.
The ML Snapshot and Discourse Roadmap help teachers design intentional groupings so students can push each other’s thinking. Collaboration is not assumed. It is structured.
All six Math Language Routines, the bilingual exchange, sentence frames, and the Audio Bridge develop mathematical communication in every modality and every language the student carries.
Content rigor is never lowered. Scaffolds raise access without lowering the mathematical expectation. Every ML in every grade encounters grade-level content from their first interaction.
Brilla never gives answers. Every response is a thinking prompt. Every exchange pushes students to justify, argue, and prove. The productive struggle is protected because that friction is where critical thinking develops.
The Math Sketchpad invites students to show their mathematical thinking in their own way. The pilot is actively asking educators: how do multilingual learners show creative thinking in math, and what would that look like in this tool?
The Daily Pulse, Brilla Noticed card, Brilliance Heatmap, and Scholar naming are all confidence-building decisions. Mathematical identity and mathematical confidence develop together or not at all.
Changes what a teacher looks for in a student. When you start from strength, you see different evidence and make different instructional decisions.
Protects the cognitive work that asset-based thinking surfaces. You cannot honor a student’s brilliance and then do the thinking for them.
Creates the conditions where the first two can happen. A student cannot take the risk of reasoning out loud in a room where they do not belong.
The outcome that makes all three sustainable. A student who sees themselves as a mathematical thinker keeps engaging even when the work is hard.
These are not name-drops. They are the intellectual foundations every design decision in Brilla was tested against before anything was built.
"The question that started all of this was simple: What would math class look like if every multilingual learner walked in knowing the room was built for them?"
We do not have a complete answer yet. These four pillars are what we believe so far. The demos show what those beliefs look like when they become a product. Educators who explore these demos and share their feedback are the ones who will tell us whether we got it right.
Every scaffold, every question, every prompt is calibrated to the mathematical and linguistic developmental stage of the student in front of it.
These students are building their first relationship with mathematical thinking, often learning to read, learning English, and learning math simultaneously. The home language is the primary vehicle for mathematical reasoning. Brilla honors that completely.
These students are making the shift from learning to count to counting to learn. The space between what they can think and what they can express in English is where Brilla lives, making that space generative, not frustrating.
Many MLs at this band are building sophisticated mathematical intuition in their home language that has never been fully activated in English academic contexts. Brilla adds the academic English register for argumentation without lowering the mathematical expectation.
These students have developed sophisticated mathematical intuition, strong pattern recognition, and disciplinary reasoning that often exceeds what standardized measures capture. Brilla adds the formal language of mathematical proof and generalization.
| Feature | K–2 | 3–4 | 5–6 | 7–8 |
|---|---|---|---|---|
| AI response length | 1 sentence | 2 sentences | 3 sentences | Full exchange |
| Primary output mode | Drawing | Drawing + writing | Writing + audio | Writing + oral argument |
| Home language support | Default on | Toggle available | Toggle available | Toggle available |
| Daily Pulse | Emoji only | Emoji + optional sentence | Emoji + reflection | Emoji + metacognitive prompt |
| Answers given | Never | Never | Never | Never |
The best classrooms we have ever walked into felt like this. Something alive. Something where the teacher trusted the students to think, and the students trusted the teacher enough to try. That trust does not come from a program. It comes from conditions that someone upstream took the time to build. Brilla exists in the middle of that web, not directing it, but nourishing it.
Leaders need visibility into ML learning patterns without compromising student privacy. Coaching language that starts with strength. Data that surfaces what is working and where to invest next. Brilla gives leaders the walkthrough brief, the debrief protocol, and school-level ML engagement trends, organized and ready before they walk in the door.
Teachers need a co-pilot for differentiation, not another thing to manage. Brilla organizes the complexity of ML differentiation so teachers can focus their expertise where it matters most: in the relationships and decisions that only a human can make. The Equitable Lesson Roadmap, the ML Snapshot, and the Discourse Roadmap are formative assessment tools (Wiliam, 2011) designed on the principle that assessment is most powerful when it feeds forward to the next instructional decision, not backward to a grade.
