UK’s Leading AI Automation AgencyTransform Your Business with AI

AI Powered Chat Bot

# ITSM Hub: AI Automation Transforms Training Operations

## Client Overview

**ITSM Hub** is a leading provider of IT service management training across Australia, New Zealand, the UK, and US, specializing in ITIL, PRINCE2, DevOps, and cybersecurity certifications. With a growing customer base and increasing demand for their virtual training programs, they needed to scale their operations while maintaining exceptional customer service.

## The Challenge

ITSM Hub faced several operational bottlenecks:

- **Manual Customer Service**: High volume of course inquiries requiring individual staff responses
- **Administrative Overhead**: Time-intensive manual updates to course schedules and pricing
- **Inconsistent Information**: Varying response quality and accuracy across different team members
- **Scalability Constraints**: Limited ability to handle peak inquiry periods effectively
- **Geographic Complexity**: Managing multi-currency pricing and location-specific course offerings

## The Solution

AI Smarter implemented a comprehensive AI automation platform comprising three integrated components:

### 1. Intelligent Web Chatbot
An AI-powered customer service assistant that handles course inquiries, pricing questions, and scheduling requests in real-time. The system provides location-aware responses and supports multiple currencies (AUD, USD, GBP, NZD).

### 2. Automated Email Processing
Smart email automation that processes course schedule changes and pricing updates sent by staff. The system automatically updates databases, maintains audit trails, and sends professional confirmation emails.

### 3. Daily Management Reporting
Automated generation of comprehensive daily reports summarizing customer interactions, escalations, and operational metrics, delivered directly to management.

## Technical Architecture

The solution leverages modern cloud technologies:

- **AI Engine**: OpenAI GPT-4o with custom prompt engineering for domain expertise
- **Automation Platform**: n8n workflow automation for seamless process integration
- **Database**: Supabase PostgreSQL with real-time capabilities and comprehensive audit logging
- **Frontend**: Responsive web interface optimized for both desktop and mobile users
- **Security**: Enterprise-grade encryption, role-based access controls, and GDPR compliance

## Results Achieved

### Customer Service Excellence
- **98% Success Rate**: AI chatbot successfully resolves customer inquiries without human intervention
- **Instant Response**: Average response time reduced from hours to under 2 seconds
- **24/7 Availability**: Customers receive immediate assistance regardless of time zone
- **Consistent Quality**: Standardised, accurate information delivery across all interactions

### Operational Efficiency
- **95% Automation**: Course schedule and pricing management now fully automated
- **Error Reduction**: 90% fewer manual data entry errors
- **Staff Productivity**: 60% increase in time available for high-value activities
- **Process Reliability**: Complete audit trails for all system changes

## Key Innovations

### Geographic Intelligence
The system automatically recognises customer locations and provides relevant course information, ensuring New Zealand customers see Australia/NZ courses with appropriate currency pricing.

### Smart Course Recognition
Advanced pattern matching recognizes common abbreviations (CSAM, ITIL, P2) and course variations, dramatically improving search accuracy and customer experience.

### Professional Communication
Automated email responses maintain ITSM Hub's professional brand standards while providing detailed, actionable information to staff and customers.

### Comprehensive Audit Trail
Every system change is logged with complete before/after tracking, ensuring regulatory compliance and enabling detailed performance analysis.

## Implementation Success

The project was delivered in 4 weeks with zero downtime during deployment. Comprehensive testing included 90 test cases across all system components, achieving production-ready reliability from day one.

**Key Success Factors:**
- Deep understanding of ITSM Hub's business processes and customer needs
- Robust technical architecture designed for scalability and reliability
- Extensive testing and validation before production deployment
- Collaborative partnership approach throughout the implementation

## Conclusion

The ITSM Hub AI automation solution demonstrates how thoughtfully implemented AI can transform business operations while enhancing customer experience. By automating routine tasks and providing intelligent customer service, ITSM Hub has positioned itself for sustainable growth while maintaining the personal touch that defines their brand.

The 98% automation success rate, combined with significant cost savings and improved customer satisfaction, showcases the transformative potential of AI when applied strategically to real business challenges.

---

**About AI Smarter**

AI Smarter specialises in developing custom AI automation solutions that deliver measurable business value. Our expertise in natural language processing, workflow automation, and enterprise integration helps organisations achieve operational excellence through intelligent automation.

**Ready to transform your operations?** Contact us at www.aismarter.ai

AI Training

Tailored half day training to support writing and image generation with AI tools.

Walked through basic to advanced prompt engineering and use of 'secret' tools, to provide high quality output.

Within 2 hours, Jason was able to master our process for waterfall prompt engineering using the CLEAR method, and we were able to produce a high quality illustration in the style of his chosen illustrator to complement a chapter in the children's book that he is writing. This alone has saved him weeks of time and the cost of employing a professional illustrator .

