New Opensource AI 70B — Now Available Free

The AI Platform That Thinks, Builds & Learns

Visual workflow automation with a built-in 70B AI model, AI reasoning agents, GPU-accelerated data science, and AutoML — all in one platform. Community tier free, Pro from £1.87/day.

🚀 Start Building Free
✓ No credit card needed ✓ Community tier always free plan ✓ Sign up in 30 seconds
Already have an account? Sign in →
▶ Watch Demo
No credit card required
Unlimited executions forever
82 TFLOPS Dual GPU included
Your data stays private
Open source
Intelligence Pipeline — Opensource AI Agent
Running
Input Trigger
Webhook / Schedule
🧠
AI Agent
ReAct Reasoning
AI Engine 70B
Opensource AI
GPU Analytics
cuDF · 82 TFLOPS
Output
Any channel
Powering next-generation AI workflows
🧠Opensource AI 70B
83.4 TFLOPS FP16
🚀260GB GPU VRAM
Unlimited Executions
🔬PySpark Workbench
🤖AutoML Pipeline
📓JupyterLab GPU
🔗19+ Integrations
🛡️100% Private AI
💷From £1.87/day
🧠Opensource AI 70B
83.4 TFLOPS FP16
🚀260GB GPU VRAM
Unlimited Executions
🔬PySpark Workbench
🤖AutoML Pipeline
📓JupyterLab GPU
🔗19+ Integrations
🛡️100% Private AI
💷From £1.87/day
Platform

Everything you need,
nothing you don't

From visual workflow automation to big data processing and machine learning — deeply integrated, endlessly powerful.

Visual Workflow Builder
Professional drag-and-drop canvas. 30+ node types, real-time execution, undo/redo, minimap, keyboard shortcuts. Build complex pipelines in minutes.
✓ Core Feature
🧠
AI Agent — ReAct Reasoning
Multi-step reasoning agent that thinks before acting. Real-time sensor inputs: time, weather, GPS, microphone, camera, live news feeds.
🚀 Exclusive Feature
🤖
Opensource AI Engine 70B
70 billion parameter AI model running locally on your GPU. Zero API costs, total data privacy, unlimited tokens. Switch models instantly.
🔒 100% Private
🔬
Data Science Workbench
Enterprise big data in your browser. PySpark, Hive SQL, Pig Latin, and GPU-accelerated analytics — all in one place.
Pro Feature
# PySpark GPU
import cudf
df = cudf.read_csv("data.csv")
result = df.groupby("dept")
  .agg({"sal":"mean"})
✓ 10M rows in 0.03s
🤖
AutoML Pipeline
Upload CSV → auto-clean → race ML algorithms → download best model → predict. No code required at any step.
Pro Feature
Leaderboard
🥇XGBoost94.2%
🥈Rnd Forest91.8%
🥉Log. Reg.87.4%
✓ Best model auto-saved
📓
JupyterLab Integrated
Full JupyterLab with GPU-accelerated Python kernels. PySpark, cuDF, and all data science libraries pre-installed. Access from your browser.
Pro Feature
Sub-Workflows & Error Handling
Call saved workflows as sub-workflows. Error handler nodes catch failures automatically and route to fallback logic or notifications.
✓ Built-in
🔑
Credentials & Security
Secure credential manager for API keys, OAuth tokens, webhooks, SSH keys. Export/import workflows, node pinning, workflow variables.
✓ Enterprise Ready
Built For Everyone

Powerful for experts.
Simple enough for anyone.

We built four features specifically to make enterprise-grade AI accessible to every team — regardless of technical skill.

🧩
For Non-Technical Users
Simple Mode — guided UI
No code. No setup. Just describe what you want in plain English — the AI Build engine generates your entire workflow automatically and loads it on the canvas.
Type "send Slack alert when GitHub issue is created" — done
60+ one-click templates — zero configuration needed
Drag and drop canvas — no programming knowledge required
Try it now
Click ✨ AI Build in the platform and describe your workflow in plain English.
For Speed-Critical Apps
Real-time AI — under 3 seconds
Switch to Groq speed mode for lightning-fast 70B responses. Perfect for live customer chat, real-time content generation, and instant AI replies — completely free.
Groq cloud delivers 70B responses in under 3 seconds
14,400 free Groq requests per day — no credit card needed
Switch between speed and depth with one click in the toolbar
Speed comparison
Groq: ~2s · Local 70B: 15-45s · Both free
📋
For Teams Without Technical Staff
One-click templates — zero config
Every template loads instantly with pre-configured nodes, connections and settings. Pick one, enter your API credentials, click Execute. That is the entire setup process.
💬 AI Chat
📧 Email AI
🐙 GitHub
📊 AutoML
📄 Doc Q&A
+55 more
60+ templates across AI, DevOps, Data Science, Marketing, CRM and more.
🔁
For Simple Automation
Zapier-style triggers — no AI needed
Not every workflow needs AI. Use Schedule, Webhook, HTTP and Filter nodes to build classic "when X happens, do Y" automations — at a fraction of Zapier's cost with no per-task pricing.
Schedule triggers — run daily, hourly, weekly automatically
Webhook triggers — react to events from any external service
Unlimited executions — flat monthly rate, no per-task fees
vs Zapier
Zapier charges per task. We charge a flat monthly rate — run 1 or 1,000,000 automations for the same price.
AI Engine

