pulse-ollama

Ollama adapter for PULSE Protocol — run local AI models with the same interface as cloud providers.

Your data never leaves your machine. GDPR-compliant by design.

# Switch from cloud to local — one line changes:
# from pulse_openai import OpenAIAdapter as AI; adapter = AI(api_key="sk-...")
from pulse_ollama import OllamaAdapter as AI; adapter = AI(model="llama3.2")

# Everything below stays EXACTLY the same
msg = PulseMessage(action="ACT.ANALYZE.SENTIMENT", parameters={"text": "I love open source!"})
response = adapter.send(msg)
print(response.content["parameters"]["result"])

Why pulse-ollama?

Feature	Cloud (OpenAI/Anthropic)	Local (Ollama)
Data leaves your machine	Yes	No
GDPR compliance	Depends on DPA	By design
Works offline	No	Yes
API key required	Yes	No
Cost per token	Yes	Free

Installation

pip install pulse-ollama

Requirements: Ollama must be installed and running.

# Install Ollama (macOS/Linux)
curl -fsSL https://ollama.ai/install.sh | sh

# Pull a model
ollama pull llama3.2       # 2GB, fast
ollama pull mistral        # 4GB, high quality
ollama pull phi3           # 2GB, efficient
ollama pull gemma2         # 5GB, Google's model

Quick Start

from pulse import PulseMessage
from pulse_ollama import OllamaAdapter

adapter = OllamaAdapter(model="llama3.2")

# See what models you have installed
print(adapter.list_models())
# ['llama3.2:latest', 'mistral:latest', 'phi3:latest']

# Ask a question
msg = PulseMessage(
    action="ACT.QUERY.DATA",
    parameters={"query": "Explain PULSE Protocol in one sentence"}
)
response = adapter.send(msg)
print(response.content["parameters"]["result"])

# Sentiment analysis
msg = PulseMessage(
    action="ACT.ANALYZE.SENTIMENT",
    parameters={"text": "I love this open source project!"}
)
response = adapter.send(msg)
print(response.content["parameters"]["result"])
# {"sentiment": "positive", "confidence": 0.95, "explanation": "..."}

# Translate
msg = PulseMessage(
    action="ACT.TRANSFORM.TRANSLATE",
    parameters={"text": "Hello, world!", "target_language": "German"}
)
response = adapter.send(msg)
print(response.content["parameters"]["result"])
# Hallo, Welt!

Supported Actions

PULSE Action	Description
ACT.QUERY.DATA	Ask questions, get answers
ACT.CREATE.TEXT	Generate text from instructions
ACT.ANALYZE.SENTIMENT	Analyze emotional tone
ACT.ANALYZE.PATTERN	Find patterns in data
ACT.TRANSFORM.TRANSLATE	Translate between languages
ACT.TRANSFORM.SUMMARIZE	Summarize long text

Configuration

adapter = OllamaAdapter(
    model="llama3.2",                        # Default model
    host="http://localhost:11434",           # Ollama server URL
    timeout=120,                             # Inference timeout (seconds)
)

# Override model per-request
msg = PulseMessage(
    action="ACT.CREATE.TEXT",
    parameters={
        "instructions": "Write a poem about AI",
        "model": "mistral",           # Override default
        "temperature": 0.9,           # Creativity
    }
)

# Remote Ollama server (shared team deployment)
adapter = OllamaAdapter(host="http://ai-server.company.com:11434")

Provider Switching

The whole point of PULSE adapters is that switching providers is one line:

# Local model (privacy, offline, free)
from pulse_ollama import OllamaAdapter as AI
adapter = AI(model="llama3.2")

# Cloud (scale, latest models)
# from pulse_openai import OpenAIAdapter as AI
# adapter = AI(api_key="sk-...")

# Everything else stays identical
response = adapter.send(msg)

Part of the PULSE Ecosystem

pulse-protocol    # Core (install this first)
pulse-openai      # OpenAI GPT
pulse-anthropic   # Anthropic Claude
pulse-ollama      # Local models (this package)
pulse-gemini      # Google Gemini
pulse-binance     # Binance exchange
pulse-bybit       # Bybit exchange
pulse-kraken      # Kraken exchange

License

Apache 2.0 — free forever.

GitHub: github.com/pulseprotocolorg-cyber/pulse-python