AI and people shouldn’t feel like two services. With us, they’re one.
Hi, my delivery was supposed to arrive on Friday and it hasn’t turned up. The tracking still says “in transit.”
Hi Sarah, I’m Rosie, Willow & Pack’s AI assistant. Sorry your box is late — I know how much you’re relying on it. Let me check order #48291 now.
Can you resend it? The girls are down to their last two meals.
One thing before I do — your notes show Poppy reacted to the new lamb recipe last month. Health needs a person, so I’m bringing in Ellie from our care team to check the box before we resend it.
Hi Sarah, Ellie here — I’ve got the full thread. I’ll swap the lamb for chicken, send it overnight at no charge, and monitor it personally. Sound OK?
Rosie Hi Sarah — did the replacement arrive, and how is Poppy?
Sarah It arrived this morning. Poppy’s much better on chicken.
Rosie Glad to hear it — I’ve flagged chicken for future boxes.
Preference updated · ticket closed · AI continued the queueRosie says she’s AI. Ellie gives her name. Both trained in our service method: trust, respect and ownership, in one conversation.
Example based on real life customer conversations with us.
One workflow, one standard and one price — however the work moves between people and AI.
A client asks: "I want the AI to start handling cancellation requests from customers." Same request, two approaches.
What we deliver beyond a standard BPO
What a typical client can expect
Transform your service economics, without disrupting quality or operations.
We optimise your processes and run AI and people as one service, trained in the same method, so every conversation stays personal.
Quality people, trained in our service method: trust, mutual respect and ownership. Personal service, not scripts.
Explore specialist services that solve a specific problem, or build your own. Trained on our methodology, with cloud and local models.
Assess how ready your operation is to adopt AI, then put the right programme in place to support your people.
What service shape suits your needs?
Choose one of our pre-built AI services, or we build one custom. Runs standalone or connected to your stack. Our team behind it for escalation, knowledge, and continuous improvement.
Talk to us about this →How you pay
We can easily price by time, conversation, customer, fixed fee, outcome, or a blend.
We operate on a standard master services agreement, with bespoke work agreements that sit underneath. See the Master Services Agreement →
Augmenting an in-house team on customer care for a dog food subscription brand.
Our team runs alongside theirs, working to the same process and standard. Same tickets, same tone — lower cost, no rebuild.
Mode mix People · Human ↔ Human
Multilingual concierge customer care across EMEA, US and LATAM.
In-market, in-language, on-brand. A dedicated team that reads Zegna as if it were their own house, backed by process that scales across three continents.
Mode mix People · Human ↔ Human · Human ↔ Brand AI
AVA, our AI out-of-hours receptionist, curated for the clinic.
AVA answers the calls the clinic used to miss — books, triages and hands off to a person when it matters. Revenue that used to sit in the answerphone now pays for the service.
Mode mix AI service · Human ↔ Brand AI
Responsible AI, an award-winning team, a flexible platform and a fair commercial partnership.
Vulnerability stops the automation, by default. Every AI decision traces back to a known source.
Proprietary30 years leading service, including inaugural CX leadership at Spotify and ASOS. Award-winning, globally distributed, and 95% of the team stay — so the knowledge compounds instead of walking out.
We connect to your systems. Your choice of data location and cloud, with no lock-in on model, data or process. Compatible with personal AI agents. Cyber Essentials certified and NHS DSPT Standards Met.
Clear contracts with no nasty surprises. Flexible terms, and notice periods that work both ways.
Built on the Service Handshake — a managed-service standard we authored and published (CC BY 4.0).
Live managed services, across AI and people.
Clients today, and the brands our team led service for before Neos Wave.
Most of our work comes from recommendation.
Your customers are already sending AI to your contact centre, and most operations can't tell when it's happening. We ran the field test that proved it — a consumer AI we built passed identity checks and negotiated resolutions inside live contact centres of multi-national brands. No agent realised.
This is Dual CX. It's the reason we built Neos Wave to run all four modes on one workflow.
Customer service used to be a two-body problem — a customer and a brand. It isn't anymore. This is Dual CX.
Your biggest line, still priced in seats and minutes — and every point of margin sits here.
A person calls, a person answers. The traditional model.
A customer talks to your AI. Voice, chat, deflection metrics.
Where the exposure is arriving. Unpriced, unmanaged, and already inside live contact centres.
The customer's own AI agent contacts you. It looks like normal traffic. It isn't.
Two agents negotiate an outcome. Fast, cheap, and almost entirely ungoverned.
An operation that can't tell who it's talking to commits without authority, discloses to unverified parties, misses vulnerability, fails machine-speed verification, and can't defend the complaint that follows.
None of it looks like a failure at the time. It looks like normal handling — until somebody asks.
Modes 1 & 2 are where cost comes out. Modes 3 & 4 are where CX gets designed next. The market has split into two camps. Neither is built for both.
Two roads. Neither goes where service is going. Talk to the team building the third.