Your own private
AI supercomputer

Run a 70 billion parameter AI model on shared GPU infrastructure. Zero API costs, complete data privacy, enterprise performance.

⚡ NVIDIA Blackwell GPU — Dual GPU

Run the largest open-source AI models simultaneously at full speed. Your data never leaves your infrastructure.

Opensource AI 70B — Best reasoning
Default
Opensource AI 8B — Fast inference
Available
Llama 3.3 — Meta AI
Available
Qwen 2.5 — Best for code
Available
Run multiple models at once
260GB VRAM
No OpenAI bills ever
GDPR compliant
Unlimited tokens
260GB
Total VRAM (Dual GPU)
83.4T
FLOPS FP16 Combined
70B
AI Parameters
£0
AI API Cost
Use Cases

Built for every
workflow imaginable

From simple automations to complex AI pipelines — our visual canvas makes it accessible for everyone.

Workflow
Webhook — System alert received
🧠
AI Agent — Analyse error log
AI Engine 70B — Generate summary
🐙
GitHub — Create issue automatically
💼
Slack — Notify team with analysis

Intelligent IT Operations

When a system alert fires, your AI agent instantly analyses the error, generates a root cause analysis, creates a GitHub issue, and notifies your team — all in seconds, automatically.

Webhook receives alerts from any monitoring tool
AI agent analyses logs with full context
AI generates human-readable issue descriptions
Automatic team notification with root cause
Error handler routes failures to backup channel
# GPU Analytics — cuDF
import cudf, time
df = cudf.read_csv("data.csv")
t0 = time.time()
result = df.groupby("region")
  .agg({"revenue": "sum"})
  .sort_values("revenue",ascending=False)
✓ 10M rows in 0.031s on NVIDIA Blackwell
✓ Significantly faster than CPU pandas for large datasets

GPU-Powered Data Science

Process millions of rows with PySpark, accelerate analytics with 260GB dual GPU using cuDF, generate AI insights, and push results to your reporting tools.

PySpark handles datasets of any size
cuDF GPU analytics — 51x faster than CPU
Hive SQL for familiar SQL on big data
JupyterLab notebooks with GPU kernels
Results pushed to any destination
Step 1 — Analyse task
Breaking down research into 4 sub-goals...
Step 2 — Read sensors
Time: 09:42 UTC · Weather: Clear · News: 23 articles loaded
Step 3 — Synthesise
Key themes: open-source AI, GPU costs -40%, enterprise adoption 67%
✅ Final Answer
3 key insights with actionable recommendations ready.

Autonomous AI Agents

Build agents that reason, plan, and execute complex tasks using real-time sensor data and our 70B reasoning model. Every step traceable on the canvas.

Real-time sensors: weather, news, time, location
Multi-step ReAct reasoning fully visible
Configurable reasoning depth (2-8 steps)
Fully autonomous — no human intervention
Output to email, Slack, Telegram or any channel
⏰ Schedule — Every morning at 8am
📊 Fetch — Google Sheets + Airtable
◈ AI — Generate daily insights report
📧 Email — Send to all stakeholders
💼 Slack — Post to #reports channel

Business Process Automation

Automate any business process visually. Connect your apps, process data with AI, and trigger actions on any schedule — no code required.

Schedule, webhook, or event-based triggers
Connect Google Sheets, Airtable, Notion
AI generates personalised content
Multi-channel delivery in one workflow
Condition branches and loop logic built-in
📁 Upload raw CSV dataset
🧹 Auto-clean — fix missing, remove dups
🏆 Race 4 algorithms simultaneously
💾 Save XGBoost as best_model.joblib
🔮 Predict on new data instantly

No-Code Machine Learning

From raw messy data to a trained, deployed ML model — without writing a single line of code. AutoML races algorithms and picks the winner.

Automatic data cleaning and preprocessing
Races XGBoost, Random Forest, SVM, and more
Handles imbalanced datasets automatically
Best model saved and ready to deploy
Upload new data for instant predictions
Integrations

Connect to
everything

19+ pre-built integrations across communication, productivity, databases, AI and IoT. Universal HTTP node connects to any REST API.

