Skip to content

Strategy and values

Where we play

Kwilo is B2B2C in India.

  • B2B, primary. Higher education (engineering, degree, universities) and corporate training. K-12 and PU colleges are supported but not where we lead.
  • B2C, secondary. School learners and JEE/NEET aspirants, plus parents. Self-serve, freemium.

The institutional problem

Indian institutions don't lack software. They drown in it. The pains we hear and read about most:

  1. Faculty turned into data-entry operators: attendance portals, parallel registers, spreadsheets.
  2. Timetables built by hand in Excel over weeks, breaking every time one lecturer changes.
  3. Fee collection as a monthly chasing ritual, with reminders sent to parents who already paid.
  4. Accreditation prep eating months of staff time, with the same data re-entered for NAAC, NBA and NIRF.
  5. Result-day errors at university scale.
  6. Admission enquiries that die before the second follow-up.
  7. WhatsApp acting as the real ERP.
  8. ERP projects that fail because staff won't adopt them.

Our answer

One record, three products, start with one. Campus, Classroom and Compliance share data, so compliance evidence is a by-product of running the college rather than a separate project. An institution can buy the part it needs first.

The system does the work. Competitors sell software staff must operate. We're moving Kwilo toward an ERP that acts: a solver that builds and repairs timetables, allocation that staffs invigilator duty, and next, agents that draft fee nudges and attendance alerts for a human to approve. Existing modules become the agent's tools. People stay in the loop for anything consequential.

AI where it's strong, math where it's not. LLMs grade against rubrics, draft content and explain. Constraint problems like timetabling go to a proper solver, because a confident wrong timetable is worse than none.

Wedge: handwritten answer sheet grading. It's the most concrete time saving we can show a college in a demo, and it pulls in courses, assessments and outcome data behind it.

The learner bet

The moat is the learner model: Kwilo remembers what a learner knows, topic by topic, over months, and uses it to pick what comes next. Kwi, the tutor persona, is how the learner feels that memory. Every surface that asks a question should record the attempt.

We build it in stages that each ship alone: rule-based mastery first, then a spaced review queue, then difficulty steering. The concept graph comes last, and only if retention proves the cheaper stages work.

Monetisation follows the taste-then-volume rule: free learners get real value, Pro removes limits and unlocks downloads at ₹399 a month.

Go to market

  • Institutions: demo-led sales. The site leads with the three products and routes buyers to a demo form. Solution pages target specific tool queries (handwritten grading, CO-PO tracking, DPDP proctoring) rather than head terms we can't rank for yet.
  • Individuals: organic and word of mouth into a free signup. Every site-to-app link carries attribution so we know what converted.
  • Mobile: Android first. Kwilo Learn for individuals; Kwilo Campus for faculty and institutional learners, in Play internal testing.

What we value

These are the rules we actually apply when a product decision is close. Each one shows up in shipped behaviour, and the examples say where.

Educators decide

AI drafts, suggests and ranks. A person signs off anything that counts. AI grading returns a suggested mark with a confidence band and the trainer accepts or edits it. Physical exam marks are finalised per sheet by a human. Consequential automation (messages to parents, money) waits for approval.

Enter it once

One record serves every product. An enquiry prefills the registration form. Registration demographics carry onto the student record at admission. Registration fees land in the same receipt sequence as every other fee. If a feature needs someone to retype data that already exists, it's the wrong design.

Never fake it

We don't show a number we haven't earned. The Practice Hub shows a dash for accuracy until a learner has attempted a chapter, and returns an honest empty state instead of a made-up plan. The landing page carries no invented ratings or press logos. Site capability grids mark what's available and what's rolling out.

Private by default

  • No analytics or session replay until the visitor accepts cookies.
  • Parental consent before a minor's face is enrolled for attendance.
  • We store APAAR ID, not raw Aadhaar.
  • Every institution's data is isolated at the database level with row-level security.
  • B2C users can pause or delete their account themselves. Deletion anonymises PII immediately.
  • AI-generated content is labelled.

Built for India as it is

Phone OTP is the default signup. Pages are prerendered to paint fast on 3G. Answer sheets are compressed on the phone before upload. Fee receipts print amounts in Indian-system words. Statutory codes (UDISE+, AISHE, affiliation numbers) are first-class fields.

Taste before the paywall

A free learner uses the real thing before being asked to pay. They practise the top chapter of every subject, hold a real study plan and preview every generated artifact. Pro sells volume and downloads, not entry.

Plain words

We write like a colleague who has seen the problem before. See Voice and tone.

What we're not doing

  • Winning accreditation for anyone. We keep the evidence straight.
  • Storing raw Aadhaar.
  • A trainer product for individuals. B2C trainer signup is held behind a waitlist while Learner gets our focus.
  • Per-feature AI limits for free users. One monthly generation pool covers everything.