Students need to be seen as thinkers first. Tools built for how they think, how they communicate, and what they already know, in every language they carry. The Student Portal gives them agency over their own scaffolding and reflection. Their thinking is always the starting point.
"Brilla builds capacity by making the brilliant work already happening more visible, more connected, and more responsive to every learner in the room."
Every tool in this vision was designed from the beginning with multilingual learners at the center. Not added on. Not retrofitted. Not simplified. Built for them, full stop. The research that grounds these design decisions is not background reading. It is the architecture.
Garcia and Wei (2014) established translanguaging as the natural cognitive process by which multilingual learners draw on their complete linguistic repertoire to make meaning. A multilingual learner is not switching between two separate language systems. They are using one integrated system that includes every language they carry.
This changes everything about how a platform like Brilla should respond to a student. When Scholar C2 begins explaining fraction reasoning in Spanish and moves into English mid-thought, that is not code-switching or a language accommodation. It is sophisticated cognitive work that should be received as complete mathematical thinking.
"Mathematical competence and English proficiency are not the same thing. Conflating them systematically underestimates what multilingual learners know and can do mathematically."
Judit Moschkovich's research on mathematical discourse with bilingual learners is foundational to every design decision in Brilla's student-facing interactions. A student who explains their reasoning in Spanish is demonstrating mathematical competence. Brilla is built to see that clearly.
Brilla surfaces student linguistic assets and mathematical reasoning patterns to help teachers design intentional groupings, pairing students who can push each other's thinking rather than sorting them by English proficiency level. The ML Snapshot identifies who draws well, who explains well, and who is ready to lead discourse.
Every interaction begins from strength. The platform identifies what a student demonstrated and builds the next instructional move from that evidence. Teacher-facing data uses language like "Student A1 showed emerging understanding of..." Never deficit framing, ever. This is a direct application of Funds of Knowledge theory (Moll, Amanti, Neff, and Gonzalez, 1992): the knowledge students bring from home and community is instructional material, not background noise.
Students toggle between English and their home language at any point without losing their place in the math. When a student responds in Spanish, Brilla receives it as complete mathematical thinking and continues the exchange. This is translanguaging as pedagogy (Garcia and Wei, 2014): honoring the full linguistic repertoire of every multilingual learner as an asset, not an obstacle to route around.
Brilla scaffolds toward conceptual understanding in every interaction because every ML deserves to know not just how the math works, but why. Every scaffold is a thinking prompt, not a worked example. This aligns with Vygotsky's Zone of Proximal Development: scaffolds calibrated precisely to the edge of a student's current capacity, not below it. When a student no longer needs a scaffold, Brilla notices.
Zwiers et al. (2017) established this principle and it is the reason the MLRs exist: when students are arguing mathematically, explaining their reasoning, and critiquing the thinking of others, they are not preparing to learn the content. They are learning it. Language and mathematics develop simultaneously in the act of discourse, not in sequence.
This is also consistent with Halliday's systemic functional linguistics (1978): language is a meaning-making resource shaped by the contexts in which it is used. Mathematical language develops most powerfully in mathematical contexts, with real mathematical purposes.
Multiple means of representation, action and expression, and engagement are not accommodations for MLs. They are the architecture of excellent instruction for every learner. Brilla builds all three into every interaction as design principles.
Math concepts through visual, linguistic, and audio modes. Students access the concept through the mode that works for them.
Students respond by drawing (Math Sketchpad), speaking (Audio Bridge), or writing (text input). Expression mode is always student-chosen.
Daily Pulse builds metacognitive engagement. The support path menu gives students agency over their own scaffolding, always.
The six Cs are not outcomes that follow learning. They are the conditions through which deep learning happens. Each one is present in Brilla's design, simultaneously and interdependently.
CAST (2018) and Golinkoff and Hirsh-Pasek (2016) arrive at the same conclusion from different directions: deep learning is multimodal, social, and built on agency. Brilla is designed at the intersection of both frameworks.
We know what is at stake when a district agrees to bring an AI tool into their schools. They are not just agreeing to a new piece of software. They are placing trust, on behalf of students, families, and educators, in an organization they may have just met. We take that seriously. This page is our attempt to be completely transparent about how Brilla handles data, what we will never do with it, and what you can hold us to.