Mike’s YouTube → Newsletter Automation

Case Study

Turning a manual copy-and-paste grind into a hands-free content engine.

⸻

1. Project Snapshot
• Brief – Mike, a fast-growing YouTube finance influencer, wanted every new video transcript transformed into a polished email newsletter—without lifting a finger.
• Goal – Kill the copy-and-paste routine, keep the weekly newsletter cadence rock-solid, and give Mike back his evenings.

⸻

2. From Pain to Objectives

The pain
• After each upload, Mike downloaded the auto-generated transcript, cleaned timestamps, wrote summaries, formatted a MailerLite campaign and queued the send—2–3 hours every release.
• Any missed step delayed the newsletter, annoying 40 000+ subscribers hungry for fresh insights.
• Manual workflow meant no time for A/B tests or list-growth experiments.

Targets
1. End-to-end automation, triggered within minutes of a video going live.
2. Reduce content-prep time from hours to seconds.
3. Preserve Mike’s “human” tone while improving consistency.

⸻

3. Solution Overview
• YouTube webhook – Fires when a new video is published; kicks off an n8n flow.
• Transcript fetch & clean – Python micro-service pulls captions via YouTube API, strips timestamps and filler words.
• AI summariser – OpenAI GPT-4o condenses ~7 000 words into a 300-word “key takeaways” section that sounds like Mike.
• Newsletter assembly – Liquid template merges summary, video link, call-to-action and sponsor slot; Markdown converted to responsive HTML.
• MailerLite API push – Email created, subject-line A/B variants added, scheduled to deploy at Mike’s optimum send time.
• Notion content hub update – The same automation writes the cleaned transcript, summary, sponsor and performance tags to Mike’s Notion database, keeping his editorial calendar and evergreen content library in sync.
• Archive & analytics – Supabase stores raw transcript, summary, click-thru stats; n8n posts a Slack notification so Mike can preview (or veto) before send.

⸻

4. Implementation Highlights
• Notion API integration – Each new entry includes video URL, summary, tags and a link to the live newsletter, creating a searchable knowledge base for future repurposing.
• Fallback guardrails – If transcript pull fails, flow pauses and pings Mike’s VA instead of sending a blank email.
• One-click resend – Missed the Friday window? Trigger a manual rerun by reacting with 📧 in Slack.

⸻

5. Benefits & Early Wins (first month)
• Time saved – ~6 hours per week reclaimed (four newsletters).
• Newsletter punctuality – 100 % on-time sends (versus 60 % before).
• Open rate – 32 % → 38 % thanks to consistent formatting and A/B subject lines.
• Subscriber growth – +11 % as Mike added a “get the notes by email” CTA to each video.
• Zero additional spend – Open-source stack; incremental OpenAI tokens cost < $20 / month.
• Centralised knowledge – Notion now houses every transcript and summary, fuelling blog posts, social snippets and future book projects.

⸻

6. Tech Stack Snapshot
• Automation – n8n (Docker on AWS ECS)
• AI engine – OpenAI GPT-4o
• Transcript fetch – Python + youtube-transcript-api (Lambda)
• Data store – Supabase Postgres (stats & archives)
• Email – MailerLite REST API
• Knowledge base – Notion API (content calendar & library)
• Monitoring – Grafana Cloud (latency, token spend, failure alerts)

⸻

7. What’s Next?
• Multilingual summaries for Mike’s Spanish channel launch.
• Dynamic RSS feed for subscribers who prefer reading on Feedly.
• Personalised section (“Top three links for you, ”) using behavioural data from Supabase.

⸻

8. Key Takeaway

Automating the transcript-to-newsletter pipeline—and synchronising every asset with Notion—freed Mike to focus on content creation, not content repurposing. The result: happier subscribers, healthier metrics and a creator who finally logs off before midnight.

⸻

Vanished VPN: Zero-Touch Support with an AI Chatbot

Case Study

How an always-on web assistant wiped out the ticket queue—and the support bill.

⸻

1. Project Snapshot
• Brief – Give Vanished VPN subscribers instant, accurate answers to common questions (setup, streaming, billing) through a web-based AI chat widget.
• Goal – Eliminate routine tickets, shrink response time to seconds and let the small team focus on product growth instead of inbox firefighting.

⸻

2. From Pain to Objectives

The pain
• 80 % of support requests repeated the same queries (device config, “Why can’t I watch Netflix?”).
• Peak-hour surges meant customers sometimes waited 24 h+ for a reply.
• Contracted help-desk licences and after-hours staffing chewed through the budget.

