Quick Start — Nexus One AI Portal

Open the chat interface

Open WebUI is your main access point — it works in any browser.

Go to http://ai.local:3001 in Chrome, Firefox, Edge, or Safari. No installation required — it runs entirely in the browser.

💡 If ai.local doesn't resolve, ask your administrator for the server IP and use http://[IP]:3001 instead.

Sign in

Use the credentials your administrator set up for you.

First person to sign up

The first account created becomes the admin automatically. Add other users from Admin Panel → Users after signing in.

Everyone else

Ask your admin to create an account, or enable open registration from Admin Panel → Settings → General.

Choose a model

Pick which AI to talk to. Start with Llama 3.1 8B for most tasks.

View all models →

Recommended

llama3.1:8b

Fast, general purpose. Good for Q&A, writing, summarising. Start here.

mistral:7b

Strong at reasoning and structured tasks. Good alternative.

llama3.2:3b

Very fast and lightweight. Use when speed matters more than depth.

Have your first conversation

Type a question or request and press Enter. Everything stays on your server.

Try these to start:

Click any prompt to copy it, then paste into Open WebUI.

Upload a document and ask about it

This is where Nexus One AI becomes genuinely useful — your documents, your questions.

In Open WebUI, click the + icon next to the message box, upload a PDF or Word doc, then ask questions about its contents.

Supported file types

PDF, DOCX, TXT, MD, CSV, XLSX. Start with files under 10 MB. PDFs must be text-based (not scanned images).

Questions to try

"Summarise this in 3 paragraphs."
"What does section 4 say about liability?"
"List all deadlines mentioned."
"Extract all the monetary values."

✓

You're set up. Explore what else is available.

Switch to the Platform Features tab above to see everything Nexus One AI can do.

🖼️

Multimodal Chat

Upload images, switch models mid-conversation, and access persistent chat history with search.

📄

Document Intelligence

Extract tables, compare documents side-by-side, and run batch processing across file sets.

🤖

Build an AI Agent

Create autonomous agents that run multi-step tasks, call tools, and act on your behalf.

🔧

Something not working?

Check the troubleshooting guide for common issues and fixes.

Intelligence

🖼️ Multimodal Chat Prompt Studio

Full-featured chat with persistent history, image upload, model switching per conversation, and markdown rendering. Search across all past conversations from the sidebar.

Go to: AI Tools → Multimodal Chat

📄 Document Intelligence

Upload PDFs and documents for extraction, comparison, and Q&A. Supports table extraction, batch processing, side-by-side document comparison, and structure parsing.

Go to: AI Tools → Document Intelligence

🎙️ Meeting Assistant

Record or upload meeting audio and get automatic transcription, AI-generated summaries, action item extraction, and follow-up email drafts — all on-premises.

Go to: AI Tools → Meeting Assistant

💬 Secure Chat Rooms

AI-assisted group conversations for your team. Create topic-specific rooms where members can collaborate with an AI moderator. Fully private, no cloud relay.

Go to: AI Tools → Chat Rooms

Automation

🤖 Agent Builder

Build autonomous AI agents that execute multi-step tasks, call external tools, read files, and act on instructions without human input for every step. Includes governance and approval workflows.

Go to: AI Tools → Agent Builder

🔄 Workflow Automation

Chain AI tasks together in visual pipelines. Trigger on schedule, webhook, or event. Connect input sources, AI processing steps, and output destinations into repeatable flows.

Go to: AI Tools → Workflow Automation

⏰ Scheduled Jobs

Set AI tasks to run automatically on a cron schedule — daily reports, nightly document processing, weekly summaries. Jobs run on the server even when no one is logged in.

Go to: AI Tools → Scheduled Jobs

🔌 Connectors

Connect Nexus One AI to your existing systems — databases, file shares, APIs, email, and enterprise tools. Connectors let agents and workflows pull live data from your environment.

Admin → Tools → Connectors

Knowledge & RAG

📚 Knowledge Base

Upload and organise documents into collections. The AI searches these at query time using RAG — so responses are grounded in your actual content rather than model training data.

