Back to the galaxy

01 AI & Automation

Intelligence that worksfor you.

Not demos — production-grade deployments. I build AI systems that understand your business, handle real workflows, and scale without friction. Six of them run live on this page.

▶ See it live Ask Dan’s AI →
80%
Ticket resolution ratehandled without a human
<2s
Average response timeend to end
24/7
Always onlineno shifts, no queues
6
live demoson this page

02 What I build

AI THAT RUNS REAL OPERATIONS.

Six kinds of system, all shipped. Each one starts from a real workflow and ends with something your team can trust.

01

AI Support Systems

LLM-powered agents that read your knowledge base, understand context, and resolve customer tickets automatically. Handles 70–85% of inquiries without human intervention.

LLMRAGVector DBClaude API
02

Automated Workflows

End-to-end process automation — ticket routing, escalation logic, response generation, CRM updates. Systems that run 24/7 without supervision.

PythonWebhooksREST APICron Jobs
03

Calendar & Booking Integrations

AI that schedules, rebooks, and manages appointments. Connects to Google Calendar, Calendly, and custom booking systems with natural language interfaces.

Google APINLPWebhooks
04

Role-Based AI Access Systems

Multi-tenant AI platforms where different user roles access different AI capabilities, knowledge scopes, and permissions. Built for enterprise environments.

AuthMulti-tenantRBAC
05

Custom AI Agents

Purpose-built agents for specific business functions — data extraction, document summarisation, competitor monitoring, internal knowledge search.

AgentsTool UseMemory
06

RAG Knowledge Pipelines

Ingest your documentation, PDFs, databases, and internal wikis into a retrieval-augmented system. Your AI answers from your own data, not hallucinations.

EmbeddingsPineconepgvector

03 Live demos Running in your browser

TRY THE AI YOURSELF.

Real logic, no backend. The classifiers, parsers and pipelines below are the same shapes I run in production — only the model calls are replaced by fast local rules, so everything works offline and nothing you type leaves this page.

Demo A · Support agent

This exact system handles support for real businesses.

Pick a ticket — or write your own. The agent classifies it, retrieves the right policy from the knowledge base, drafts the reply and updates the CRM. Low-confidence tickets go to a human.

  • Ticket ingestion → intent + entities
  • Context retrieval from a RAG knowledge base
  • Response generation and CRM update — all automated
Riddle Support Agent · inboxProcessing
  1. Ingest
  2. Classify
  3. Retrieve
  4. Draft
  5. CRM

Demo B · Workflow automation

An inbox that routes itself.

Every email is read, classified and sent down the right branch — refund ticket, bug report, CRM lead or calendar booking — then the team is notified and everything is logged. Pick an email or write one and press Run.

  • Intent classification with a confidence score
  • Entity extraction: order numbers, amounts, dates, emails
  • Branching, retries and an audit log for every step
workflow · inbox-router.ymlArmed

        

      

    


    
    

      

        
Demo C · Booking assistant

        
Book, move and cancel in plain language.

        
The assistant understands dates, times and names, checks the calendar for conflicts, suggests the nearest free slot and keeps track of what “it” means in the conversation.

        
Try: book a meeting with Sarah tomorrow at 2pm · what’s free on Friday? · move it to 4pm · cancel it

      

      

        
calendar assistant · this weekSynced

        

          

          

            

            

            

          

        

      

    


    
    

      

        
Demo D · Bot console

        
Bots you can actually talk to.

        
Three real bot platforms I’ve shipped — wired to a working command parser right here in your browser. Pick a platform, type a slash command, hit enter. Try /help in any of them.

        
No backend. Pure JS demo of the same parsing layer I run in production.

      

      

        

        
RB
RiddleBot#0001@riddle-bot · v2.4.0
ONLINE

        

        

        
>

      

    

  







  

    

      

        
04 LLM · RAG · Vector DB

        
ASK DAN’S AI.

      

      
A working RAG demo, running entirely in your browser against a knowledge base about Dan. Every question is matched against 22 documents — watch the three best matches light up in the vector space.

    

    

      
      

        
DRAsk Dan · RAG · KNOWLEDGE BASE · LIVEOnline

        
  1. 01Embed query
  2. 02Vector search
  3. 03Rerank
  4. 04Generate

        

        

          
        

        

        
Engine · Knowledge-base RAG — instant, offline, free

      

    

    
In production this exact pipeline points at Pinecone, Chroma or Qdrant, embeds queries with OpenAI text-embedding-3-large, retrieves the top-k matches by cosine similarity and pipes them into Claude or GPT-4. The retrieval and citation logic here is real — only the embedding lookup is replaced by keyword scoring.