The platform assigns anonymous identifiers from the moment of onboarding. No name. No photo. No birthdate. No address. No personal identifier of any kind ever enters the Brilla system. A student is Scholar A1 to us, and that is all we need to serve them well.
That boundary is not a default setting that can be changed. It is a design constraint we built into the architecture from the beginning. We made it impossible to store student names because we do not believe it should ever be possible.
Student names are translated to anonymous identifiers by the district’s own system before any data reaches Brilla. We never see the name at any point in the process.
Districts retain full ownership of all data generated during a Brilla engagement. That means the district can export everything, delete everything, or end the engagement at any time, and Brilla will honor that request without question, without delay, and without retaining anything afterward.
Multilingual learners are disproportionately represented in communities that have historically had the least power over what happens to their data, their health data, their financial data, their education data. They are also the students Brilla is designed to serve. That is not a coincidence we take lightly.
We built the privacy architecture the way we did because we believe the communities that have been most harmed by data misuse deserve the most rigorous protection. Not adequate protection. Rigorous protection.
We also know that trust is not established by a privacy page. It is established by consistent behavior over time. This page is our starting commitment. What we do with your data, or more precisely, what we do not do, is how we earn the right to keep working with your community.
If you ever believe we have violated anything on this page, we want to hear from you directly before you hear from a lawyer. That is how we intend to operate.
Brilla’s privacy architecture was designed against the Future of Privacy Forum’s generative AI vetting checklist (April 2024) from the first day of design work. We did not wait for a district to audit us. We audited ourselves first.
Privacy is not only about what we do with data. It is also about who can access the platform in the first place. A tool that only works for students with reliable broadband, new devices, or strong English proficiency is not an equitable tool, regardless of its privacy policy.
We welcome questions before you commit to anything. Use the pilot interest form to reach us and we will respond personally, not with a form letter.
When a student opens their portal, Brilla asks them to choose a nickname. This can be a superhero, a book character, or anything they like. The AI calls them by this nickname throughout their session.
The nickname is stored only for the duration of that browser session. It is never sent to a server, never attached to any data record, and never shared with anyone. The next session, the student chooses again.
Teachers and leaders see student first names alongside Scholar codes in their dashboards. This connection happens through your district’s own rostering system during setup.
The name-to-code mapping lives only in the school’s local browser session, not on Brilla’s servers. Brilla only ever processes the Scholar code. Real names are translated to codes before any data reaches Brilla, and they stay there.
“The students Brilla is built for have already been failed by too many systems that treated their data as a resource instead of a responsibility. We are not going to be one of them.”
Every student chooses a nickname at the start of each Brilla session. A superhero name, a character, anything they want. That nickname is what appears in the student portal during the session. It is never stored by Brilla, never shared with anyone, and disappears when the session ends. The nickname gives students a sense of identity and ownership in their learning space without any personal information ever leaving the classroom.
Each student is assigned a Scholar code by their teacher during class setup. Scholar codes are the only identifiers that move through Brilla's systems. A code like A1 or C2 carries no personally identifiable information. Student first names live only in the teacher's local browser session and are never transmitted to Brilla's servers. Teachers and leaders see name and Scholar code together. The platform never stores a student's real name.
Teachers can set the class to Independent, Partner Explore, or Small Group mode from their Lesson Setup. This controls what students see in their portal. Mode assignment is the teacher's decision and is not stored as student data. No additional information about a student is collected when their mode changes. The mode setting applies to the session only and resets the next time the teacher sets up their lesson.
Each week Brilla sends a text message to families in their home language. The text is generated from session data and contains a specific, asset-based description of what their child demonstrated that week, a real-world home activity connected to the lesson, and a link to the full Family Portal.
The link opens directly to the family's portal with no login required. Family phone numbers are collected during the school onboarding process and are used solely to send the weekly text. They are never shared, sold, or used for any other purpose. Families can opt out at any time by replying STOP.
Brilla is being built as an AI platform. We think that comes with serious responsibilities, especially when the humans in the room are children, and especially when those children are multilingual learners whose linguistic assets have too often been overlooked by the systems meant to serve them.
