ErdosFL

ErdosFAI · part of ErdosFL · Erdos Federated Agentic Intelligence

The AI platform your engineers
wish they had time to build.

ErdosFAI is an agent lifecycle platform — build, test, deploy, reinforce, and govern AI agents from one observable, audited runtime. Five tightly-integrated layers covering composable agent cards, multi-agent orchestration, human-in-the-loop safety, continual improvement, and real-time observability.

Explore the stack Quickstart
  • Dependency-light core
  • Runs offline, no key
  • Audit-ready
  • Apache-2.0

The agentic lifecycle

Many tools. One platform.

01

Build

Compose agent cards from prompts, skills, memory, tools, and MCP servers.

BLUE
02

Test

Score agents against rigorous evaluation suites before they ship.

RED
03

Deploy

Run multi-agent pipelines with human-in-the-loop gates.

BLUE
04

Reinforce

Learn from every trajectory — agents that improve release over release.

GREEN
05

Govern

Policy enforcement, cost, and a tamper-evident audit trail.

GREEN

Enterprise agentic technology

Five tightly-integrated layers

Each layer is a real Python package. The core is standard-library only, so it reads, runs, and ships anywhere.

01

Intelligence Layer

Composable agent cards — prompt, skills, memory, MCP servers, and a typed tool library wired into one reusable, versioned object.

erdos_fai.intelligence
02

Orchestration Layer

Multi-agent pipelines — chain agents into steps that pass artifacts forward, with approval gates between them. Low-code by construction.

erdos_fai.orchestration
03

Safety Layer

Multi-approver human-in-the-loop governance, admin-configurable write gates, and PHI/PII-aware redaction on every input and output.

erdos_fai.safety
04

Learning Layer

Trajectory-driven evaluation and optimization, dataset synthesis, and refined memories written back to agents — versioned upgrades.

erdos_fai.learning
05

Insights Layer

Real-time cost & ROI dashboards, agent metrics, event-replay tracing, and a hash-chained, tamper-evident audit trail.

erdos_fai.insights

Build agents.
Keep control.

Don't hand your agentic future to a single model provider — own the lifecycle.

Get started →

How it fits together

One governed runtime, end to end

03 · SAFETY — policy · redaction · HITL write gates 05 · INSIGHTS — tracing · cost/ROI · audit trail 02 · ORCHESTRATION — pipeline 01 Intake 01 Triage 01 Referral 04 · LEARNING — trajectory → evaluate → refine ↺

Get started

From idea to a governed pipeline in a few lines

Install

# clone, then from the repo root
pip install -e ./erdos-fai
# live Claude calls (optional)
pip install -e "./erdos-fai[anthropic]"

Build & run a pipeline

from erdos_fai import Agent, AgentCard, Pipeline, Step

triage = Agent(AgentCard(
    name="Triage",
    system="You triage support tickets.",
))
pipe = Pipeline("support", [
    Step("triage", triage, output_key="triage"),
])
print(pipe.run("Dashboard is down").final_output)

No API key? It runs offline on the built-in echo provider. Set ANTHROPIC_API_KEY to run on claude-opus-4-8.

The product suite

Plan · Build · Test · Govern — in one platform

Agent Builder

Compose deep agents — skills, tools, memory, MCP — with guardrails baked in.

Evaluation & QA

Continuous scoring, regression detection, and per-metric reports.

Continual Learning

Agents adapt in production, getting smarter with every run.

Governance & Audit

HITL approvals, write gates, cost control, and tamper-evident logs.