Sell more. Serve more. Be more.

Inside EPIC CSX. A working demonstration of how Customer Contact, Service & Experience should feel in 2026.

For twenty years I have led contact, service, and experience transformations across more than six sectors and thirty countries. Banking. Aviation. Telco. Retail. Luxury travel. BPO. Financial services. Public sector. Teams of forty. Teams of four thousand. Greenfield builds and brownfield turnarounds.

The pattern is always the same.

Contact centres are run as cost centres. Measured on AHT, abandonment, and headcount. Squeezed. Offshored. Automated. And leaking revenue.

Every. Single. One.

This is what I built to fix that, and what it could look like inside your business.

The thesis: your contact centre is a revenue engine

Most operators have it the wrong way around. They treat the contact centre as the place customers go when something has gone wrong, and they design it to make those interactions cheap.

Cheap interactions cost more than they save. They cost you sales conversion. They cost you retention. They cost you upsell. They cost you the next booking. They cost you the recommendation. They cost you the lifetime value of customers who chose someone else next time, quietly, without complaining.

The EPIC CSX operating model treats every interaction as a commercial moment. Sales conversion, retention, upsell, and rebooking are designed into the service experience, not bolted on as scripts.

AI first. Human always one tap away.

The line that defines EPIC CSX is the one most operators get wrong. AI is not a replacement for human service. It is the layer that decides, in real time, when AI adds the value and when only a human can.

The AI just decided. James Harrington at 94% match. Six channels offered. The customer chooses.

When MJ types her first message, the AI captures every entity, builds the context, scores intent, and matches her to the right specialist Ambassador in ninety seconds. Then it does the thing most systems never do.

It hands over.

James Harrington is offered through whichever of the six channels MJ prefers. WhatsApp, phone, video, callback, email, live chat. Every channel arrives inside one Portal. James sees the entire conversation history before he opens his mouth. Zero repetition.

This is the operating model in one move. AI handles what AI is best at. Humans handle what only humans can. The customer never feels the seam.

What has been reimagined

I built EPIC CSX over a decade ago, before AI was the answer to every question.

Thirty-six capabilities. Four pillars: Easy · Personalised · Intuitive · Contextual. Twenty employee experience principles across Time · Information · Motivation · Empowerment.

When AI arrived, I did not throw EPIC CSX away. I added AI to a foundation that was already working.

That is the difference. Most operators are layering AI onto broken processes. The customer does not feel the AI. They feel the brokenness, faster.

I do it the other way around.

Meet MJ. One guest, eight relationships, one operating model.

I built a working demonstration. Not a deck. Not a video. A live, interactive walkthrough that takes a senior decision-maker through fourteen chapters of one guest's journey, end to end.

Meet MJ. Platinum member, EPIC-2847591, twelve thousand four hundred and fifty points.

MJ is travelling with her husband David and six others to celebrate twenty-five years of marriage. Twelve to nineteen June. Ocean Penthouse. Budget thirty thousand pounds.

David has a severe peanut and tree nut allergy.

Watch what the operating model does.

Chapter 1 · Recognised before she types

Twelve voyages, £148K LTV, James Harrington matched at 94%, David's allergy protocol live. All before MJ has typed a word.

The portal opens MJ's profile before she sends her first message. Twelve-voyage history. £148K lifetime value. Severe allergy protocol. A 94%-matched Ambassador named James Harrington.

MJ types one natural-language message. The system extracts five live entities in real time: anniversary, eight guests, Mediterranean, June, severe nut allergy. The Ambassador does not type any of this. They act on it.

This is what I mean by Easy. No forms. No "have you got your booking reference." No friction.

Chapter 2 · The quote builds itself

A living quote, not a PDF. Shared in one tap. Group built automatically. David's allergy enforced before anyone replies.

James gets a complete brief: profile, allergy, loyalty, LTV, intent score, preferences. Zero context loss across the handoff.

The quote builds itself from the conversation. Ocean Penthouse £22,500. Chef's Table ×3 £900. Anniversary Enhancement £650. Couples Spa £1,800. Curated Excursions £2,400. Platinum 15% discount applied automatically.

Total: £24,013. £5,987 under MJ's £30,000 budget, flagged in green. The AI spots the gap and proactively recommends three nut-safe enhancements that still leave £4,927 on the table.

This is what I mean by commercial. The Ambassador is not selling. They are advising. And the basket is still bigger than the brief.

Chapter 3 · One enquiry, eight relationships

Six of eight have voted on the anniversary dinner. Chef's Table leads with five votes. The AI tracks every signal and nudges the two who haven't voted yet.

