Example sessions across grades and subjects — how Saya drives deeper thinking, broader participation, and measurable pedagogical impact.
These case studies are drawn from South Asian classroom contexts where Saya was first deployed. The pedagogical patterns — IRE cycle disruption, Bloom level distribution, participation breadth — apply identically in classrooms following Cambridge, IB, GCSE, CBSE, WAEC, or any other framework. The curriculum names change. The learning science does not.
The scenarios below are illustrative examples based on Saya's design and session architecture. Student names are fictitious.
Primary · Mathematics
Grade 5 · Classroom
Starting challenge
Only 3 of 28 students were contributing to discussion. The class had completed the worksheet correctly but couldn't explain why the rule worked.
Session extract
We've established that ½ and 2/4 are the same. Can anyone tell the class why?
Because if you multiply both by 2 you get the same thing.
Hamza, that makes sense for 2/4 — but what if I said ½ equals 50/100? Would the multiply-by-2 rule still explain it?
Oh — you'd have to multiply by 50 instead.
So does the rule change depending on which fraction you're comparing to, or is there a deeper pattern?
(spontaneous discussion begins)
Outcome: Participation breadth rose from 3 voices to 11 within 8 minutes. Bloom level shifted from recall to analysis without the teacher re-explaining the concept.
Secondary · History
Grade 9 · Classroom
Starting challenge
Students were listing causes from the textbook without connecting them. Discussion was flat — correct answers, no reasoning.
Session extract
We've listed four causes: nationalism, imperialism, militarism, and the alliance system. Which was most important?
The assassination of Archduke Franz Ferdinand.
Aisha, historians sometimes call the assassination a spark rather than a cause. What's the difference — and does it matter for how we assign responsibility?
A spark needs fuel. So the fuel was already there.
If the fuel was already there, what would have ignited it if the assassination hadn't happened?
Outcome: The class moved from listing to causal reasoning. Three students who hadn't spoken in 20 minutes entered the discussion organically after Saya's second question.
Upper Secondary · Ethics
Grade 12 · Classroom
Starting challenge
A polarised class — two confident voices dominated. The teacher wanted more positions surfaced without escalating tension.
Session extract
[private] This topic is politically sensitive. Use quiet voice to bring in the students who haven't spoken.
read: Two students are debating loudly — most others are disengaged. risk: none. suggest: Use quiet voice toward the back row — they look like they have views but don't want to enter the current dynamic.
I'm curious — does anyone think the answer to where free speech ends changes depending on which country you're in, or is there a universal principle?
I think context always matters. What's acceptable in one place can be dangerous in another.
Outcome: Seven students contributed who hadn't spoken before. The teacher used the private channel twice to redirect Saya's focus — both commands executed correctly.
Home Learning · Science
Grade 7 · Home Companion (parent-led)
Starting challenge
Parent wanted to support her child's revision without dominating the session. Child was memorising definitions rather than understanding the process.
Session extract
Sara has been studying photosynthesis all week. She can recite the equation but I'm not sure she understands what each part means.
Sara, what do you think the plant is actually trying to do — what problem is photosynthesis solving for it?
It needs to make food?
Yes — so it's making its own food instead of eating it. What does it need to make that food, and where does each ingredient come from?
Water from the roots... and CO2 from the air... and — oh, the sun gives the energy to put them together.
Exactly. So what do you think happens to a plant in a dark room — not immediately, but over a few days?
Outcome: Sara built the conceptual model herself rather than reciting it. Parent reported she asked follow-up questions independently the next day.
Saya's design is grounded in decades of classroom participation research. These are the core concepts that shaped its architecture.
Initiate–Respond–Evaluate is the dominant interaction pattern in most classrooms: teacher asks, one student answers, teacher evaluates. Research consistently shows this pattern concentrates participation in 3–5 students and suppresses independent reasoning. Saya is specifically designed to disrupt IRE by entering after a student responds — asking a follow-up rather than letting the teacher evaluate — which redistributes the conversational floor.
Cazden (2001), Sinclair & Coulthard (1975)
Bloom's revised taxonomy identifies six cognitive levels: remember, understand, apply, analyse, evaluate, and create. Most classroom interactions cluster at the first two levels. Saya's architecture targets Level 3 and above by design — every response is classified by a local skills engine before any AI call is made, ensuring the tier of engagement is intentional rather than random.
Anderson & Krathwohl (2001)
In a typical 30-student classroom, research shows that fewer than 8 students account for 80% of verbal contributions. Saya tracks speaker attribution per session and uses quiet-voice interventions and strategic silence to create entry points for under-represented voices — without singling out students or creating embarrassment.
Tobin (1987), Rowe (1986) — wait time research
Teacher wait time research shows that extending the pause after a question from 1 second to 3 seconds significantly increases the number and quality of student responses. Saya extends this further — it can remain silent for an entire exchange if the teacher has issued a silence command, modelling for students that thinking time is valued.
Rowe (1986)
Every session generates a full analytics report. These are the metrics Saya produces and what they tell you.
Proportion of the class who contributed at least once. Saya tracks unique speaker names per session.
Every message — from teacher, student, and Saya — is classified across six Bloom levels by a local skills engine.
Teacher, student, and Saya contributions measured as % of total word count.
Session-level aggregate of language patterns classified as energised, engaged, flat, or strained.
Saya's single most important suggested action for the teacher's next session, generated at session end.
We are building a library of real classroom outcomes from teachers using Saya across subjects and grade levels. If you have run sessions and seen measurable impact, we would like to feature your experience.