Brilla protects productive struggle by never providing direct answers to math problems. It provides thinking prompts that help students find the answer through their own reasoning. The moment Brilla gives an answer, a student loses a learning opportunity. We will never make that trade.
AI systems trained on historical data can reproduce historical patterns. Brilla commits to regular bias auditing of all AI outputs, specifically checking whether scaffold suggestions differ by language background, because every student deserves the same quality of support.
No district enters a Brilla pilot without explicit informed consent from leadership, staff, and families. Consent is not a checkbox, it is an ongoing conversation. Families receive the Bulletin in their home language so they can stay informed and opt out at any time.
Brilla will not use black-box AI. Every scaffold suggestion, every heatmap signal, every coaching output will be explainable in plain language to any educator who asks. If we cannot explain why the AI suggested something, we will not surface it.
Every AI output in Brilla is a suggestion, not a directive. Teachers can dismiss, modify, or ignore any recommendation. Leaders can override any heatmap signal. Students can close any scaffold. The human is always in charge. Always.
Brilla commits to ongoing quality auditing of all multilingual outputs, because every student whose home language is something other than English deserves the same precision, warmth, and mathematical integrity in every scaffold they receive.
Data governance must be foundational to any educational AI implementation, not retrofitted after the fact.
"When our languages, identities, and histories are missing from the data, we are either misrepresented or made invisible." Brilla’s multilingual equity auditing is a direct response.
60% of K–12 teachers used AI tools in 2024–25. Ethical frameworks must keep pace with adoption. Human oversight must remain central.
Systems must maximize equitable access to AI tools so that ALL students benefit, with specific attention to multilingual learners and under-resourced communities.
Human oversight must remain central to education-related AI systems. AI augments, never replaces, educator expertise.
Multilingual contexts and long-term conceptual understanding remain underexplored in AI math research. Brilla exists precisely to fill that gap.
"We are not neutral about AI. We believe it can be a force for equity, but only if the people building it are willing to be held accountable. We are."
Delivered every two weeks in the family’s home language, the Brilliance Bulletin is not a translation of school jargon. It is an invitation into the learning, a celebration of what students are building and a bridge between school and home.
WIDA’s research establishes that when families engage with their child’s academic content in their home language, mathematical confidence deepens and academic language grows, because learning is never just a school-day event.
The Bulletin is not a one-way communication tool. It is a two-way relationship between the platform, the school, and the community that makes learning possible.
Six weeks of language development data for your MLs, organized across the three dimensions Brilla observes in every student interaction. In a live deployment this updates after every lesson. This simulation shows what that data looks like in practice.
Moving from "I don't understand" toward "I started but got stuck" a measurable growth signal.
Bilingual toggle use declining: growing independence in English mathematical language.
Moving from single answers to claim-with-evidence. Discourse dimension strengthening.
Surge in weeks 5-6 after teacher began giving explicit discussion role assignments.
What you are seeing is a simulation of six weeks of Brilla observations. In a live deployment, every student portal interaction, support path selection, bilingual toggle use, and discourse participation event is logged and surfaces here. The three dimensions shown (Discourse, Sentence, and Word/Phrase) align directly with WIDA Proficiency Level Descriptors and California CA ELD Standards so teacher observations connect to annual ELP assessment data.
The strength you identified is the right anchor for your debrief. Name the specific moment, not the general pattern. "I watched the student at table 3 explaining with her hands" is more powerful than "your MLs were engaged." Specific observations build specific teaching moves.
Anchor the next step to one specific moment in tomorrow’s lesson. Instead of "give MLs more time," try: "Before you open the whole group share, give table 3 60 seconds and the sentence frame. That one move changes who gets to talk."
This is simulated progress data illustrating what Brilla would track over six weeks. In a live platform, data updates after every lesson session. Student first names are connected to Scholar codes through your district’s rostering system, which Brilla never stores. Nicknames are chosen by students at the start of each session.
Marcus (Miles Morales) moved from selecting "I don't understand" to "I started but got stuck" across 6 weeks, a measurable shift in self-assessment and entry point confidence. Amara (Hermione) increased sketchpad use every week and is producing longer written explanations alongside her drawings.