Most systems treat group bookings as eight separate transactions. EPIC CSX treats them as one intelligent journey with eight connected commercial profiles.

One tap shares a living quote with the whole party. Seven additional guest profiles created on the share. Dietary requirements captured instantly. David's severe allergy and Sophie's vegetarian both flagged. A live group poll asks where the anniversary dinner should be: Chef's Table wins five votes, Terrace two, Ocean Grill one, AI auto-nudges the non-voters.

When Ryan asks about flights inside the group chat, the AI answers him contextually and schedules a callback. Nobody had to relay the message. Nothing fell through.

One enquiry. Eight commercial relationships, each one growing.

Chapter 4 · Video Intelligence

Twenty seconds of video. Sentiment analysis live. Emotional dip detected the moment the allergy is mentioned. Before MJ has finished speaking.

MJ records a 20-second video. Not a phone call she has to schedule. Not a form she has to fill in. A video.

The system runs real-time speech-to-text. Pulls live entities from her words. Tracks second-by-second sentiment alongside the playback. Flags an emotional micro-dip at the exact moment she mentions David's allergy.

It then ranks the next best actions, with measurable sentiment lift attached to each one: Recommend Ocean Penthouse. Book Chef's Table. Activate allergy protocol across nine venues (CRITICAL). Escalate to James (RECOMMENDED).

This is the channel that will define the next decade of premium service.

But the differentiation is not that MJ can send a video. It is what comes back.

James responds in video. Same medium MJ used. Suite, dining, allergy protocol all visible alongside his words.

James does not type a reply. He records his own video back to her, while the system surfaces the relevant suite, dining, and protocol cards alongside his words. "MJ, happy anniversary to you and David, twenty-five years, that's wonderful. I've got the perfect setup in mind for your group..." He addresses every concern she raised, by name. He shows what she asked to see.

This is what differentiated luxury service looks like in 2026. Not faster typing. Not chatbot scripts. Eye contact, voice, presence, and full context. Both ways.

I have built it.

Chapter 5 · Mentioned once, enforced everywhere

The Head Chef personally takes ownership. Not a notification. A relationship. David's safety, named, owned, on record.

The allergy follows David, not MJ. Always David. Never MJ.

Mentioned once in conversation. Propagated automatically across fourteen touchpoints: every restaurant, every bar, room service, the galley, the spa, the buffet. Head Chef Marco Bellini personally acknowledges the protocol in a confirmation thread to MJ herself. Name, signed, on record.

Zero manual follow-up. The protocol follows David everywhere on board.

This is Personalised at the level it actually has to operate. Safety. Identity. Recognition. At scale, on a moving ship, across a thousand staff who never met him.

Chapter 6 · The portal in her pocket

Day three on board. MJ is relaxing on Deck 7 between the spa and dinner. She has a small question. She opens the portal in her pocket.

One question, asked from a sun lounger. Six channels in the operating model, one truth, no repetition.

"Is there a sunset yoga class this evening? I'd love to do something relaxing before dinner."

The AI Concierge answers in seconds. Tonight's Sunset Flow at 6:30pm, Sky Deck, 45-minute Vinyasa, mats and towels provided, perfect before her 8pm dinner reservation.

But MJ wants something more.

"Actually, could we do something for the whole group? A few of us were talking about it at lunch."

AI recognises the group booking moment. It does not improvise. It bridges to James in the same thread, with full context.

This is the moment most operators get wrong. A group request inside a 1:1 service chat needs human judgement. The AI sees it. It does not transcribe and forward to a separate team. It does not improvise.

It bridges.

James appears in the same thread, within seconds, with the entire context already in front of him. He confirms a private group session at 6:15pm on the Sky Deck. Sunset views, dedicated instructor, mats and towels, chilled coconut water laid out for arrival. The same chat thread MJ opened with her one quiet question.

James posts the booking to the group chat. The whole group says yes within seconds.

James posts to the group chat. The replies arrive in seconds. MJ. David. Claire. Tom. The whole group joining the experience James and the AI orchestrated together. No emails. No phone calls. No "I'll loop everyone in." Same channel. Same context.

One question. Eight bookings. Same channel.

Chapter 7 · When things go wrong

Brand Sentiment crashes from 8.5 to 3.2 in seconds. £89,000 of lifetime value at risk. The system sees it before James has picked up his phone.

Day three of the voyage. David is offered food with traces of tree nuts at the Pool Bar buffet.

Brand Sentiment crashes from 8.5 → 3.2 live on the panel. The system detects it in seconds. James is notified. Revenue-at-risk is calculated live: £89,000 of LTV at risk.