📧
Email
💼
Slack
✈️
Telegram
🎮
Discord
🐙
GitHub
📊
Sheets
📓
Notion
🗃️
Airtable
💳
Stripe
🎯
Jira
🐘
PostgreSQL
Redis
📡
MQTT IoT
🤗
HuggingFace
🌤️
Weather
📰
News API
🔮
Replicate
🐦
Twitter/X
🐷
Pig Latin
🔗 + Any REST API via HTTP Node
Pricing

Simple, honest pricing.
Community tier always free to start.

All plans include unlimited workflow executions and the full Opensource AI Engine 70B. Advanced data science tools unlock in Pro.

💡
£1.87 / day
Get the complete Professional platform — Data Science Workbench, AutoML, JupyterLab and GPU Analytics. Cancel any day, no questions asked.
Cancel the same day
No monthly lock-in
Full GPU access
Start Free →
Community
£0/month
Self-hosted on your own infrastructure. Full platform access, forever free.
Visual workflow builder (30+ nodes)
AI Agent + Local 70B LLM
All 19+ integrations
Limited workflow executions
Webhook & schedule triggers
Templates library
Full REST API access
Data Science Workbench
AutoML Pipeline
JupyterLab + GPU notebooks
Get Started Free
Enterprise
£137.99/month
Dedicated infrastructure, compliance controls, and white-glove support.
Everything in Professional
Priority server access allocation
SSO / SAML / LDAP
On-premise deployment
Custom integrations
GDPR compliance reporting
Dedicated support engineer
4-hour response SLA
Team training & onboarding
Custom contract & invoicing
Contact Sales
Community

Loved by developers
and data scientists

Early Access
Be among the first to use it
We're in early access. We're not going to show you fabricated testimonials — instead, sign up free, try it yourself, and tell us what you think. Your feedback directly shapes the product.
Try it free → Send feedback
New Feature

Build any workflow
just by describing it

No drag and drop needed. Just type what you want in plain English — our AI engine reads your description and builds the entire workflow automatically in seconds.

1
Open the platform
Go to opensource-ai-agent.uk/app and the AI engine loads automatically.
2
Click ✨ AI Build in the toolbar
Find the glowing ✨ AI Build button at the top of the workflow canvas.
3
Describe your workflow in plain English
Type something like "Every morning fetch news, summarise with AI, and post to Slack" — or pick from the quick examples.
Workflow appears on canvas instantly
The AI builds and connects all the nodes automatically. Click ▶ Execute to run it immediately.
✨ Try AI Build Free →
✨ AI Workflow Builder
Describe your workflow
Every morning at 8am, fetch the latest AI news, summarise the top 5 stories using AI, and post the digest to our Slack #news channel
Quick examples
Send AI summary to Slack daily
Monitor errors → GitHub issues
Translate text automatically
Analysing your description...
Selecting best node types...
Building workflow structure...
4
Connecting nodes & configuring...
5
Loading onto canvas...
✓ Workflow Built — Ready to Execute
⏰ Schedule
📰 News API
◈ AI 70B
💼 Slack
Hybrid Intelligence

Always on. Never interrupted.

Your local 70B AI runs free on shared GPU. When it's busy, traffic automatically shifts to cloud providers — so your users never wait.

Automatic Failover Chain
🖥️
Local GPU
£0 / call
Blackwell · 70B model
4 parallel workers
PRIMARY
if busy/down
🚀
Groq Cloud
£0 / call
Llama 3.3 70B
✓ Active — 14,400 req/day
FREE TIER
if quota hit
Together.ai
£0.001 / 1K
Llama 3.3 70B
Pay as you go
FALLBACK 2
if needed
🤖
OpenAI
£0.15 / 1K
GPT-4o-mini
Most reliable
LAST RESORT
Failover happens in under 1 second — users never notice
💰
Zero AI costs by default
Your local 70B model handles everything for free. Cloud fallbacks only activate when needed — keeping your bills near zero even at scale.
£0 for 99% of requests
No OpenAI subscription needed
Cloud costs pennies when used
4 parallel AI workers
Ollama runs 4 simultaneous requests on your NVIDIA Blackwell GPU. No queuing, no waiting. Multiple users get instant responses at the same time.
4 concurrent AI requests
Flash attention enabled
Model stays warm (30 min cache)
🛡️
Auto-restart monitoring
Every service is monitored every 30 seconds. If anything crashes, it restarts automatically — before users even notice a problem.
5 services monitored 24/7
Auto-restart in under 5 seconds
Live health dashboard
⚡ View Live Hybrid Dashboard →
Free No credit card · No limits · Open source

Start building AI-powered workflows today

Community tier always free for self-hosted. Professional plan from just £1.87/day with no commitment. The most powerful AI workflow platform ever built.

🚀 Launch Platform Free ▶ Watch Demo
No credit card
Unlimited executions
Cancel anytime
Open source