Start a conversationWe published one of the first frameworks to help companies handle AI-to-AI service. Now we're evolving it into a protocol-neutral standard for anyone to adopt.
The eight questions buyers ask us most — answered plainly. If yours isn't here, talk to AI Maria below.
Those platforms sell you software. You still need to configure it, connect it to your data, write and maintain the knowledge, handle escalations, monitor quality, and staff a team for everything the AI can't do. That's a real programme of work.
Neos Wave sells you the outcome. We bring the AI, the people, the knowledge base, the escalation team, the QA, and the reporting — as one managed service, in the tools you already use. You don't buy a platform and then hire a team to run it. You buy a service. We run it.
If you already own one of those platforms and want to keep it, we work with it. Our AI runs alongside or on top of your existing stack, not instead of it.
Traditional BPOs bill by the hour or per handled contact. Their unit economics only work when volume grows and headcount grows with it. Adding AI to a BPO means bolting a tool onto a labour model — the incentives don't line up.
We're built AI-first with people where they matter most: judgement, empathy, complex cases, edge cases, and continuous improvement. Our AI handles the volume; our team handles the exceptions, teaches the AI, and owns the customer relationship. You get one service, one standard, one price — not a labour arbitrage play.
The result: BPO-style accountability for the whole service, without the BPO-style cost curve.
Because we're one service made of AI and people, we don't price by seat or per ticket. We price by what actually matters to you: time (dedicated capacity), conversation (per handled interaction), outcome (per resolved case or booked appointment), project (fixed scope), or a blend.
For context on where we sit in the market: standalone AI platforms typically charge $0.99 to $2.00 per resolved ticket, with per-seat licences on top. Traditional BPOs typically charge $4 to $6 per resolved ticket, or an hourly rate per agent. We price competitively against either approach — and unlike either, you're not paying for a platform and a team separately.
All engagements sit under a standard Master Services Agreement with bespoke work agreements underneath. Talk to us and we'll build a shape that works.
No. Our AI is grounded in your knowledge: your policies, your product data, your past tickets, your tone of voice, your escalation rules, your compliance requirements. Nothing about your customer is answered from a language model's general training data.
We build and maintain a governed knowledge base for you. Our team owns it — keeping it accurate, adding new content as your product changes, retiring outdated answers. When the AI responds, it's citing that knowledge base, not making things up.
We use best-in-class language models from OpenAI, Anthropic, Google and open-weight providers as the reasoning layer — whichever one is right for the task. But the answers themselves come from your knowledge, retrieved and cited. That's the standard we wrote into the Service Handshake.
For a proof of value on a single, well-defined use case: 3 to 5 days. That's a working AI agent connected to your knowledge, handling real conversations in a sandbox, with our team behind it.
For a full production service: typically 4 to 8 weeks, depending on how many channels, how many use cases, integrations required, and how mature your knowledge is when we start. We do the knowledge lift for you — you don't need a documentation project to begin.
We'd rather start narrow and expand than boil the ocean. Most engagements begin with one channel and one use case, then grow.
It hands off to our people, with full context. The customer doesn't restart the conversation, doesn't repeat themselves, doesn't get shunted into a different queue. To them, it's one service.
Our team — not yours — handles the escalation, unless you specifically want your team in the loop. We agree escalation rules with you up front: what the AI does, what it doesn't do, what always goes to a human, what needs a manager. Those rules are written into the service, monitored, and improved over time.
Behind the scenes, every escalation teaches the AI. Cases that used to need a human become cases the AI handles next time — with our team reviewing and approving the improvement, not just letting the model learn unsupervised.
We operate to enterprise standards. We hold Cyber Essentials certification, have met the NHS Data Security and Protection Toolkit standards (Standards Met), and are UK GDPR compliant. Customer data stays in-region, is encrypted at rest and in transit, and is never used to train third-party models.
On AI safety: we don't let the AI answer questions outside its knowledge scope, we monitor for hallucinations and unsafe responses in real time, and every AI response is auditable back to the source it cited. Our Service Handshake standard defines the guardrails.
For regulated sectors (financial services, healthcare, legal), we can operate to sector-specific compliance requirements — tell us your regulator and we'll walk you through how we'd align.
Email, live chat, WhatsApp, SMS, voice, phone, in-app messaging, and social DMs. We work inside your existing helpdesk (including Zendesk and other major platforms on request) — the customer never sees a Neos Wave-branded surface unless you want them to.
Voice specifically: we design and run AI voice agents for inbound and outbound, in the same managed model. If you want to hear one, talk to AI Maria in the section below — that's the same technology.
Talk to AI Maria — she can talk about Neos Wave, the Service Handshake, Dual CX, and how we work. Or leave a message.
Ask about Neos Wave, the Service Handshake, Dual CX, or how we work. Voice, in the browser — no dial-in.
Prefer email? hello@neoswave.com — or leave the details below.