Four minutes from incident to James on the line. Four parallel actions. Each one carrying a measurable sentiment impact. Recovery, designed.

James calls within four minutes of the incident. Not after the email of complaint, not after the negative review, but in real time. The AI briefed him before he dialled. He does not ask "what happened?" He says "I know what happened, and here is what I have done."

Four parallel actions fire alongside that call. Personal phone call. Complimentary upgrade. Private dining with the Head Chef. Extended spa package. Each one carries a measurable sentiment impact tracked live on the panel.

Chapter 8 · Service became revenue

Brand Sentiment ends at 9.4. Higher than before the incident. Greek Islands rebook follows. Service became revenue.

Brand Sentiment ends at 9.4. Higher than before the incident. Two months later, the AI spots a rebooking signal in MJ's browsing behaviour and generates a personalised offer: Greek Islands. September. £26,200. James sends it on WhatsApp. MJ rebooks.

LTV £148K → £193K. +30.4%.

One enquiry, fourteen chapters, forty-seven touchpoints, six channels, eight commercial relationships, one coherent journey. Service became revenue. Every metric proves the operating model.

Brand Sentiment, not NPS

NPS is a survey that asks customers what they thought after the fact. It is slow, voluntary, and gamed.

Brand Sentiment is a continuous, real-time, behavioural metric. It updates from conversation tone, browsing activity, positive and negative language, CSAT ratings, and behavioural signals. It moves second by second. It tells the operator what the customer is feeling right now. Not what they remembered to tick on a form three weeks later.

You cannot recover what you cannot see.

This is the measurement layer modern CX has been waiting for. It is also the metric that makes the executive case for transformation defensible at board level. Brand Sentiment moves before churn shows up in the numbers. By the time churn shows up, it is too late.

Trust through openness

This is the part most operating models get wrong.

Customer support pages are usually designed for the brand, not the customer. Buried. Hard to find. Deflecting. A black box where complaints go to die.

EPIC CSX inverts it.

Thirty-eight interactions across six channels. MJ sees everything. James sees everything. Nothing hidden.

Every WhatsApp, every video, every call, every email, every chat, every SMS, every callback, every Portal action. All of it logged automatically as it happens. Not summarised from memory. Not copied manually after the fact. Captured in real time. Categorised. Time-stamped. Channel-tagged.

MJ sees the same view James does.

One to five stars on every interaction. Each rating triggers a contextual response. Brand Sentiment updates live. Behavioural, not a survey.

And she rates each interaction one to five at the moment it happens. The five-star moment when David's allergy is flagged across every venue. The three-star moment when the quote needs more clarity. The five-star moment when all eight guests confirm. Each rating triggers a contextual response in MJ's voice and context. Brand Sentiment updates live.

Nothing hidden. Nothing summarised from memory. Nothing buried. No "let me check what my colleague said." No "I'll come back to you on that."

Openness AND trust as an operating model, not a marketing claim.

What this can offer, anywhere

This is not built only for premium cruise. The operating model applies wherever the customer's choice, return, and recommendation are commercially material.

Wealth management. Healthcare concierge. Premium automotive. Fine dining and Michelin-starred kitchens. Luxury hospitality. Singapore Grand Prix corporate hospitality. Intimate high-profile band performances. Insurance. Banking. Telco. Retail. Aviation.

The framework is sector-agnostic. The capabilities are not.

The track record across my career proves it lands on both sides of the commercial ledger:

• £38M annual OPEX saved. 4,000+ FTE, 12 sites, 3 continents. Global airline group.
• 40% mortgage sales uplift. Regulatory rating rescued. Major UK bank, 3,000 FTE.
• 90%+ CSAT and 2 million customers acquired in 6 months. Greenfield digital bank, 200% over acquisition target.
• Abandonment 90%+ → under 10% in 6 months. Global services, 2,000+ FTE, 70+ sites.
• 99% rebooking intent. Premium travel operator, full-service transformation.
• 75 → 400+ FTE in 12 months at 150% revenue vs target. Greenfield BPO.

Savings AND sales. Acquisition AND retention. Turnaround AND scaling. Every engagement.

This is what differentiates the EPIC CSX operating model. The commercial models drive both sides of the ledger because they are designed to. Service becomes the place revenue is made, not the place customers are sent when something is wrong.

Ready to see it live?

This page is the summary. The walkthrough is the proof.

Step inside MJ's actual journey. Click through fourteen chapters. See the quote build itself. Watch the Brand Sentiment line move in real time. Open the Support tab and see every interaction logged. Feel what the operating model feels like, end to end.

Ten minutes. No registration. No deck.