How we work

How the work is put together

We are not a consultancy and we do not sell decks. We build systems. Every piece of work has to run cleanly from something a person did to value coming back to them — and at any moment we can say exactly where a project sits on that line.

The spine

Activity → Evidence → Verification → Tokenization → Exchange → Regeneration

In plain words: do it, show it, check it, record it, use it, give back. Six steps, in order. We do not fold the checking into the recording, and we do not call an activity worthwhile before we have said what would count as proof. What each step means →

Stage 01Activity
Someone learns, teaches, makes, tends or cares for something. We help shape it so it can actually be finished, and so it can be seen.
Stage 02Evidence
Whatever shows it happened — work, logs, assessments, records. We help decide what counts, and how it gets captured and stored, with consent.
Stage 03Verification
Someone checks that evidence against criteria set in advance, with or without help from software. We design the rubrics and the workflow, and name the gaps.
Stage 04Tokenization
One Impact Unit goes onto the contributor's record. We say which kind applies, and we stay strict about it — units are earned, never handed out.
Stage 05Exchange
The checked result can now open doors: a marketplace, a recognition scheme, a policy. We never let our own units get confused with anything financially regulated.
Stage 06Regeneration
The loop that keeps the system going and keeps it fair. We work out how value returns to contributors and their communities, and we name anything extractive.
Foundation principles

Four principles behind every decision

When they get in the way of what is convenient or fast, the principles win.

i · Systems thinking first

Every problem is a system: what people are rewarded for, how they behave, what the rules say, and what happens next. Map it before proposing anything. Name who is involved, what moves between them, what loops back, and how it breaks.

ii · Indic & ethical lens

Rooted in Indic philosophy — balance, dharma, care for what we hold in common, and not doing violence to yourself — alongside plain modern ethics: ask permission, collect as little as possible, take nothing that is not offered, and treat every contributor with dignity.

iii · Open by default

Open formats by default, tools that work with other tools, built for public good. Anything that locks people in has to be argued for. What we write and design is Creative Commons unless there is a reason otherwise.

iv · Product + capacity

Every piece of work leaves behind something that runs and people who understand it. We judge it by what is still there after we leave.

Where the work is recorded

The three ledgers

Everything we do gets written into one or more of three ledgers — three separate books. We always know which one we are writing to. What each ledger holds →

SDG 4 · Quality EducationKnowledge
Resources, pathways, curricula, mentor time, and reusable learning artifacts.
SDG 16 · Strong InstitutionsImpact
Verified impact claims, milestones, evidence packs, and verification sign-offs.
SDG 8 · 11 · 13Growth
Personal and institutional evolution; capability maturity; governance development.
The four kinds of unit

One checked activity, one unit earned

When something spans more than one kind, we pick the main one and note the other in the evidence.

LIU

Learning Impact Unit

Teaching, tutoring, mentoring, running a class, writing curriculum.

SIU

Skill Impact Unit

Trade training, craft skill, professional upskilling, proven competence.

CIU

Cultural Impact Unit

Oral tradition, art, community ritual, keeping a language alive, passing culture on.

GIU

Green Impact Unit

Tending an ecosystem, caring for water bodies, growing trees, recording what changes.

Units cannot be sold or transferred. They are credentials — proof of what someone did — not investments. One activity earns one main unit, and we never pile several onto a single unchecked activity.

How claims get checked

Nothing gets recorded until it has been checked

We start at the lightest level the stakes allow, and never build third-party checking into a small first pilot. The four levels →

Level 0Self-reported
Verified by the contributor themselves. For early pilots, proof-of-concept, and low-stakes iterations — flagged clearly as unverified.
Level 1Peer verified
Verified by peers, mentors, or community reviewers. For knowledge listings, portfolio rubrics, and community learning circles.
Level 2Independent assessor
Verified by a qualified third party with defined credentials. For medium- and high-stakes programs, milestone grants, and funder-facing claims.
Level 3SSE-aligned
Verified through a regulated, SIA-equivalent workflow. For annual impact reporting and regulated disclosure.

For anything involving children, we need written consent from a parent or guardian before any AI touches their data, before any review, and before anything is recorded. This is a hard stop, not a guideline. The full policy is public.

The quality gates

Four tests everything has to pass

Could Sita Tai — a semi-literate woman in a Konkan village who teaches five children in her home to read — take part in this, have her work checked, and get real recognition for it, without being embarrassed, without a fight, and without depending on anything she cannot reach or afford?
— The Sita Tai Test. The architecture exists to make her contribution constitutionally visible.
  • The fifteen-minute test. Can a teacher run this in under fifteen minutes a week?
  • The ledger-ready test. Which of the three ledgers does this belong in, and what would make it ready to go in?
  • The real-claim test. Does this get us an actual checked result — or just a nicer framework for producing one someday?
System design rules

Non-negotiable

Agents propose; policy decides

Software can draft, suggest and model. It never writes a real record or a checked claim on its own. A person signs off, every time.

Evidence-first

We do not publish a claim until its evidence is filed, stored so it cannot be quietly swapped, and locked to the people allowed to see it. No evidence, no claim, no unit.

Everything is versioned

Everything we publish carries a version number and says where it came from. Nothing counts as the real version without a date and a source.

The line with financial regulation is absolute

Impact Units are not investments, and anything that is regulated runs on a separate track. We never build anything that blurs the two.

Scaffolding ≠ substance

A tidy framework with no real data, no decisions and no constraints is not finished work. It is preparation for finished work, and we say so.

It has to be deployable

If all a session produces is more structure and nothing that meets the real world, we flag it. The point is a checked result, not a prettier diagram.

See it at work

The programs are where all of this meets the real world.