Commercial AI extracts
An extraction tower: it pulls value from users and optimises for engagement, profit, scale. The user is the product.
Socratus · Digital PracticeWe build the technology that lets community and institutional knowledge surface, travel, and return.
An extraction tower: it pulls value from users and optimises for engagement, profit, scale. The user is the product.
An instruction tower: well meant, but top down, experts decide what communities need, then build for them.
A horizontal commons: communities already hold what matters, so we surface it and help it travel between them, peer to peer.
Asked about forest foods for state nutrition programs, communities didn't just say “include them.” They proposed value-added forms, Mahua laddoos, Kaku powder, that are child-friendly, culturally familiar, and easy for the state to distribute. The wisdom is in that detail, and a tick-box survey strips it away.
1.4 billion people. Most of their knowledge has no channel to flow through.
Odisha's forest department worked out how to restore both forest incomes and the ecosystem, pairing river desilting with reopened honey-gathering, instead of fencing people out. Every protected reserve in India hits the same wall. But that solution sits in one district's files, and the next reserve starts from scratch.
The same hard-won lessons get learned over and over, and never shared.
Using AI for public good: not private wealth, and not the well-meant, top-down nonprofit model where experts decide what communities need. Hybrid, humble, and oriented to people and places.
A tool that unleashes human connection and creativity, not a central oracle that predicts and manages us.
The application layer is cultural infrastructure, the way the newspaper and the novel were to print. Not frosting on a foundation-model cake.
Spoken knowledge. Report anything, from anywhere, in any language.
Geo-spatial knowledge. Any neighbourhood becomes a living archive.
Written knowledge. Decades of fieldwork, made conversational.
Connects all three, and surfaces the patterns no single source could see.
Sense, synthesize, deliberate, design, deploy, learn: the loop closes when local knowledge travels, shapes a decision, and returns in a form people can use.

Report anything, from anywhere, in any language. Each report is a small act of citizenship; a campaign gathers thousands. Launching one takes four steps.
Name the wisdom you are looking for: a need, a practice, a place.
Any of 22+ languages and the dialects beyond; any geography.
Share a link or handle. Anyone can speak up from a phone.
Voices arrive, are transcribed and translated, and become a Data Story you can act on.
Voice-first · Multilingual · Web-first

voices recorded · land rights, livelihoods, governance · five states
50+ voices on the question “How can we make our villages wiser?”
ರೈತ ಯಾವಾಗಲೂ ಸ್ವಾಭಿಮಾನದಿಂದ ಬದುಕಬೇಕು, ಯಾರ ಮುಂದೆಯೂ ತಲೆಬಾಗಬಾರದು. ರೈತ ಈ ದೇಶದ ಬೆನ್ನೆಲುಬು; ಆ ಗೌರವ, ಆ ಸ್ಥಾನಮಾನ ಅವನಿಗೆ ಸಲ್ಲಬೇಕು.

Local Open Knowledge Access. Turn any neighbourhood into a living archive, readable to itself.
“We cannot respond to what you do not notice. We cannot deliberate about what you cannot perceive.”
Discover and exchange local knowledge: the neighbourhood as a living archive.
A “digital twin” where every marker is a story, a lesson, or a call to action tied to a real place.
A mobile app that uses AI to organise and connect knowledge, wherever you stand.
Pilots: Bhitarkanika, mapping the national park and its resources · Melukote, promoting local eco-tourism.


A resident photographs a roadside bloom and adds one line: “Flower cluster of the Silver Oak tree.” LOKA identifies the species and turns the sighting into a structured, searchable record:
Structured biodiversity data from everyday sightings.
Understand what is around you, and why it matters.
Unravel the world's objects and share what you find.
Knowledge gathered is not knowledge used.
Field notes leave with the staff who hold them. Patterns across years of fieldwork never get synthesised. Reports get filed and forgotten.
Named after the fungal networks that thread a forest with no centre, Mycelium is the engine that reads documents across the ecosystem and weaves them into connective tissue: a living graph where organisations, events, insights, and signals become linked nodes.
“What have we learned about women's market access in Odisha?”
Answered from 20 years of fieldwork, not two weeks of document review.
Collective wisdom lives not only in space, scattered across a community today, but in time, across everything an organisation has already learned.


Wisdom + systems. The knowledge commons itself: a container, not a dashboard.
It holds the corpus and integrates every source, voice through VoiceReport, place through LOKA, documents through Mycelium, into one searchable commons, writing Articles, Recipes, Findings, Data Stories, each cited to its source.
Every claim has a source
Organisations control what publishes
No anonymous claims
Live now 353 orgs · 236 signals · +126/week

Mycelium builds one knowledge graph across every organisation's documents. Wystem surfaces it at three levels, from a single organisation out to the whole ecosystem.
Articles and Recipes drawn from a single organisation's corpus.
The questions and voiced demands the ecosystem is raising. What people are asking for.
Patterns that span many organisations, wisdom no single org could see alone, surfaced so the next one never starts from scratch.





782 voice submissions across four states, spoken in Odia and Hindi, transcribed and translated, flowed into Wystem.
It surfaced what no single voice could: communities four states apart independently named the same 20-year extinction window, and the same answers, heritage “camps” and ₹20,000 micro-grants for women.





Each Data Story is generated by Wystem from a VoiceReport campaign, and credited back to the voices that made it.


Wystem read two field documents from Socratus, found a repeatable method buried inside, and catalogued it: a four-step recipe with context, approach, and steps, scoped entirely to those two sources. Now any group facing the same problem can pick it up, instead of starting over.
The method itself: speak to the crocodile, scold it like a child, project calm, and treat the animal with respect rather than lethal force.
We surface what communities know; we do not improve or instruct them.
Preserve the texture and lived experience while making it navigable at scale.
Whoever is hardest to reach should find it easy to be heard. We design with the unheard in mind.
It cannot be manufactured by better algorithms. It has to be earned.
“The willingness to share what you know is the most valuable thing in the system. No amount of engineering can manufacture it.”
Live VoiceReport campaigns: Atmashakti's Indigenous Yatra and Melukote Vani. Voices recorded to date: 300K across the Melukote constituency, 12K on indigenous rights, 10K commons stewards, 5K climate-atlas panchayats.
LOKA mobile app with AI-tagged spatial discovery; pilots in Bhitarkanika and Melukote.
Wystem live: 353 organisations, 159 projects, 471 published insights, and 236 signals, with continuous synthesis; multilingual across 8 locales.
At scale, across 22+ languages.
Of any place or campaign.
Articles, Recipes, Findings, Data Stories.
Patterns no single org could see alone.
Every claim cited; orgs control what publishes.
The systems behind all of it.
Shared infrastructure for the sector, a commons your practice helps build and own.

Show up and use the platforms. The commons is only as alive as its use.
Hold them as shared infrastructure, not someone else's product.
The consent and relationships that let wisdom travel safely.
Resources to keep building and maintaining what we all rely on.
In return: your knowledge made searchable, cross-org patterns on demand, voice and geo intake as a shared service, and a public, citable commons in 8+ languages.
“The answers to our hardest problems already exist, with the people living them. Our work is to help that wisdom surface, travel, and be heard.”
The Digital Practice Rajesh Kasturirangan · Manoj Kurien · Srinivas Mangipudi · Mithun Sheshagiri · Ashrith Shetty · Smriti Tiwari · discoverloka.org · voicereport.in · wystem.ai