  







  

    

      

        
05 Document intelligence

        
Paste an invoice. Get clean data back.

        
The reader finds the vendor, invoice number, dates, line items and totals, checks that the maths adds up, and flags anything suspicious — overdue dates, totals that don’t match, changed bank details.

        
  • Structured JSON out, ready for your accounting system
  • Anomaly checks before anything gets paid
  • Bank details are masked in the output

      

      

        
document reader · extract.pyReady

        

          

          

            
Document

            
Extracted

          

          

          

        

      

    

  







  

    

      

        
06 Two Brains

        
LLM & RAG, OR CLASSICAL ML.

      

      
LLMs aren’t always the right tool. For pattern recognition with clean labels, time-series forecasting, latency-sensitive trades, or anything regulated — classical ML is faster, cheaper, deterministic, and easier to audit. I ship both, sometimes in the same system.

    

    

      
Path A · LLM + RAG
The reasoning brain.
Hosted LLM with a retrieval layer over your private knowledge. Trades reasoning ability and natural-language flexibility for higher latency and per-token cost. Best when answers must be explained, when domain knowledge changes weekly, or when the user is in the loop.

        
  • Claude 3.5 Sonnet / GPT-4 / open-weights (Llama, Mistral, Qwen)
  • Pinecone · Chroma · Qdrant · Weaviate · pgvector
  • OpenAI text-embedding-3-large / Cohere v3 / BGE-M3
  • Tool use, structured JSON outputs, prompt caching
  • Python execute-engine validates LLM output before it touches anything real

        
Best for: support, content, agents, analyst flows

      
Path B · Classical ML
The signal brain.
Pure pattern in, probability out. Trained on your labeled history. No language layer, no hallucinations, sub-millisecond inference, fully deterministic. The right choice when speed and certainty matter more than explainability.

        
  • XGBoost · LightGBM · CatBoost for tabular & time-series
  • LSTM · GRU · Transformer encoders for sequences
  • scikit-learn for the rest (regression, classification, clustering)
  • PyTorch when a custom architecture earns its keep
  • MLflow or W&B for experiment tracking, walk-forward validation

        
Best for: trading, fraud, churn, ranking, anomaly detection

    

    
// And yes — I ship hybrids. LLM for the strategy layer, ML for the execution layer.

  







  

    

      

        
07 Architecture

        
HOW THE STACK FITS TOGETHER.

      

    

    
    
      
  1. 01Data Ingestion
    Docs, PDFs, APIs, DBs → structured chunks
    2. 
      
  2. 02Vector Store
    pgvector / Pinecone semantic search index
    3. 
      
  3. 03LLM Layer
    Claude API — RAG context + system prompt
    4. 
      
  4. 04Automation
    Python workflows, webhooks, CRM sync
    5. 
    

  







  

    

      

        
08 Use Cases

        
INDUSTRIES I’VE DEPLOYED AI IN.

      

    

    

      
E-Commerce
Order & Returns Automation
AI agent handles order status queries, initiates returns, and processes refund requests without a human agent touching the ticket.
→ 78% of support load automated

      
SaaS / Tech
Technical Support Knowledge Base
RAG system trained on documentation, changelogs, and past support tickets. Users get accurate answers instantly, 24 hours a day.
→ <1.8s average resolution time

      
Hospitality
Booking & Reservation AI
Natural language booking agent integrated into the property's calendar system. Handles availability checks, bookings, and modifications.
→ 60% reduction in front desk load

      
Finance
Document Intelligence
AI pipeline that ingests contracts, invoices, and reports — extracts structured data, flags anomalies, and summarises for decision makers.
→ 90% reduction in manual review time

      
Healthcare Admin
Patient Intake Automation
Collects patient info, checks appointment availability, pre-fills forms, and routes to the correct department — all via a conversational interface.
→ 3x intake throughput

      
Internal Operations
Company Knowledge Assistant
Internal AI trained on your SOPs, HR policies, and product documentation. Employees get instant answers from your own knowledge base.
→ Hours of search time saved daily

    

  






  

    
09 Contact Open to full-stack roles · Dresden / remote

    
Let’s build something
people don’t log off from.

    

      Daanriddle@gmail.com
      
    

    

      Discord ↗
      GitHub ↗
      Tebex ↗
      Docs ↗
      About me →