Every Student Gets a Speaking Partner. Every Teacher Gets to See It.
The individual product provides real-time spoken practice with an AI tutor. An institutional version is only a planned concept: no university, school, academy, or pilot can be offered until the provider, legal, reporting, safeguarding, and load gates described on this page are complete.
Discovery only · No institutional deployment or price offered · 6 individual-product languages
The Problem Is Arithmetic, Not Teaching
One teacher, twenty-five students, fifty minutes. Divide it out and each student produces spoken language for a minute or two per class — through no fault of the teacher. Speaking is the skill exams grade and employers hear, and it is the one skill a group class structurally cannot supply enough of.
The individual product supports one-to-one practice. Teacher visibility is a planned institutional capability and is not available today.
Built for the Institutions That Teach Languages
Three planned institutional paths. None is deployable until the provider and legal gates are complete.
Universities & Language Centres
Thousands of students face accredited-level requirements and mobility deadlines; conversation classes cap at a dozen seats. Give exam-bound and Erasmus cohorts daily speaking practice, with evidence your language centre can put in front of its own stakeholders.
- A future pilot design could align with an academic calendar
- Speaking-exam preparation with rubric-graded assessments
- Adult cohorts remain provider-authorization gated
Schools & Education Authorities
For ministries, regional authorities, and school networks reviewing digital solutions for foreign-language teaching: capped per-student budgets a procurement office can actually plan around, teacher visibility, and an honest, written account of what exists today versus what is roadmap.
- A bounded commercial model would require a future written agreement
- All cohorts are provider-gated; minors require express authorization and safeguards
- Complements — never replaces — teachers and assistants
Language Academies & Official Schools
Your students pay for outcomes, and speaking is the outcome they judge you on. A future authorized path could add structured daily conversation practice between lessons. Course branding, instructor reporting, and seat licensing are product and contract concepts rather than deployed capabilities.
- Per-seat licensing is a proposed future model
- Exam-prep scenarios: IELTS-style interviews and more
- Start timing is agreed only after provider, legal, product, and load gates pass
Practice That Compounds, Not a Toy That Talks
The individual product's learning loop exists today. Institution-level oversight and reporting do not; those remain planned work before any authorized pilot:
1. Speak and be corrected
Free-form spoken conversation in realistic scenarios, matched to the student's placement level. Corrections arrive in the moment the error is made — when the student can immediately retry the sentence.
2. Every mistake is logged
Each error lands in the student's personal Mistake Ledger, categorised by type — grammar, vocabulary, pronunciation — building a precise map of what this student, specifically, gets wrong.
3. Drilled from their own errors
Micro-drills are generated from the student's own logged mistakes — never a generic worksheet. Days later, spaced probes re-test old errors in new contexts; a mistake only retires when the student proves it stayed fixed.
4. Planned teacher evidence
Before any institutional launch, class-scoped roles and reports would need to be built and verified: who practised, how many minutes, which error categories dominate, and whether corrections are sticking. No teacher report or export is available today.
Preview a reflection prompt for speaking practice
The public score tool lets educators and adult learners examine frequency, challenge, feedback, and reflection without creating an account or sending the answers to LinguaLive.
What Exists Today vs. What Is Roadmap
Most vendors will show you their roadmap and call it their product. We keep the columns separate — and we suggest you demand this table from everyone you evaluate, including us.
Individual adult product today
- Live voice tutor in 7 languages on web and iOS; not authorized here for institutional use
- Scenario library + custom scenario authoring
- Placement test, Mistake Ledger, personalised drills, spaced re-testing
- Rubric-graded speaking assessments
- Individual daily usage caps, enforced server-side
- Privacy policy that names our AI provider plainly
Required before any pilot
- Build and verify class access codes and bulk seat provisioning
- Build and verify pseudonymous account enrollment
- Build and verify class-scoped reports and approved export formats
- Build and verify per-organisation data and replay controls
- Build and verify teacher roles and scenario assignment
- Complete documented load and safeguarding acceptance tests
On the roadmap
- Live teacher dashboard (reports ship first; dashboards follow evidence)
- EU-resident AI processing
- Formal data-processing agreement pack for schools
- Provider authorization for the intended institutional use
- Under-18 cohorts only with express provider permission, age safeguards, residency, and a school-grade DPA
- Single sign-on and platform integrations
Only the first column describes current individual-product functionality, and even that is not an institutional offer. The other columns are prerequisites or roadmap; we will not claim they have shipped before they are independently verified.
