Solvyren AI documentation

From your first conversation to reusable intelligent workflows, this guide covers the features available in the Solvyren workspace, how they behave, and the controls that keep you in charge.

Quick start Open workspace

The signed-in Solvyren AI workspace

Chats & projectsOrganize ongoing work and temporary tasks Create an AgentCombine instructions, files, tools, and Skills Connect intelligenceUse Solvyren or your own compatible API credentials Personalize the experienceAdjust language, layout, messages, and speech

Quick start

Create and sign in to your accountRegistration opens a private workspace associated with your account.
Select an intelligent serviceUse Solvyren AI or connect a compatible external provider from the model menu.
Start a conversationWrite a message, attach a file, or enable a Skill or available tool.
Organize and reuseMove work into projects and save presets, prompts, memory, or an Agent.

Chats & projects

A new chat is the entry point for every task. Find and filter past conversations from the sidebar, group related work in projects, open multiple conversation paths, or use Temporary Chat when a task should stay outside normal history.

ProjectsKeep related conversations and working material together

Multi-conversationExplore more than one response path from the same task

Temporary ChatUse a separate mode for work that should not enter regular history

Drafts & messagesLocal drafts, Markdown, long-message collapse, and smooth streaming

Create a project for a group of ongoing tasks

Models, external APIs & presets

The model selector brings Solvyren AI and user-connected providers into one place. Add external credentials only through the provider key interface—never paste a key into a chat message. Presets save a model and parameter configuration for repeatable use.

Select Solvyren AI or connect a compatible provider

Files, attachments & tools

Upload a local file from the composer or enable an available tool or Skill for the current task. The client can resize oversized images before upload and turn very long pasted text into a file for more reliable processing.

PrivacyUpload only content you are authorized to process, and remove sensitive information that is not necessary for the task.

Add a file, Skill, or available tool to the current conversation

Agents

An Agent is a reusable assistant designed for a specific goal. The builder combines a name, description, model, category, instructions, tools, Skills, file context, conversation starters, and optional support details.

Define the role, goal, inputs, and expected deliverable.
Connect only the tools and files the task genuinely requires.
Use conversation starters to make the Agent's scope clear.
Validate with small tasks before expanding automation.

Agent Builder: instructions, models, tools, Skills, and file context

Skills & prompts

A Skill packages a repeatable method with instructions, references, and tool constraints. A prompt stores a reusable task instruction or output format. Both can be created, searched, managed, and called from a conversation.

SkillsBest for multi-step work with rules, references, and controlled tools

PromptsBest for quickly reusing a focused instruction or output structure

Find, create, and reuse consistent ways of working

Memory & bookmarks

Memory stores preferences or background that should be available across conversations. You can create, filter, edit, disable, or delete saved memories. Bookmarks keep important content easy to revisit. Never store passwords, private keys, recovery phrases, or access tokens in memory.

Model parameters & context

The parameters panel supports custom identity and instructions, sampling controls such as temperature and Top P, penalties, context and output limits, stop sequences, reasoning controls, verbosity, file resend behavior, image detail, web-search capability, and advanced streaming options. Availability varies by provider and model.

Adjust generation, reasoning, and context behavior for the current task

Speech

Speech settings cover speech-to-text and text-to-speech. Choose an engine, language, and voice; adjust recording sensitivity and playback rate; and configure automatic transcription, sending, or playback. Your browser may ask for microphone permission when voice input is used.

Voice input, transcription, playback, and conversation mode

Personal settings

Each account can independently control theme, language, type size, UI scale, message sending, keyboard commands, Markdown and LaTeX behavior, notifications, layout, and accessibility.

Area	Available controls
Appearance & layout	Theme, language, font size, UI scale, chat direction, and mobile layout
Sending & editing	Enter-to-send, in-progress follow-ups, local drafts, image resize, and long text
Message display	Markdown, LaTeX, reasoning details, tool details, and rich-text copy
Notifications & access	Unread count, reply notifications, sounds, and keep-screen-awake

Personalize sending, commands, messages, conversations, and prompts

Data & privacy

The Data & Privacy panel controls whether saved memory participates in responses. It also provides conversation import, file and shared-link management, archived chats, provider API keys, credential revocation, local speech-cache cleanup, and conversation clearing.

HTTPS protects data in transit.
Never place passwords, private keys, recovery phrases, or API tokens in messages, memories, project descriptions, or prompts.
External providers apply their own billing, retention, and data-processing policies.
Sign out after using a shared device.

Start using Solvyren AI

Create an account and enter your intelligent workspace.

Get started Sign in