Questions about shardr,
answered plainly.

shardr is a white-label AI agent platform. You add a branded agent chatbox to the website or app you already run, and every end user gets their own agent with its own memory. The agent answers from your company’s knowledge and performs actions your product already supports.

Companies that already have a product and a user base, and want a conversational layer inside it rather than a separate assistant. Your users stay in their existing account, in your existing interface.

No. shardr adds a chatbox to the product you already run. Your app keeps its screens, its data and its actions; the agent proposes, and your app performs.

Yes. The chatbox carries your brand inside your product. Your users are talking to your agent, not to a vendor’s assistant.

Each end user has their own agent with its own memory, rather than one shared bot. Preferences, history and context belong to that person and carry into their next conversation.

What the user has shared and your configuration allows: preferences, goals, constraints and past decisions. A new conversation starts with that context already in place, so users do not introduce themselves twice.

Yes. Your team can open a user’s memory in the console and correct it. A correction keeps the history: the old fact is retired and linked to its replacement, and staff corrections are marked as such.

From the content you connect: your catalogue, your policies, your support articles, your uploaded documents. Each agent version names the sources it is allowed to use, so two agents in the same company can have different reading lists.

The ones your product already performs and you choose to expose — adding to a basket, saving a plan, booking a session, preparing a rule. The user approves the suggestion and your app does the work, with the confirmation returning to the same conversation.

Yours. You connect your own provider credential and choose a model that key can serve. The calls stay on your provider account, and the key remains yours to rotate or revoke.

Your provider bills your company directly, at its own prices. shardr adds no markup and your provider’s bill never passes through shardr.

The agent platform: the agent, memory, knowledge, actions and your team’s console. The charge scales with active end users rather than the number of messages they send.

One person in your app in a given month. The same user talking twice or forty times is still one active end user; their extra conversations change your model usage, not the number of users shardr counts.

Yes, at two levels: a model-spend cap for each end user, and an overall monthly cap for the environment. You choose when alerts fire and what happens at the limit — refuse new requests, or switch to a cheaper model.

Production and staging. Published agent versions never change, so you can tell exactly which version answered a given conversation.

No. Corrections retire a fact and link it to its replacement rather than erasing the record, so support can see what the agent believed at the time and what it knows now.

With the first release. shardr is working it out with the first companies using the product, including what active-end-user billing means for them.

shardr is in early access. Email hello@shardr.ai to join, and you can help shape what ships — pricing included.