Use the aixplain OS

The aixplain OS is what you use when you sign in to aixplain, at app.aixplain.com. It is where you chat with Omni, browse the Marketplace of models, tools and integrations, and build, test, deploy and monitor your own agents, with your workspace's credits, API keys and members in one place. We run it for you on cloud, so there is nothing to install. This article explains how to sign up, how to find your way around Chat, Studio, the Dashboard, Settings and your workspace, how agents run and are governed, and what the OS can't do yet.

On your own servers, the same OS runs at your organization's address: see Run the aixplain OS on-prem. For what the OS is made of, see What is the aixplain OS?.

Sign up and sign in

1. Go to app.aixplain.com.
2. Sign up with your email address or a social sign-in such as Google.
3. Verify your email address if asked.
4. Follow the welcome. It includes a card for connecting WhatsApp or Telegram if you want to message Omni from your phone.

To sign in later, go to app.aixplain.com and use the method you signed up with. Social sign-in accepts only verified email addresses and matches your exact email address.

You can browse the app and the Marketplace as a guest before you sign up. Private pages and actions open a sign-in or email-verification dialog.

Find your way around

The rail on the left switches between the parts of the OS. Next to it, the side nav holds your work: the workspace selector at the top, then your chats. The side nav is collapsible, chats still running show a spinner, and the user menu has a download menu for aixplain Desktop.

Koder sits in the rail next to Omni and opens Koder, aixplain's coding agent. For Koder on your computer, see Get started with Koder.

How it fits with aixplain Desktop and the Agents SDK

One aixplain account signs you in to all three, and they share your workspace's agents, credits and API keys.

An agent you save with the SDK is saved to your workspace, so you can open it, edit it and check its runs in the app. To install the desktop app, see Get started with aixplain Desktop.

Chat

Chat is where Omni lives in the browser. Type a request in the chat box, which has a + menu, and connect apps from the Ready for you row. Omni works in your connected apps, makes documents, slides and spreadsheets, runs tasks on a schedule or a trigger, and builds agents with you.

Each chat message carries the trace of the run behind it, so you can open the steps that produced an answer. HTML in a reply opens in a Code or Preview view, and the preview runs sandboxed.

The default agent behind chat is called the workspace agent. You can set it up with your own tools, integrations and models from the agent library.

For the full guide, see Get started with Omni.

Studio

Studio is the place in the app where you build. The previous studio app was replaced by the new app at app.aixplain.com, and the previous studio now shows a banner inviting you to switch.

You can also build an agent by describing it to Omni, with Koder in aixplain Desktop or in code with the Agents SDK. For every way to build, test, deploy and schedule an agent, see Build and deploy agents.

Build an agent

The agent builder puts a chat next to Canvas, Code and Tracer tabs, so you can build an agent and test it in the same place, with a live trace of each run.

1. Open the agent builder.
2. Add integrations and pick their actions, then add tools, skills and inspectors. You can give a tool its own display name on this agent, such as a sales Slack and a support Slack.
3. Add an existing agent as a sub-agent if the job needs a team. An attached sub-agent keeps its own settings.
4. Set the agent's maximum iterations and, if you need a fixed shape, an output template in JSON.
5. Test it in the chat beside the builder and read the trace.

The builder saves your work in your browser and warns you before you leave with unsaved changes. Drafts you haven't saved sit in the Unsaved tab. An agent's drawer has a Channels tab, where its channel connections are listed and you can revoke each one on its own.

Find your tools and integrations

The capability library lists your tools and integrations together. Integrations show a description, a Custom toggle gathers the integrations you build yourself, and MCP servers have their own tab. Pinned integrations appear first in lists.

Add knowledge

Knowledge, built on aiR, stores your documents in a hosted index that agents search by meaning to ground their answers. You create a knowledge collection through the aiR integration and add documents or files by link. Files are parsed into text and split into chunks.

• Search defaults: a collection can store its own search and ingest defaults, and reranking can be turned on per collection. It is off by default.
• Browse and edit: the Data page lets you browse, search and edit an index.
• Limits: each ingest request takes up to 100 documents, and documents with no readable text are reported instead of being added empty.

For code examples, see the knowledge base docs.

Add inspectors

Inspectors check an agent's input, steps or output while it runs, then let the run continue, edit the content, rerun a step or stop the run. An inspector can be a plain-language prompt run on a model, your own Python function or a guardrail from the Marketplace. Step inspectors also take effect on team agents that follow tasks or use a planner, so an output an inspector blocks never becomes the answer.

Add inspectors to an agent in the agent builder, or with inspectors=[...] in the Agents SDK.

Agent runtime

Every agent you save runs on the Agent Engine, the part of the OS that runs agents. On cloud, your agents run 24/7 on aixplain's managed infrastructure, and each run is billed.

On each run, an agent plans its own steps and calls models and tools. It stops when the task is done or a budget is reached, and every run returns its output with its steps, credits used and run time. The trace lists every step: the models and tools called, the credits used and the run time. In a team agent, each delegation carries the sub-agent's own steps, so you can see what it did.

Swapping the model on a live agent needs no rebuild, and the agent keeps its owner, policies, inspectors and run history.

Deployment

A saved agent is deployed: it runs on cloud with its own ID and an API that your teams and customers reach with a secure key.

• Call it: by its ID from the Agents SDK, the REST API or MCP, with an API key from your workspace. You can also call an agent through the OpenAI-compatible chat completions API by passing agent- followed by the agent ID as the model.
• Keys: create keys in Settings → API keys (see Settings below). A key can carry request, token and credit limits.
• Limits and monitoring: every deployed agent comes with rate limits, a budget, monitoring and tracing.
• Versions: agents are versioned, with separate staging and production, and you can roll back to an earlier version.
• Your own servers: you can start on cloud and move to your own servers later, with no rebuild. Deploying to your own servers needs the Enterprise plan.

