Service Description

This page describes the feature set of the Uprelic Services. It forms part of your contract (Individual Terms §§ 1.6, 2.1; Business Terms § 3): it defines what the Services are, in the version in force when you conclude your contract. How the Services may change over time is governed by the Terms (Individual Terms § 22; Business Terms § 21), and earlier versions of this page remain retrievable via the version selection.

Which plan includes which limits (Credit allowances, parallel generations, scheduled Automations) is stated on the pricing page. Which AI model providers serve requests is stated on the Subprocessors & AI Providers page.

Chat and the assistant

A chat interface in which an AI assistant answers questions, writes and edits, and carries out multi-step tasks. You choose the model from the list available on your plan; requests are processed by the model provider serving your selection. The assistant can plan tasks, call tools, search and read the web, run code, and — with your permission — operate your connected third-party accounts. Guests can view shared chats and continue a shared chat in a guest session.

The assistant is personalised: the Services derive and store notes about your projects and preferences from your usage to improve later answers. You can manage this in the settings.

Knowledge, files and search

You can upload files and organise them in projects. Uploaded and connected content is parsed and indexed so the assistant can search and cite it. Web search and page reading are available to the assistant, and fetched pages may be cached. Per-file visibility controls apply in shared projects.

Connected accounts and Actions

You can connect third-party accounts (for example messaging, email, cloud storage, project management, design and meeting tools, and further tools via an integration catalogue). Connected content can be read and indexed at your instruction. With permissions you grant per service (read / write / delete), the assistant and your Automations can perform Actions in a connected account on your behalf — sending a message, creating or updating an item, editing a file. Actions are governed by Individual Terms § 9 (Business Terms § 11); some Actions cannot be undone. Disconnecting an integration stops access; previously ingested data can be deleted.

Automations

You can configure workflows that run unattended — on a schedule, on a webhook, or on an event in a connected account. Automation steps can use AI models, connected accounts (including Actions), web fetching and file processing. The number of active scheduled Automations is a plan limit.

Code execution

The assistant can execute code in isolated sandbox environments on our infrastructure, at your instruction or as a step of a task you gave it. Sandboxes are ephemeral; files you stage into a sandbox and outputs return into your workspace.

Documents and artifacts

The Services produce and store structured outputs ("Artifacts"): text documents, documents with tracked changes, PDF reports, spreadsheets, diagrams and images. Artifacts are versioned, can be edited collaboratively, and can be shared. Uploaded images can be analysed; images can be generated and edited.

Collaboration and public content

Projects support roles and visibility levels (private, link, domain-based, public). Chats, files and Artifacts can be shared via links, including with guests. Projects can contain forums with posts and comments, in which the assistant can also reply; public forum posts and public projects are reachable by anyone and indexed by search engines. Content standards for public content are in the Acceptable Use Policy.

Voice

Dictation converts your speech to text on our infrastructure; you confirm the transcript before it is used. Audio processed as part of Automations may be transcribed by an AI provider listed on the Subprocessors & AI Providers page.

Apps

The Services are available as a web application and a Slack integration. Where we offer iOS or Android apps, paid plans remain purchases made on the web; the mobile apps do not offer in-app purchases.

Accounts and plans

Registration uses email or a supported sign-in provider. Accounts have an immutable public handle and a display name. Plans, allowances and limits are described on the pricing page; Credits are described in the Credits section below. Account deletion removes your data as described in the Terms and the privacy policy.

Data export

You can export your data in your account settings: you request an export and receive a download link by email. The export is one or more ZIP archives containing your data as JSON files — your profile, one file per chat, one file per project, your forum posts and your derived memories. Uploaded files, Artifacts and your billing records are not part of this bundle: they are accessible and downloadable directly in the product. Automation configurations are provided on request. All of them are included when we perform an export for you on request ([email protected]) under the Terms' switching and portability clause.

Credits

This section is the Credit Guide your Terms refer to (Individual Terms § 6.2; Business Terms § 14): what consumes Credits, what does not, and how consumption is measured. It deliberately contains no per-model rate card — what you actually pay is measured per request and shown in your usage view.

What a Credit is

A Credit is the unit of account in which we measure your use of the AI features of the Services. It is not money, not electronic money and not a voucher. What Credits cost in currency, and which allowance each Plan includes, is stated on the pricing page and in checkout.

What consumes Credits

Credits are consumed by operations that do AI work for you:

• chat messages and everything the assistant does to answer them,
• agent tasks, including Actions in Connected Accounts,
• Automation runs,
• transcription of audio,
• analysis, generation and editing of images,
• document generation (for example reports and spreadsheets).

What consumes no Credits

• reading, browsing and searching what is already in your workspace,
• storing and organising files,
• sharing, publishing and forum activity without AI assistance,
• account, billing and settings operations,
• on the free tier: use of the included model (the free tier has no Credit balance at all — see Individual Terms § 4).

What determines consumption

How many Credits an operation consumes depends on the work it causes:

• how much text is processed — both what goes in (your prompt, the files and context the assistant reads) and what comes out,
• which model you select — more capable models consume more than efficiency models,
• how many steps an agent task takes — a task that searches, reads, executes code and writes a document consumes more than a single answer, and a longer-running task consumes more than a short one,
• the size and length of media — minutes of audio, number and size of images and files.

Because tasks differ, consumption for the same kind of request varies. We therefore do not promise fixed per-operation prices; we promise an exact meter instead.

The meter

Every Credit-consuming request is metered individually and recorded. Your usage view in the settings shows your consumption within the current billing period. These records are the basis on which your balance is calculated, and they are what you can refer to if you believe a deduction is wrong (Individual Terms § 6.7).

Your spend is bounded

You can never spend more Credits than you have. When your balance reaches zero, AI processing continues on the model included in the free tier until your allowance renews or you buy a top-up: the models selectable on your Plan become unavailable, we say so in the product, and the free-tier processing itself costs nothing and consumes no Credits (Individual Terms § 2.4). Plan limits (such as parallel generations) bound concurrent use.

Changes

A change to the consumption behaviour that increases what you pay for comparable use is a price increase and follows the consent rules in Individual Terms § 6.6 (Business Terms § 14.9). We announce changes to this guide as described in the Terms; earlier versions remain retrievable via the version selection on this page.