Open protocols. Open standards. Ship your first AI agent in 5 minutes.
FAOS gives you the building blocks to create, deploy, and orchestrate AI agents across 25 industries. Python SDK, TypeScript SDK, REST API, and three open protocols that make your agents talk to anything.
from faos import FaosClient
client = FaosClient(api_key="sk-...")
agent = client.agents.create(
industry="banking",
role="compliance-analyst",
skills=["kyc-review", "aml-monitoring"]
)
result = await agent.run(
"Review KYC for account 4821"
)
print(result.output)
# -> Structured compliance reportA modular, protocol-native platform designed for builders who need enterprise-grade agent orchestration without the enterprise-grade complexity.
Web, Mobile, CLI, Slack, Teams
AG-UI ProtocolAgent Engine, Workflows, Knowledge, 25 Modules, 700+ Skills
A2A ProtocolYour Data, AI Models, Enterprise Tools
MCP ProtocolEvery layer is independently extensible. Swap AI models without changing agent logic. Add industry modules without modifying the core. Connect new tools through MCP without rewriting integrations.
FAOS is built on open standards so your agents can talk to users, talk to each other, and talk to every tool in your stack.
Agent to User Interface
Connects AI agents to any user interface — web, mobile, desktop, Slack, Teams, or custom. Build once, render anywhere.
import { FaosStream } from '@faos/sdk';
const stream = client.agents.stream("compliance-agent", {
task: "Review Q4 filings",
onUpdate: (event) => updateDashboard(event),
onApproval: (request) => showApprovalModal(request),
});Model Context Protocol
Gives agents secure, structured access to external tools, databases, APIs, and file systems. One protocol, unlimited connections.
agent = client.agents.create(
role="data-analyst",
mcp_servers=["salesforce", "postgres", "slack"],
)
result = await agent.run(
"Pull Q4 revenue from Salesforce, compare to forecast "
"in our database, and post summary to #finance"
)Agent to Agent
Enables agents to discover, communicate, and delegate tasks to other agents — across your organization or across organizations.
workflow = client.workflows.create(
agents=["kyc-agent", "risk-agent", "compliance-officer"],
pattern="sequential-with-review",
escalation="human-in-the-loop",
)
result = await workflow.run(
"Onboard new corporate client: Acme Holdings Ltd"
)Python SDK, TypeScript SDK, and a REST API that covers everything. The SDKs are in public preview — request early access.
pip install faos # public previewFull async support. Type-safe. Built for AI/ML engineers who want to create agents, run workflows, and manage industry modules.
npm install @faos/sdk # public previewFirst-class TypeScript support for frontend and full-stack developers building agent-powered applications.
curl -H "Authorization: Bearer sk_..." https://api.faosx.ai/v1/agentsLanguage-agnostic access to the full FAOS platform. Every SDK operation is also available through REST.
Every FAOS industry module comes with pre-built agents, domain-specific skills, workflows, and knowledge structures. Use them as-is, extend them, or build your own.
Cybersecurity
45+ agents
Manufacturing
40+ agents
Investment Management
40+ agents
Banking & Finance
35+ agents
Logistics & Supply Chain
35+ agents
HR & Talent
30+ agents
Professional Services
30+ agents
Healthcare
30+ agents
Aviation
30+ agents
Agriculture
30+ agents
Marketing & MarTech
30+ agents
Banking · Vietnam
25+ agents
Insurance
20+ agents
Capital Markets · Vietnam
20+ agents
Software & SaaS
20+ agents
Systems Integration
20+ agents
Travel & Hospitality
15+ agents
Retail & E-commerce
15+ agents
Insurance · Vietnam
15+ agents
Automotive
15+ agents
AgriTech
15+ agents
Gaming
15+ agents
PropTech · Vietnam
15+ agents
FinTech
15+ agents
EduTech
15+ agents
Every module follows the same architecture. Adding a new industry is a YAML configuration — no code changes to the core platform.
No infrastructure to set up. No models to train. Install the SDK, create an agent, run it.
pip install faos # public previewfrom faos import FaosClient
client = FaosClient(api_key="sk-...")
agent = client.agents.create(
industry="banking",
role="compliance-analyst",
)result = await agent.run(
"Review the KYC documentation for "
"new account application #4821"
)
print(result.output)
# -> Structured compliance review with
# findings, risk score, and recommendationsQuestions, contributions, feedback — the FAOS developer community is where builders connect.
Our published papers and open, reproducible code. See exactly how FAOS works under the hood.
Browse the RepoCompounding Intelligence — essays on building the agentic enterprise, with a new issue most weeks.
SubscribeHands-on guidance for your use case. Architecture review, integration planning, or POC build.
Book a WorkshopFAOS is built on open standards. The protocols are open, the SDKs are MIT-licensed and entering public preview, and our research is published in the open. No vendor lock-in, no proprietary traps.
The Python and TypeScript SDKs and protocol clients are MIT-licensed and entering public preview. Build without lock-in.
AG-UI, MCP, and A2A are open standards. Your agents are not locked into FAOS — they interoperate with any compliant platform.
Our research program publishes papers and reproducible code in the open at github.com/frank-luongt/faos-research. Verify the results yourself.