Agno Demo

5 agents, 1 team, 1 workflow served via AgentOS. Each agent learns from interactions and improves with every use.

Overview

Agents

Agent	Description
Dash	An adaptive data agent that queries and interprets your data — improving its understanding of your schema, metrics, and priorities with every interaction.
Pal	A personal agent that learns your preferences, context, and history.
Gcode	A lightweight coding agent that writes, reviews, and iterates on code. No bloat, no IDE lock-in — just a fast agent that gets sharper the more you use it.
Scout	A self-managing context agent that researches, drafts, and refines information stored in s3 buckets.
Seek	A self-learning research agent that investigates complex topics over time, building persistent knowledge that compounds across sessions.

Team

Team	Description
Research Team	Seek and Scout working together as a team.

Workflow

Workflow	Description
Daily Brief	A workflow that sources and surfaces new developments (using seek), tracks metrics (using dash), and produces a daily digest (using scout).

Architecture

All agents share a common foundation:

Model: OpenAIResponses(id="gpt-5.2")
Storage: PostgreSQL + PgVector for knowledge, learnings, and chat history
Knowledge: Dual knowledge system — static curated knowledge + dynamic learnings discovered at runtime
Search: Hybrid search (semantic + keyword) with OpenAI embeddings (text-embedding-3-small)
Learning: LearningMachine in AGENTIC mode — agents decide when to save learnings

Getting Started

1. Clone the repo

bash

git clone https://github.com/agno-agi/agno.git
cd agno

2. Create and activate the demo virtual environment

bash

./scripts/demo_setup.sh
source .venvs/demo/bin/activate

3. Run PgVector

bash

./cookbook/scripts/run_pgvector.sh

4. Export environment variables

bash

export OPENAI_API_KEY="..."      # Required for all agents
export EXA_API_KEY="..."         # Optional (Exa MCP is currently free)

5. Load data and knowledge

bash

cd cookbook/01_demo

python -m agents.dash.scripts.load_data
python -m agents.dash.scripts.load_knowledge
python -m agents.scout.scripts.load_knowledge

6. Run the demo

bash

python -m run

7. Connect via AgentOS

Open os.agno.com in your browser
Click "Add AgentOS"
Add http://localhost:7777 as an endpoint
Click "Connect"

Evals

Test cases covering all agents, team, and workflow. Uses string-matching validation with all or any match modes.

bash

# Run all evals
python -m evals.run_evals

# Filter by agent
python -m evals.run_evals --agent dash
python -m evals.run_evals --agent seek

# Verbose mode (show full responses on failure)
python -m evals.run_evals --verbose

Agno Features Demonstrated

Feature	Where
LearningMachine (AGENTIC mode)	All 5 agents
CodingTools	Gcode
ReasoningTools	Gcode
SQL Tools	Dash, Pal
MCP Tools	Seek (Exa), Scout (Exa), Dash (Exa), Pal (Exa)
Knowledge (hybrid search)	All agents
Persistent Memory	Pal, Seek
Teams (coordinate mode)	Research Team
Workflows (parallel steps)	Daily Brief
Scheduled Tasks	Daily Brief
AgentOS	run.py