The Two Questions Every Evaluator Asks First
"Where does student voice data go?"
Speech is processed by Google's Gemini service, which we name in our privacy policy. With session resumption enabled, Google documents that Live conversation state may be retained for up to 24 hours, and provider abuse-monitoring records may be retained for up to 55 days. LinguaLive correction-replay cleanup targets clips older than 30 days on a best-effort basis, and account deletion immediately attempts an owner-scoped purge. Per-organisation replay controls are not deployed. Institutional deployment remains blocked until provider authorization and data terms are confirmed.
"What will this actually cost?"
No institutional price or contract is currently offered. If authorization is obtained, the proposed model would define the cohort, term, usage cap, price, and any overage treatment in a written agreement before access is enabled.
What a Future Pilot Could Look Like
Illustrative future structure only: work starts after provider authorization, legal review, and written acceptance gates. One term or semester. A defined cohort — an exam-prep group, a mobility intake, or a teacher-training programme. Fixed price. Defined exit.
- Weeks 0–2 — setup. Classes provisioned via access codes, teachers onboarded (one hour is enough), placement tests run, classroom load tested against the agreed acceptance criteria.
- During — daily practice, fortnightly evidence. Students speak every day inside their caps; teachers receive per-class reports every two weeks.
- Mid-point — the honest review. Full data against the four agreed measures. If it isn't working, this is the built-in moment to stop — payment milestones are structured around it.
- End of term — outcomes report. Pre/post assessment movement, correction-to-mastery rates, teacher survey — plus a costed per-seat proposal for the following year, priced on your cohort's real usage.
This is pre-authorization discovery, not a quote, launch commitment, or offer of institutional access.
Institutional FAQ
Is there anything to install?
Once provider and legal authorization is complete, the planned institutional path runs in the browser rather than requiring an app rollout or MDM project. Class-code access and pseudonymous pilot accounts are implementation commitments for an authorized pilot, not capabilities available to deploy today. (An iOS app exists for individual adult learners.)
Which languages does LinguaLive teach?
Seven languages today: English, Spanish, French, German, Italian, Japanese, and Korean. A future authorized institutional deployment would normally start with English speaking practice, with the same account covering the other languages at no extra cost.
Is student voice data stored? What does the AI do with it?
Speech is processed by Google's Gemini service. Because session resumption is enabled, Google documents that Live conversation state may be retained for up to 24 hours; provider abuse-monitoring records may be retained for up to 55 days. LinguaLive may also save short, private correction replays, whose cleanup targets clips older than 30 days on a best-effort basis. Per-organisation replay controls are planned rather than deployed. Institutional use remains blocked until the applicable provider authorization and data terms are confirmed.
How much does it cost?
No institutional product, pilot, or price is currently offered. If the provider and legal gates are resolved, the planned commercial model would use a defined cohort, term, usage cap, and written price rather than an open-ended usage commitment. Any actual offer would be documented in a future institution-specific agreement.
Can it be used with students under 18?
No. New institutional cohorts—including adult cohorts—remain blocked until Google and counsel confirm an authorized provider path for the intended use. Any future cohort that includes minors additionally requires provider authorization that expressly permits under-18 access, enforceable age and institutional safeguards, and the required residency and school-grade data-processing agreement. EU residency or a DPA alone is not sufficient.
Does it replace teachers or conversation assistants?
No, and we decline to be positioned that way. The planned institutional concept is a practice room between classes: it would add speaking repetitions and, only after the reporting work is built and verified, give teachers bounded progress evidence. Curriculum, assessment judgement, and the human relationship would stay with your staff.
Can teachers create their own scenarios?
The individual product already includes custom scenario authoring. Teacher roles, class-level assignment, and institution-scoped controls are planned capabilities, not features available for institutional deployment today.
How do we know whether it worked?
The proposed evaluation framework has four measures: minutes of speaking per learner; the share of logged mistakes later passed in spaced re-tests; pre/post rubric-graded speaking assessments; and a structured teacher survey. Class reports, institution exports, and outcomes reporting still have to be built and verified before any authorized pilot.
Help Us Understand the Speaking-Practice Gap
If you are evaluating future speaking-practice tools, write with your requirements. We can share the current product boundary and authorization prerequisites, but we cannot offer a pilot, institutional access, launch date, or price yet.
Universities & language centres · Education authorities · Academies & official language schools