Leitagu knowledge hub

Leitagu is a web app that helps people in Estonia find furniture from Estonian stores that goes with what they already own, fits their room and is available when they need it. You describe what you need or add your room, answer a few short questions, and get matching products with price, size, stock and a link to the store. This page explains Leitagu in brief and maps the rest of the hub.

#What Leitagu is

The name is Estonian for "let it be found". The product was called Sobib before it was renamed.

Buying one more piece of furniture to go with what you have is slow. Wood tones and finishes are hard to compare from photos, every store names colours differently, and the products are spread across many websites. On top of that, you need to know whether the piece fits the wall and whether you can get it this week or only in two months.

Leitagu works as a chat. It asks up to four short questions (what, what size, what budget, how soon), understands free text in Estonian and English, and searches one catalogue of products from several stores. It compares tone, undertone, material, finish, style, legs, handles and fronts one by one, groups the results by how close they are, and explains each match in one line. You can model your room from a template, in an editor or from a photo, see a product in a 3D view of the room, and ask what else would fit in the free wall and floor space.

Leitagu does not sell furniture or take orders. Every product links to the store's own product page.

#Where things stand (October 2026)

• Built.
  • The shop (apps/web): accounts, several chats per person, reusable rooms, matching, "what should I add" suggestions, a 3D room view, availability filters and favourites, in Estonian and English.
  • The admin app (apps/admin): products, users, read-only views of chats and rooms, an audit log, and Stores, where the team creates stores and their invite links.
  • The partner app for stores (apps/partner): analytics, product corrections, data connections (Shopify, a product feed or a CSV upload) and team management.
  • The research CRM (apps/crm), loaded with a dataset of 99 Estonian furniture companies and 10 organisations.
  • This knowledge hub (apps/docs).
• Not real yet. All 50 products, prices, stock levels, store cities and delivery times in the shop are example data, and the app labels them as such. No real store has connected or supplied data yet. An outreach email to stores is drafted.
• Planned. Real products from stores that connect, use without an account, price-drop and new-match alerts, shared lists for couples, and matching that also uses product images. See What is not built yet.

#Map of the hub

#Technical sections

These go into more detail than the product pages:

• Platform architecture: how the apps, the database, sessions and the AI calls fit together, and how to run, deploy and test them.
• Partner integrations guide: how a store's product data gets into Leitagu, for store staff and for our team.
• Analytics definitions: what we count and how, from impressions to store visits.
• Internal CRM: the team's research and sales tool for potential partner companies.
• Reading this hub as an AI agent: machine-readable versions of every page.

#How to read this hub

• Every page starts with a short summary, so a page makes sense when read on its own.
• Statements about how Leitagu behaves come from the code in apps/ and packages/ and the READMEs next to it. Anything planned but not built is marked as planned.
• Every page is also available as Markdown at its address plus .md. /llms.txt indexes the hub, /llms-full.txt holds every page in one file, and /index.json lists the pages with their headings. See Reading this hub as an AI agent.
• Estonian terms appear in parentheses, for example sideboard (puhvetkapp). The glossary lists them all.
• The source documents behind these pages are in the repository: docs/leitagu-requirements.md (requirements and user stories), docs/leitagu-handover.md (decisions and build plan), docs/furniture-agent-estonia.md (research notes), docs/retailer-outreach-email.md (the pitch to stores) and data/research/README.md (the research dataset of potential partners).