One platform. Costs you can see. Knowledge that stays.
Governing AI means knowing what it costs, what it emits, who uses it and what it answers from. iYan replaces scattered subscriptions by one governed platform: budgets, routing to the right model, a carbon figure on every answer, and a knowledge base your teams share.
1
platform instead of five subscriptions
CO₂
estimated on every answer
5×
usage levels per plan, no surprise
100 %
of answers from your documents cite their sources

What an ungoverned AI stack costs.
Not only in dollars: in time, in duplicated knowledge and in decisions nobody can explain.
Subscriptions that add up
A chat tool, an automation tool, a document tool, a scheduler, a support inbox: each with its seats, its invoice and its own copy of your data.
No view of consumption
Who used which model, for what, at what cost? Most organizations cannot answer, so they cannot set a budget either.
Knowledge scattered
Procedures live in inboxes and personal folders. People recreate what exists, and knowledge leaves with the person.
An invisible footprint
Each premium request has an energy cost. Without a figure, nobody can choose a lighter model when it would do.
Consolidate, measure, share.
The three things a governed platform must do for you.
One stack instead of fragmentation
Chat, knowledge base, boards, studio for documents and media, automations, functions, dashboards and connectors in one place, under one identity and one audit journal.
Measured and budgeted usage
Every request carries a cost in units and a CO₂ estimate. Quotas per plan and per user, consumption history and a monthly forecast in each profile, model allowlists per organization.
Knowledge that stays with the organization
Documents are indexed locally, answers cite their sources, folders are shared by permission, and the knowledge stays in your perimeter when people move on.
Spend the right token on the right task.
The mechanisms that keep the bill and the footprint down.
- 1
Fast model by default
Routine work runs on a light model; frontier models are chosen on purpose, per task, per plan.
- 2
Cost in units
Each model declares how many units a call costs. Heavy models consume the quota faster, so the trade-off is visible before it is paid.
- 3
CO₂ per answer
A carbon estimate per thousand tokens, per picture or per video is shown with every result and summed in the consumption history.
- 4
Quotas and budgets
Plans carry levels of usage; administrators set per-user quotas, allowed models and feature access. No bill shock.
- 5
Decisions explained
A local decision engine decides when to ask a question, when to upgrade the model and when to stop. Its decisions are logged.
- 6
Automations instead of repeats
Scheduled or event-driven workflows with retrieval, masking, OCR and notification steps replace repeated manual prompts.

Centralise what your teams know.
Drop documents, folders or whole archives into the vault: PDF, office files, scanned pages read by OCR, transcripts. They are extracted, chunked and indexed with embeddings inside your perimeter, and searchable in natural language.
Every grounded answer cites the passages it used, so a reader can open the source. Folders are shared by invitation with precise rights; boards, dashboards and shared files follow the same rules.
Prompts, functions and automation templates live in catalogues an administrator curates: the organization's way of working is captured once and reused, instead of living in a few people's heads.
From five tools to one governed platform.
What a migration looks like.
- 1
Inventory
Which tools, which seats, which data, which real cost.
- 2
Perimeter
Hosting choice, allowed models, masking rules, audit retention.
- 3
Knowledge
Import documents, set folders and rights, validate first answers.
- 4
Teams
SSO, quotas, prompt and automation catalogues, a short training.
- 5
Measure
Consumption, CO₂ and adoption dashboards from the first month.
Before and after, in a 15-person team.
Indicative, based on the migrations we run; your figures come from the inventory.
| Indicator | Scattered tools | iYan |
|---|---|---|
| Subscriptions to manage | 4 to 6 | 1 |
| Places where your data lives | 4 to 6 | 1 |
| View of who spends what | ||
| Carbon figure per request | ||
| Answers with cited sources | Depends | |
| Knowledge kept after a departure | ||
| Audit journal across tools |
Frequently asked questions
How does iYan reduce AI costs?+
Three ways: one subscription replaces several tools; routine requests run on a light model and only chosen tasks use a frontier model; quotas and budgets stop uncontrolled use. The unit cost of every model is visible in the administration.
How is the CO₂ figure computed?+
Each model carries an estimate in grams of CO₂ equivalent per thousand tokens, or per generated picture or video, set by the administrator from published measurements. It is shown with the answer and accumulated per user per day. It is an estimate, not a measurement of your servers.
Can I cap what a user or a team spends?+
Yes. Plans carry usage levels; administrators set per-user quotas, allowed models, and which features a user can access. Consumption and a monthly forecast are visible in every profile.
Which documents can the knowledge base read?+
PDF, office documents, text and markdown, images and scanned pages through OCR, audio transcripts, and whole archives. Documents are indexed inside your perimeter; nothing is sent elsewhere for extraction unless you choose a cloud model.
How are answers kept trustworthy?+
Grounded answers cite their sources; the knowledge base keeps versions; a local decision engine asks a question instead of guessing when the request is ambiguous, and the whole exchange is in the audit journal.
What happens when someone leaves?+
Their folders, boards and automations are owned by the organization and can be reassigned. Access is revoked in one place through SSO. Nothing lives in a personal account outside the platform.
Govern AI instead of chasing it.
An assessment with your team: inventory of tools and costs, perimeter, first knowledge sources, and the governance rules to set.