Agentic chat
Not autocomplete — an agent that senses context, plans a route, and acts with tools, streaming every step back to you.
Hosted SaaS and self-hosted enterprise deployment
An agent that can execute real work, not just answer questions.
isA understands context, plans next steps, and uses tools to complete tasks, with memory that carries across sessions. Start on the hosted workspace or deploy the enterprise edition in your own environment.
No credit card to start · Free tier included · Enterprise deployment keeps data inside your environment
Agent session
Plan my launch week and draft the announcement.
Sense— context · 14 tools · memory
Plan— route the task
Act— execute & stream
Done — 5 tasks scheduled across launch week and a draft announcement saved to your workspace.
8
apps on one shared memory
60+
built-in tools in every session
75+
models, routed automatically
3
ways to run — desktop, cloud, or self-hosted
Product surfaces, shown as they run.
The same agent core powers the hosted workspace, the local desktop experience, and the enterprise console. The animations below model the live isagent.io surfaces: user work happens in the app, while deployment, access, models, and audit controls live in the console.
Plan a 3-city launch trip next week and add it to my calendar.
web.search · flights
calendar.create_event ×6
Booked SF → NYC → Austin and added 6 events to your calendar.
Launch week
Mon · San Francisco — kickoff
Wed · New York — press
Fri · Austin — partners
Message the agent…
Hosted SaaS · for individuals and teams
Agent workspace
One agent, every app, one shared memory — with live tool use in each session.
- Agentic chat with live tool use
- 8 apps, one shared memory
- 60+ built-in tools
- Persistent memory across sessions
- Proactive automation & scheduling
- 75+ models with policy-aware routing
- Telegram channel live — more channels on the roadmap
Dashboard
Welcome back, admin@acme.io
API requests
771
Tokens used
16.8M
Active agents
12
Est. cost
$8.75
Governance · live
Deploy 'support-bot' approved
API key created · prod
Member added · dev@acme.io
Model routed · auto
Self-hosted · for organizations
Admin console
The full control plane — govern agents, run the model stack, and prove compliance.
- Self-hosted in your cloud or on-prem
- White-label branding & SSO
- Governance: RBAC, approvals, audit log
- Model platform: serving, fine-tuning, GPU
- Full observability: traces, metrics, health
- Bring your own keys (BYOK)
- Data sovereignty, SLA & dedicated support
One platform. Three ways to run it.
The same agent core runs on your own machine, hosted in the cloud, or self-hosted in your enterprise — pick the one that fits, switch when you need to.
On your machine
Desktop
macOS · Windows · Linux
Install and run the agent locally with one command. Standalone keeps everything on-device and private; edge adds cloud sync when you want it.
- One-line install, runs offline
- Standalone — fully local & private
- Edge — local + cloud sync
$ curl -fsSL https://isagent.io/install.sh | bashor pip install isa-mate
For individuals and teams
Hosted SaaS
isagent.io
Sign up and start quickly. The hosted workspace is managed, updated continuously, and metered by credits for individuals and teams.
- Fast access, no infrastructure setup
- Free tier, then usage-based plans
- Apps, tools, and skills in one workspace
For enterprises
Self-hosted & white-label
Your cloud or on-prem
Deploy the full platform inside your own environment — in production today on dedicated on-prem clusters. Your data never leaves your perimeter.
- Runs in your cloud or data center
- White-label branding & SSO
- Data sovereignty, audit & SLA
Built to act, not just answer.
Everything an agent needs to carry real work end to end — under one account, one memory, one workspace.
Persistent memory
It remembers your projects, preferences, and past work across every session — no re-briefing.
Multi-app workspace
Chat, research, trading, images, and content generation — distinct apps that share one memory and one account.
Tools & skills
Tools and curated skills are available in the workspace so the agent can execute tasks, not only draft responses.
Proactive automation
Schedule tasks, watch for events, and let agents act on a calendar — work that continues when you step away.
Model routing
Routes requests across available models based on task fit, policy, availability, and cost-performance requirements.
Common workflows it can run.
The same agent core supports different work patterns, from personal productivity to research, operations, data analysis, and automation.
For teams & businesses
Customer operations
Answer, triage, and resolve on the channels your customers already use — with your knowledge base behind every reply.
Chat · Knowledge · Channels
Research & reporting
Competitive scans, market briefs, and due diligence — researched on the live web and delivered as finished documents.
Hunt · Omni · Documents
Workflow automation
Multi-step business processes that run on a schedule and hand control to a human when judgment is needed.
Automation · Human-in-the-loop
For individuals
Data analysis
Point it at a CSV, a database, or a URL — get queries, charts, and models back, explained in plain language.
Data Scientist
Markets & portfolio
A quant workspace that screens, watches, and writes a daily report — you open the day already briefed.
Trading
Creative production
Images, brand assets, and long-form drafts — generated from templates and refined in conversation.
Dream · Omni
More than a chat box.
The same agent works three ways — talk to it, design with it, or hand it the browser. Switch modes mid-task; it keeps your context and memory.
Chat
Agentic conversation with live tool use, extended thinking, and runnable code artifacts.
Design
A creative canvas — generate images and designs, then preview and refine them inline.
Browse
Hand it the web — the agent drives a real browser in a live viewport, observing and acting for you.
For developers
Or build with the SDK.
An OpenAI-compatible client over the same agent platform. Install, point it at the gateway, and ship.
- OpenAI-compatible chat & streaming API
- One endpoint, 75+ models — routed for you
- Python SDK, REST, and webhooks
$ pip install isa_model
from isa_model.inference_client import AsyncISAModel async with AsyncISAModel(base_url="https://api.isagent.io") as client: resp = await client.chat.completions.create( model="claude-latest", messages=[{"role": "user", "content": "Plan my launch week."}], ) print(resp.choices[0].message.content)Want the full setup guide in your inbox?
Put an agent to work today.
Start free on the hosted product, or talk to us about running it inside your own cloud.