Skip to content

Demos

Working demo applications that showcase LOCO-Agent with real frameworks.

Google ADK -- Customer Support

3 ADK agents (triage, support, escalation) sharing a bounded Gemini API pool. Live Gemini API calls with LOCO scheduling.

Repo: github.com/ArielSmoliar/loco-adk-demo

git clone https://github.com/ArielSmoliar/loco-adk-demo.git
cd loco-adk-demo && pip install -e .
python run_mock.py              # no API key needed
python run_mock.py --capacity 1 # heavy contention
Agent Model Weight Role
triage gemini-2.5-flash 1.5 Classify tickets
support gemini-2.5-flash 1.5 Handle simple tickets
escalation gemini-2.5-pro 3.0 Handle complex issues

AutoGen -- Enterprise Security Pipeline

8 AutoGen agents with per-agent budget enforcement. Shows cost governance in action -- investigators hit their budget cap, the scheduler rejects further work and frees slots for other agents.

Repo: github.com/ArielSmoliar/loco-autogen-demo

git clone https://github.com/ArielSmoliar/loco-autogen-demo.git
cd loco-autogen-demo && pip install -e .
python run_mock.py                      # reject mode
python run_mock.py --budget-mode alert  # allow over-budget with warnings
python run_mock.py --capacity 1         # heavy PTU contention
Role Count Model Weight Budget
Coordinator 1 gpt-4o 3.0 uncapped
Analyst 4 gpt-4o-mini 1.0 20 each
Investigator 2 gpt-4o 3.0 12 each
Responder 1 gpt-4o 3.0 15

AWS Bedrock -- Document Processing Pipeline

8 agents across 4 roles processing financial documents through a multi-model Bedrock pipeline. Demonstrates v0.3 PolicyEnforcer with budget + access control + rate limiting. Reviewer (Opus) hits its budget cap; confidential documents (SARs, audits) are restricted to the reviewer agent via AccessPolicy.

Repo: github.com/ArielSmoliar/loco-bedrock-demo

git clone https://github.com/ArielSmoliar/loco-bedrock-demo.git
cd loco-bedrock-demo && pip install -e .
python run_mock.py                  # 3 policies active
python run_mock.py --capacity 1     # heavy contention
python run_mock.py --rate-limit 5   # tight rate limits
Role Count Model Weight Budget Label Access
Intake 3 Claude Haiku 1.0 30 each public, internal
Analyst 2 Claude Sonnet 2.0 40 each public, internal
Reviewer 1 Claude Opus 5.0 25 all (incl. confidential)
Summarizer 2 Claude Haiku 1.0 uncapped public, internal

Built-In Examples

cd loco-agent
python examples/burst.py           # 8 agents, simultaneous work arrival
python examples/fairness.py        # 10 agents, sustained load, Jain's fairness
python examples/webhook_spike.py   # Background load + urgent webhook spike
python examples/mdash_security.py  # Multi-model cost routing (55 agents)