Monthly long-form ยท founder voice ยท published the first Tuesday of each month

Essays.

What the post-BYJU's Indian K-12 AI landscape actually looks like from inside it. What schools should ask vendors. What DPDP changed. What the regulators didn't say out loud. Written by Bishwarup, edited by no-one.

Coming up โ€” the monthly calendar

Each month, one long-form essay. Topics drawn from what we're actually seeing in partner schools, what regulators are publishing, and what founders are asking on calls. The titles below are committed; drafts are in progress.

Jun 2026Curriculum

The CBSE Class 3 AI mandate โ€” what it actually says, and what it doesn't.

A line-by-line walkthrough of the 30-October-2025 notification + the April 2026 rollout, separating what's compulsory from what's interpretation.

Read โ†’
Jul 2026Safety

What the AI flags when your child types "I want to die" โ€” and what happens next.

The three-tier escalation protocol, the counsellor-notification cadence, and why the school never sees the actual words.

Read โ†’
Aug 2026Vendor due-diligence

30 questions to ask every K-12 AI vendor before signing โ€” the unedited version.

Hosting, data residency, DPA terms, training-on-student-data clauses, content-filter pass-rates, escalation SLAs, the works. With model answers.

Read โ†’
Sep 2026Policy

Why your school's AI use policy should be three pages, not thirty.

Trustees love long policies. Teachers ignore them. Parents don't read them. Here's the three-page version we ship โ€” and why it works.

Read โ†’
Oct 2026Field notes

Six months in Silchar โ€” what running AI literacy in Assamese-medium CBSE schools taught me.

The literacy gap is real. The mother is the gatekeeper. The principal listens to the trustee. Three things I got wrong in the first month.

Read โ†’
Nov 2026Procurement

The price of "free" AI for schools โ€” what the consumer-AI vendors don't tell principals.

The economics of free-tier AI for schools, who pays (your child's data), and why the school's "we're saving money" calculation is wrong.

Read โ†’
Dec 2026Year in review

2026 โ€” the year DPDP woke up Indian schools.

A timeline: what happened, what stuck, what was theatre, and what changes in 2027 when the Children's Data rules go live.

Read โ†’
Jan 2027State boards

Why Karnataka, Chhattisgarh, Assam, and Tripura are the next four state-board AI markets โ€” in that order.

The state-board landscape is no longer "wait for CBSE." Four state-board mandates already exist. Here's the deep map.

Coming โ†’
Feb 2027Parents

The "FOGS" decade โ€” why post-BYJU's parents stopped buying ed-tech and what they buy instead.

Fear Of Getting Scammed replaced Fear Of Missing Out. Here's what that means for any B2B-school AI vendor.

Coming โ†’
Mar 2027IB / IGCSE

What the IBO's March 2026 Mumbai principles actually mean for Indian IB schools.

Appendix 6 wasn't a mandate, the Mumbai principles weren't either, but together they're the closest thing the IB has to an AI standard.

Coming โ†’
Apr 2027Govt schools

The three stacked mandates in govt schools โ€” NEP + NCF + MoE โ€” and what each actually funds.

National Education Policy, National Curriculum Framework, MoE 30-Oct-2025. The funding doesn't follow the framework. Why.

Coming โ†’
May 2027Children's data

What changes on 13 May 2027 โ€” the day DPDP Children's Data rules go live.

Verifiable Parental Consent flow, the Schedule IV exemption, the Significant Data Fiduciary threshold. A school-by-school countdown.

Coming โ†’
Jun 2027Teacher

What teachers actually want from AI โ€” and how every vendor (including us) keeps getting it wrong.

The "AI for personalised learning" pitch is a teacher-replacement pitch in disguise. Here's the pitch that lands.

Coming โ†’
Jul 2027Open source

Why we're open-sourcing our policy templates after 25 schools โ€” and not before.

The pre-25 conservatism, the post-25 commitment, the safety-classifier follow-on in 2027. A long answer to "but won't competitors copy?"

Coming โ†’
Aug 2027Indic NLP

The Indic NLP safety gap โ€” why no English-trained model is safe for Bengali / Tamil / Marathi children yet.

What the current models miss, what we're building, why this is the moat.

Coming โ†’
Sep 2027Vendor consolidation

Why the next 5 years of Indian K-12 AI will look like 2002โ€“2007 of Indian SaaS โ€” and what that means.

The Freshworks / Zoho / WebEngage / Postman pattern, applied to schools. Long, cheap, B2B, and won by founders who stay.

Coming โ†’
Oct 2027Field notes

Year-one with 100 schools โ€” what we shipped, what we got wrong, what we're rebuilding.

An honest retrospective. Numbers, decisions, regrets, lessons.

Coming โ†’
Nov 2027International

UAE + Singapore + Gulf โ€” the Indian-curriculum school networks outside India.

The 1,200+ Indian-curriculum schools abroad. What's different about selling there. What's the same.

Coming โ†’
Dec 2027Year in review

2027 โ€” the year the safety-classifier and the policy-library both went open.

A year-end retrospective. What stuck. What broke. What we're betting on next.

Coming โ†’
Jan 2028Investors

What investors get wrong about Indian K-12 AI โ€” and what the smart ones get right.

The "TAM is huge, ARPU is tiny" trap. The B2B-only thesis. The DPDP moat. A long letter to anyone who's pitched on us.

Coming โ†’
Feb 2028Trustees

A short letter to the school trustee who keeps asking "what's our AI strategy."

Five paragraphs you can forward to your trustee. Plain English. No buzzwords. No "transformation."

Coming โ†’
Mar 2028Two-year anniversary

Two years of Deshika โ€” what stayed true, what changed, what surprised us.

The principles we wrote on day one. The ones we still hold. The ones we revised.

Coming โ†’

One thoughtful essay a month.

Published the first Tuesday of every month. No marketing. Unsubscribe in one click. Topics: post-BYJU's Indian ed-tech, DPDP enforcement, child-safe AI, the Indic safety gap, what the brain is teaching us.

Subscribe to the monthly