Sofia (Spidergirl) has selected "I don't understand" in 4 of her last 6 sessions. She is completing every session but has not yet shifted to a higher support path. More entry-point scaffolds before she opens the problem may help.
Diego (Black Panther) is showing active reclassification signals based on his current proficiency level in the selected framework. His bilingual toggle use has dropped by 70% over 6 weeks, signaling growing independence in English mathematical language. Worth flagging for your next EL coordinator check-in.
The AI responses you saw in the demos are pre-written illustrations of what Brilla would do. They are not live. A real platform would respond dynamically to each person’s actual input and adapt to real classroom data.
With that in mind, we want to know what you think about the vision, the design, and whether it matches what your students and school actually need.
Use the form below or open it in a new tab. Either way your feedback goes directly to the Aligned Impact team.
Opens in a new tab. Takes about 5 minutes.
Every response is read by a person on the Aligned Impact team. If you leave your email we will respond personally.
We compile feedback themes and share what we are learning with pilot partners.
The pilot is a 6 to 8 week co-design engagement where your feedback directly shapes platform decisions.
The Brilla pilot is a co-design partnership. Schools that join are not beta testers. They are founding partners whose feedback directly shapes what gets built next. Every step of the process is designed to protect students, support teachers, and give leaders real visibility into what is working.
Before anything is signed or scheduled, we sit down with school leadership to assess fit. We talk through what Brilla does, what data it collects, how that data is protected, and what it is never used for. We explain the Scholar code architecture, the nickname system, and the family text system in plain language.
We complete the formal data agreement together. We also support the school in communicating with families before the pilot begins. Families receive an introduction to the platform that explains what Brilla is, how their child's information is protected, and how the Family Portal works. For multilingual families this communication goes out in their home language.
Before training begins, we confirm that the technical foundation is in place. We work with your IT team to verify device compatibility, check network access, confirm that the platform loads on the devices students actually use, and identify your technical point of contact for ongoing support.
We build a training plan together. Not a generic one. A plan that accounts for your school's schedule, your teachers' starting points, and what your leaders need to be able to do on day one. We determine the sequence, the timing, and who needs what level of support.
Training is not one-size-fits-all. Every audience gets what they need in the format that works for them.
Every pilot school has a named Aligned Impact contact. Not a ticketing system. Not a general inbox. A person who knows your school, your teachers, and your context. We commit to a response time and we keep it. Questions, issues, and observations surface through a simple channel that does not add to anyone's workload.
Twice a year we meet with school leadership and teachers together to review what the data shows, what educators are experiencing, and what Brilla needs to do differently. Student surveys go out at the same cadence. We ask students directly what is working, what is hard, and what they wish the tool did. Their voice is not optional. It is the point.
At the end of each pilot cycle, your school receives a co-design brief — a plain-language summary of what your students, teachers, and leaders contributed to Brilla's development, what the data showed, and what changed in the platform as a result of your feedback. Pilot schools are recognized as founding partners in Brilla's story.
Tell us about your school, your students, and what you are hoping Brilla can do. We will reach out within two business days.
Express Interest →The Family Portal is phone-first and requires no login. Families open a link from their weekly text message and land directly on their child's Brilliance Bulletin. No account. No password. Just their child's week at a glance, in their home language.
Weekly Brilliance Bulletin with the student's breakthrough moment, growth note, session stats, real-world home activity, and a mood signal families can send back to the teacher.
Full archive of all curriculum newsletters deployed by the teacher, organized by date. Labeled as module or lesson letters. Available in the family's home language where translations exist.
The full portal is available in English, Spanish, and French in the demo. The weekly text message also previews in each language. Additional languages are in development based on pilot partner needs.
The demo shows Sofia's family portal in full, including the Brilliance Bulletin, home activity, newsletter archive, and weekly text preview. Language switching is live.
Have a question about the pilot, the research, your state’s assessment framework, or anything else you saw on this site? We want to hear from you.
Opens in a new tab. Takes about 2 minutes.
Questions about the pilot, your state’s assessment framework, data privacy, partnership or collaboration, something you want to see built, or anything else on your mind.
Prefer email? Reach us at:
hello@aibrilla.com
The pilot is free. It takes 5 minutes to express interest and a real person follows up within 5 business days.