Admin → Manage → Knowledge Base

📊 RAG Quality Dashboard

Monitor retrieval accuracy, chunk quality scores, and query-document relevance. Identify which collections return good results and which need reindexing or tuning.

Admin → Tools → RAG Quality

🧪 AI Eval Suite

Run standardised test suites against your models and prompts. Compare output quality, track regressions, and validate that fine-tuned models perform better than their base versions.

Go to: AI Tools → AI Eval Suite

⚡ Model Benchmarking

Run speed and quality benchmarks across installed models. Compare tokens/second, TTFT (time-to-first-token), and task-specific accuracy to choose the right model for each use case.

Admin → Tools → Benchmarking

Control & Safety

🛡️ Guardrails

Define content filters, keyword blocks, PII redaction rules, and prompt injection detection. All queries pass through guardrails before reaching the model and after the response is generated.

Admin → Tools → Guardrails

🔀 Model Router

Automatically route queries to the most appropriate model based on task type, complexity, or cost. Send code questions to CodeLlama, simple tasks to the fast 3B model, complex reasoning to 70B.

Admin → Tools → Model Router

🔑 API Keys

Generate and manage API keys for programmatic access to Nexus One AI from your own applications. Set per-key rate limits, allowed models, and expiry dates.

Admin → Tools → API Keys

🎓 Model Training

Fine-tune models on your own data using QLoRA — efficient training that fits on your existing GPU. Upload a dataset, configure training parameters, and produce a domain-adapted model.

Admin → Manage → Training

⚙️ This section is for system administrators. Regular users can skip to First Steps.

Verify all services are running

Check the dashboard for green status on all core services.

Dashboard →

Open the Dashboard and confirm Ollama, Open WebUI, ChromaDB, and the Cezen backend all show as running. If any are red, check the troubleshooting guide.

Create user accounts

Add accounts for each team member before they try to log in.

Users →

Go to Admin → Users to create accounts. Assign roles: Admin for IT staff, User for everyone else. You can also enable open registration if your network is sufficiently restricted.

Load AI models

Ensure the recommended models are installed and available.

Model Manager →

Go to Admin → Model Manager to see installed models. The recommended starter set: llama3.1:8b, mistral:7b, nomic-embed-text (required for RAG). Pull additional models as needed.

Set up guardrails

Configure content filters appropriate for your organisation before users start.

Guardrails →

At minimum, enable PII redaction and keyword blocks for sensitive terms. Guardrails can be adjusted at any time — start conservative and loosen rules as you understand your team's usage patterns.

Seed the Knowledge Base

Upload your organisation's core documents so users can query them immediately.

Knowledge Base →

Create collections by department (HR Policies, Finance SOPs, Legal Contracts). Upload the documents and let ChromaDB index them. Users can then query these collections from any AI tool. Check RAG Quality after indexing to confirm retrieval accuracy.

✓

System ready for your team

Share the portal link and the First Steps tab with your users.

📊

Usage Analytics

Monitor who's using what, query volumes, and model performance across your team.

📋

Audit Log

Full record of all queries, file uploads, and admin actions for compliance.

🔀

Model Router

Set routing rules so the right model is used automatically for each task type.

🔌

Connectors

Connect to your existing systems so agents and workflows can access live data.

Up and Running in 5 Minutes

Open the chat interface

Sign in

First person to sign up

Everyone else

Choose a model

Have your first conversation

Upload a document and ask about it

Supported file types

Questions to try

You're set up. Explore what else is available.

Multimodal Chat

Document Intelligence

Build an AI Agent

Something not working?

🖼️ Multimodal Chat Prompt Studio

📄 Document Intelligence

🎙️ Meeting Assistant

💬 Secure Chat Rooms

🤖 Agent Builder

🔄 Workflow Automation

⏰ Scheduled Jobs

🔌 Connectors

📚 Knowledge Base

📊 RAG Quality Dashboard

🧪 AI Eval Suite

⚡ Model Benchmarking

🛡️ Guardrails

🔀 Model Router

🔑 API Keys

🎓 Model Training

Verify all services are running

Create user accounts

Load AI models

Set up guardrails

Seed the Knowledge Base

System ready for your team

Usage Analytics

Audit Log

Model Router

Connectors