Targets
1. Cut human-handled tickets to near-zero.
2. Deliver answers in < 10 s, 24 / 7 / 365.
3. Achieve a 100 % reduction in third-party support spend within one quarter.

⸻

3. Solution Overview (simplified)
• AI brain – OpenAI GPT-4o fine-tuned on VPN docs and 50 000 historic tickets, producing natural, brand-aligned replies.
• Automation hub – n8n orchestrates chat flow, confidence checks, fallback rules and data logging; tweaks take minutes, not sprints.
• Knowledge base – Supabase Postgres stores markdown articles, network-status API data and usage tips; update once, surface everywhere.
• Web widget – Lightweight React component embedded on vanishedvpn.com; loads under 50 kB and works across desktop and mobile.
• Daily updater – Python cron parses release notes and blog posts, refreshing KB entries so the bot stays current without manual effort.

⸻

4. Implementation Highlights
• Cold-start training – domain fluency achieved in 48 h.
• Smart fallback – if confidence < 90 % or account lookup needed, n8n opens a ticket with the bot’s draft answer for rapid human review.
• Dynamic status checks – bot pings an internal API to confirm server uptime before advising region switches.
• Usage telemetry – every Q&A logged to Supabase for trend analysis and KB gap spotting.
• Web-first deployment – adds < 100 ms to page load time.

⸻

5. Benefits & Early Wins (first 60 days)
• Support cost – external help-desk contract cancelled: 100 % reduction.
• Ticket volume – down 92 %; remaining 8 % are edge-cases routed to engineers.
• Average response time – 6 h → 4 s.
• Customer CSAT – 4.3 → 4.9 / 5.
• Staff focus – 15 hrs/week freed for roadmap features and marketing.
• Self-healing KB – bot improves with every interaction; no brittle intent trees.

⸻

6. Tech Stack Snapshot
• Automation & routing – n8n (Docker on AWS ECS)
• Data store – Supabase Postgres (Sydney region)
• AI model – OpenAI GPT-4o (function-calling for API look-ups)
• Scraping/updates – Python + Feedparser (daily cron via AWS Fargate)
• Frontend widget – React + Stitches, served from CloudFront CDN
• Monitoring – Grafana Cloud (latency, token spend, fallback rate)

⸻

7. What’s Next?
• Multilingual expansion – Spanish & Japanese models in Q3.
• Pro-active tips – contextual “Did you know?” nudges based on user agent and geolocation.
• Slack & Discord bots – reuse the same knowledge base to support power users where they hang out.

⸻

8. Key Takeaway

With a lean open-source stack and a purpose-trained GPT-4o model, VanishedVPN turned support from a cost centre into zero cost. The web AI chatbot now resolves virtually every customer query in seconds, delighting users and freeing the team to vanish into bigger, bolder projects.

⸻

Ibiza AI Concierge

Case Study

Ibiza AI Concierge
An always-on, multi-channel assistant that sells the dream while saving the team.

⸻

1. Project Snapshot
• Brief – Build a friendly, knowledgeable AI assistant that handles enquiries on WhatsApp, Facebook Messenger and Instagram DM.
• Goal – Answer anything Ibiza-related, upsell VIP offers and—when a prospect shows serious buying intent—spin up a group chat with human sales lead Karen to close the deal.

⸻

2. From Challenge to Objectives

The pain:
• Enquiries roll in around the clock via phone, WhatsApp and social DMs.
• Staff are swamped during peak hours and rarely available after office hours.
• Missed or delayed responses mean lost opportunities to upsell high-margin experiences (e.g. VIP tables, boat charters).

Targets:
1. One workflow for every channel, 24 / 7.
2. Instant, accurate answers whatever the time zone.
3. Automatic hand-off to a human only for premium sales.

⸻

3. Solution Overview
• n8n orchestration – the automation backbone gluing channels, AI and data together (advanced features, minimal licence cost).
• WAHA on Coolify / AWS – scalable WhatsApp API endpoint.
• Supabase Postgres – single source of truth for venues (≈ 2 700 events), VIP pricing, chat history.
• Weekly Python scraper – refreshes event data; n8n loader pumps updates into Supabase.
• Dynamic knowledge base – the assistant pulls fresh info on demand, so every reply is up-to-date.
• Smart escalation – detects boat-charter interest, opens WhatsApp group with Karen plus context dump.

⸻

4. Implementation Highlights
• One virtual employee trained on “all things Ibiza”, fluent across three channels.
• Omnichannel webhooks feed a single n8n workflow, ready for future channels in minutes not weeks.
• Real-time upsell engine draws on the VIP table to surface tailored offers mid-chat.
• Stateless hosting keeps monthly infra spend < £80 while scaling effortlessly in peak season.
• 300+ QA tests ensured tone, accuracy and escalation triggers worked exactly as designed.