To run an agent on a schedule or when an event arrives, add a trigger (see Triggers below). For code examples, see Build and deploy agents.

Control panel

Every agent you build on the aixplain OS is a service you own, and the OS governs it from one place. The tools you connect keep their own access rules, and you add more.

• An owner: every agent has a named owner from the start, so each result traces to a person.
• Inspectors: inspectors check every input, step and output, then edit, rerun, stop or escalate a run that breaks a rule.
• Approvals: agent changes, and Omni and Koder actions, wait for your approval. Anything that sends or spends waits for your Allow.
• Budgets and limits: set a budget on an agent to cap what a run can cost, and request, token and credit limits on each API key.
• An audit trail: every run records who ran it, each tool call and every decision or approval, with the time and cost of the run. Runs export.
• The Dashboard: every run, its trace and its cost, in one view (see below).

Admins set the policies, limits and model access for their workspace. On Scale, in early access, Owners and Admins also set access and budget policies for every member and group. For roles and budgets, see Manage your workspace.

Dashboard

The Dashboard (Analytics in the app bar) shows your agents' runs, usage, cost and traces. Each agent row shows recent runs and the last run date. You can filter runs by session, open a tab scoped to one agent, and reorder tabs, which keep their agents, date range and pins.

Run traces show the date with each run time and can open the raw JSON.

See usage

The workspace usage page shows what your workspace spends and breaks it down by API key, with each key's owner and masked key in the picker. Model and tool calls made during an agent's run are attributed to that agent, so usage by agent adds up. Every run's cost also shows in its trace.

Settings

Open Settings from your profile circle at the bottom of the rail. Settings holds your workspace's members, roles, billing, usage and API keys, and your own profile.

Create API keys

API keys belong to one workspace and use its assets and credits. Each workspace holds up to 10 keys.

1. Open Settings → API keys in the workspace the key should use.
2. Select Create API key.
3. Enter a name that says where the key runs, and optionally an expiration date.
4. Select Create, then copy the key.

A key can carry request, token and credit limits. A credit limit can be a lifetime cap or reset daily, weekly or monthly.

Billing

Owners and Admins top up credits from Settings → Billing. For plans, credits and top-ups, see Your account, plans and billing.

Your workspace

Everyone has a personal workspace, and you can create or join team workspaces, where members share the team's assets and credits. Switch workspaces from the workspace selector at the top of the side nav. Members can open workspace settings, with admin-only actions hidden. Pending workspace invitations show up across the app.

For roles, sharing, budgets and Enterprise workspaces, see Manage your workspace.

The Marketplace

The Marketplace lists 900+ models, tools, integrations, skills and agents you can use, each with its price in credits. Agents and the Agents SDK reach all of them with one API key.

• Models: each model's page shows its input, output and cached-input prices in credits, its actions and functions, and its data retention. The Models page has a Data retention (ZDR) filter for models that keep nothing.
• Tools: web search, databases, Python script tools and more. A tool's page has its own tab for trying it, and its actions read as documentation.
• Integrations: connect apps such as email, calendar and chat tools, and MCP servers, including servers that protect their endpoint with OAuth. When you connect an integration, you choose which actions to connect.
• Skills: skills published by other teams appear on the Marketplace and in the agent builder's picker, and you can download one as a zip from its page.

Search covers models, tools, integrations, skills and files, and you can filter by category and developer. Your team's own assets show up in results too. See Find models, tools and integrations in the Marketplace.

Triggers

A trigger runs an agent on a schedule or when an event arrives, such as a new email in a connected Gmail account. Triggers in the rail lists them, and an agent's drawer has a Triggers tab for that agent's own.

Your first steps

1. Ask Omni for one real task. Name the apps or files it should use, and connect an app when it asks.
2. Look around the Marketplace. Open a model you might use and check its price and data retention.
3. Build a small agent in Studio. Give it one tool, test it beside the builder and read the trace.
4. Create an API key and call the agent from your own code. The Agents SDK quickstart runs a first agent in a few minutes.
5. Check the Dashboard and the usage page to see what the run cost.

Pricing and usage

Usage on cloud is paid with prepaid credits, and one credit is one US dollar. Each workspace has its own balance. Models and tools are priced by their provider, and an agent run costs the sum of its model and tool calls plus 20% for orchestration, inspectors, memory and tracing. Purchased credits expire 12 months after purchase.

When a workspace has no credits left, the app blocks new runs and shows a warning. Owners and Admins top up from Settings → Billing. For plans and what uses credits, see Your account, plans and billing and Pricing.

Privacy and data

Your data is always yours: we never sell it or train on it. To keep a model from retaining what you send, pick one marked zero data retention in the Marketplace. Signing out clears your account data from the browser. For more, see Privacy and data and Security.

Limitations

• Omni on the web and Omni in aixplain Desktop don't share sessions or settings yet.
• Deploying agents to your own servers, in your VPC or air-gapped, needs the Enterprise plan.
• Single sign-on (SSO) isn't available yet.
• Each workspace holds up to 10 API keys, and credits can't move between workspaces.
• Each knowledge ingest request takes up to 100 documents.

Frequently asked questions

Get help

• Learn the SDK and API: read the Agents SDK docs.
• See what changed: read the release notes for the latest platform updates.
• Talk to us: for Scale, Enterprise or anything else, contact us.
• Report a security issue: email care@aixplain.com. Remove credentials and private data from logs and screenshots first.