Three source types
Text, file (PDF, DOCX, TXT, MD up to 50 MB) or website – plus the knowledge Brand AI creates during setup.
Knowledge stores are the stable context source of your brand: facts, prices, products and team in one place. You do not train an AI model with them, you give it context – the right passages for every post instead of general world knowledge only.
With many AI tools, quality depends on the prompt or chat history at hand. Context gets lost between sessions, thins out with every message or has to be pasted in again and again – and results drift. In ALLMEDIA, the context is fixed to the channel. That is the decisive difference.
How it works
This is what knowledge stores look like in the app.
Folders such as Product / Service or Legal keep things tidy – as a list or as a canvas.
A knowledge store is like a briefing the AI reads along with every post. You set it up once, keep it current and decide where it applies.
Current prices, plans and terms – so no post quotes old figures.
Product data, features and use cases.
Who is responsible for what, with roles and contacts.
Answers to the questions people keep asking.
Mission, vision and values of your brand.
Details and wording that always have to be right.
Personal data of your customers, such as orders, invoices or emails, does not belong in it.
Text, file (PDF, DOCX, TXT, MD up to 50 MB) or website – plus the knowledge Brand AI creates during setup.
Manual or automatic – hourly to monthly, with last and next sync.
For example brand identity, product/service, audience, tone of voice, legal and FAQ/knowledge. The app suggests a folder structure, in list or canvas view.
Every channel uses exactly the knowledge that fits it.
On each post you see which context went into it.
Amy suggests new knowledge sources and updates – you confirm.
Knowledge stores do not train a model. Your knowledge serves as context for your posts: the matching passages are passed along with every post. For the AI model you choose per channel between models labelled “EU” and “No data retention” and globally available models.
In a chat, the result depends on the prompt and history at hand: context thins out with every message, gets lost between sessions and has to be pasted in again. In ALLMEDIA, the context is fixed to the channel – every post starts with the same maintained briefing, the same for every team member.
As current as your sources. Website stores sync on your interval or at the push of a button, files you replace, texts you edit directly. The app shows when the last and the next sync take place.
PDF, DOCX, TXT and MD up to 50 MB. Scanned PDFs are read by AI text recognition, which costs credits.
For each knowledge store you add one URL whose main content is fetched. For more pages you create more stores.
Maintaining knowledge stores is included in all plans, including the Free plan. Website syncs and text recognition for scanned PDFs use credits – the app shows how many.