⸻

5. Benefits & Early Wins
• 24 / 7 coverage – no enquiries lost after club closing time.
• ≈ 30 % uplift in upsell revenue (VIP tables, boat charters) within four weeks.
• Response time cut from hours to sub-10 seconds.
• Human workload reduced by ~20 hrs per week, freeing agents for high-touch concierge tasks.
• Cost efficiency – open-source stack + n8n achieved ROI inside two months and slashed SaaS bills by > 70 %.
• Actionable insights – every message logged to Supabase for marketing analysis and continual AI tuning.

⸻

6. Tech Stack Snapshot
• Go-between: n8n (automation & routing)
• Hosting: Coolify on AWS
• Messaging: WAHA (WhatsApp API)
• Data layer: Supabase Postgres
• Scraping: Python + BeautifulSoup (weekly cron)
• AI engine: OpenAI GPT-4o (knowledge + intent detection)

⸻

7. What’s Next?
• Sentiment scoring to flag unhappy guests instantly.
• Dynamic pricing suggestions based on real-time ticket availability.
• Voice notes → text → AI replies for hands-free travellers.

⸻

8. Key Takeaway

A single, well-orchestrated AI assistant can turn a chaotic, multi-platform enquiry stream into a seamless concierge and sales channel—delivering lightning-fast answers, richer customer experiences and a measurable boost to the bottom line, all while keeping operating costs firmly under control.

⸻

Automation of Web Enquiries

Case Study

Automating Enquiries for Binding Financial Agreements (BFAs)
From white-knuckle manual replies to fully-orchestrated AI conversations

⸻

1. The Challenge

Responding to BFA enquiries used to be painfully manual:
1. Each webform submission landed in an inbox.
2. A team member looked up (or created) the contact in OntraPort.
3. A bespoke email was drafted (new BFA vs. review).
4. Someone tried to phone the prospect to confirm receipt and next steps.

The result? Slow response times, patchy follow-up and an ever-growing spreadsheet of “we’ll call them tomorrow”.

⸻

2. Objectives
• Cut first-response time to under five minutes.
• Eliminate manual data entry and email drafting.
• Guarantee every prospect receives a confirmation call.
• Provide real-time pipeline visibility.

⸻

3. The Solution

We designed an end-to-end, event-driven workflow that talks back:
• GoHighLevel (GHL) replaced OntraPort as the single source of truth, unlocking richer native automation features and driving a significant reduction in platform costs.
• A custom webform instantly creates a contact and opportunity in GHL.
• Smart email templates auto-select wording (new BFA or review) and send immediately.
• AI Voice Assistant “Alex” (Retell AI + ElevenLabs) dials the prospect within seconds, confirms the email, answers FAQs and—thanks to live calendar integration—books an appointment on the spot.
• n8n workflows keep everything in lock-step: log call transcripts, advance pipeline stages and alert the team only when human input adds value.

⸻

4. Implementation Highlights
• GHL Roll-out & Data Clean-up – migrated and scrubbed legacy data; built a tailored BFA pipeline.
• Webform → Contact/Opp Creation – JSON payloads mapped in n8n; lead appears in CRM in under a second.
• Dynamic Email Logic – Liquid variables + conditional snippets ensure tone-perfect replies every time.
• AI Dial-back – Twilio number + Retell AI delivers a natural Aussie voice, 24 × 7.
• Calendar Booking – live availability surfaced to Alex; no more back-and-forth scheduling.
• Auto-Notes & Transcripts – call details pushed to GHL for full context before any human follow-up.

⸻

5. Outcomes (first six weeks)
• Average response time: 3 h 45 m → 58 s
• Follow-up consistency: 67 % → 100 %
• Conversions: ≈ 50 % uplift (more enquiries turning into booked consultations)
• Admin time saved: ≈ 12 h per week (one quarter of an FTE)
• ROI: project costs recouped in under two months
• Pipeline visibility: live Kanban replaced static spreadsheets

⸻

6. Tech Stack

Layer Tool
CRM & automation brain GoHighLevel
Workflow orchestration n8n
Voice interface Retell AI
Natural Aussie voice ElevenLabs
Telephony Twilio (+61 3 … dedicated number)
Language cognition OpenAI

⸻

7. What’s Next?
• Sentiment analysis on call transcripts to pre-qualify prospects.
• SMS fallback when calls go unanswered.
• Self-service portal for clients to upload supporting documents.

⸻

8. Key Takeaway

Automation isn’t about replacing people—it’s about eliminating the dull bits so people can focus on high-value conversations. By giving prospects an instant, personalised touch (email and phone), we transformed a laggy process into a slick, always-on customer experience—and paid for the project in less than eight weeks.