Questions and answers

What is published today and what we scope with you.

How do you keep our running costs down?

We match model capability to task difficulty rather than routing simple work through expensive models, and we challenge any subscription that can't show measurable value.

Also asked as: How do you stop AI bills spiralling?

Is the introductory fit call free?

The introductory 20-minute fit call is free.

Also asked as: Does the first call cost anything?

How long is the fit call?

The fit call lasts 20 minutes and is free.

Also asked as: How much of my time will this take?

What does missed-call recovery cost?

Missed-call recovery is £1,200 fixed at the current founding rate. The free 20-minute fit call checks whether it suits your setup before anything starts.

Also asked as: How much for the SMS text-back thing?

What do the new sales, recovery and evidence products cost?

Each engagement is scoped individually, so there is no price list for these products yet. The fit call is free, and if work can be scoped you get a written price before anything starts.

Is there a monthly retainer after the first project?

The published appointment-business catalogue has optional monitoring from £120 a month. Monitoring for the new sales, recovery and evidence products is scoped individually.

Also asked as: Do you do ongoing support, or is it one-off?

Do your prices include VAT?

No — every figure on this page excludes VAT. Where it applies, it's added on the invoice, never hidden inside the headline price.

How much is the AI Opportunity and Vendor Review?

£750, fixed, whatever we end up recommending. The free fit call is the fastest way to get one booked.

Also asked as: How much is the front-desk vendor review?

What will we actually be paying for each month?

Whatever recurring components your build genuinely needs (hosting, any vendor fees, and support), each itemised, with what it does and what happens if it's removed.

Are you the cheapest option?

Our pricing is designed around the outcome you need at the lowest sustainable total cost, counting support and switching as well as the headline price.

Also asked as: Why are you more expensive than a freelancer?

Who holds the accounts, and what happens if you stop working with us.

Who owns the AI subscriptions and API keys, us or you?

You do, throughout: every account, subscription and key is created in your name, never ours.

Whose hosting account does this run in?

Yours, wherever practical: you pay the infrastructure provider directly rather than through us, so there's no hidden mark-up and no access that depends on paying us.

Also asked as: Do you resell hosting?

What happens to our system if we stop working with you?

It keeps running: where the architecture allows it, the infrastructure and data sit in your accounts, so leaving costs you our support, not your system.

Also asked as: Are we locked into your agency? · Can we take this elsewhere later?

Do you take commission from vendors you recommend?

If there is any, it's disclosed before the recommendation is made, not after. We also compare at least two options every time.

Are we locked into one AI vendor?

No: every vendor is treated as swappable by design, models included. Nothing is architected around one provider.

How do we know a recommendation isn't just what earns you the most?

The test we apply is whether we'd recommend the same component if we earned nothing from it, and any referral income is disclosed to you before the recommendation, not after.

Where it lives and who touches it.

Do you have a Data Processing Agreement we can sign?

In preparation, pending legal review. If it's a procurement requirement, that's worth flagging on a call early.

Is our data encrypted?

In transit, yes, that's confirmed. At rest it's planned but not yet a standing policy across every hosting choice, so it's worth pinning down for your specific setup.

Does our data get sent to external AI companies?

Only where an external provider is genuinely the right component, and you'd know which one and why before it's used. Where a private deployment fits better, that's the design.

Where is our data hosted?

Client operational workflows use EU/UK regions by default, subject to the chosen vendor; website enquiries currently use Cloudflare's globally replicated Workers KV. UK-only requirements must be checked for the specific workflow before contract.

Also asked as: Does our data stay in the UK?

How long do you keep our data?

The retention control is in progress: the current website-enquiry KV record has no automatic expiry, and workflow periods are set per engagement. We will not claim a fixed period until it is implemented.

Can the AI run on our own infrastructure instead of a third party?

Often yes: a suitable open model can be deployed into hardware or cloud you control, where the privacy, economics and workload justify it.

Also asked as: Can we self-host the model?

Will an AI provider train on our data?

That depends on the specific provider's current contract terms, which we check per vendor rather than generalise. It's a fair question to put to us about any provider we propose.

The current status of each item.

Will our customers know they're talking to a bot?

Yes: every automated voice or chat interaction discloses that it's automated. That's a hard-coded rule, not a setting.

Are you a bot?

Yes. The website chat is an automated assistant, not a person; it answers routine questions, can book a call and hands anything requiring judgement to a person.

Also asked as: Am I talking to a real person? · Is this a human?

Will this make us GDPR compliant?

No supplier can hand you compliance, and we won't claim to. What we can do is design the data flow, retention and access controls so your own compliance work has less to fight.

Also asked as: Are you compliant?

Do you carry cyber insurance?

Not yet, that's planned, not currently in place.

Does anything happen fully automatically, with no human involved?

No client-facing action runs fully unattended: every workflow has a defined escalation and review point, and special-category data never gets a solely automated decision.

Are you ICO registered?

Not yet, deliberately deferred until the company itself is registered.

Do you have professional indemnity insurance?

Not yet: professional indemnity insurance is planned, not currently in place.

Also asked as: Are you insured?

Fallback, monitoring and the handover plan.

What happens after launch? Are we on our own?

No, support is a core phase rather than an upsell, because models, APIs and integrations keep changing after go-live.

Also asked as: Who do we call when it breaks?

If the automation fails tomorrow morning, how do we keep working?

That answer has to exist before launch: critical workflows keep a manual recovery path, and enquiries are preserved rather than lost while the automation is down.

How would you know if something broke?

Monitoring is part of what's delivered, not an extra: failed jobs, model errors, fallback activation and cost anomalies are recorded rather than discovered by you.

What happens if the AI provider goes down?

Critical workflows are built with fallback paths, so a provider outage moves the work to an approved alternative rather than stopping it.

Also asked as: What if the model is unavailable?

What uptime do you guarantee?

We design for graceful degradation, so a failure moves work to the agreed fallback route rather than stopping your business.

Vendor choice, careful scoping and the right use of automation.

Will our staff have to understand all this?

No: the complexity stays in the engineering layer. What your team sees should be a clear business process, not the routing and fallback underneath it.

What if our current software already does this?

Then we'd say so: the review deliberately checks what you already pay for and haven't switched on, and "keep what you have" is a legitimate outcome.

What decisions does the AI make on its own?

In approved B2B sales and first-party recovery campaigns, rules and AI can check, match, summarise, send the agreed messages, run routine follow-ups, classify replies and route work. Research and evidence workflows start read-only or draft-only; audience and channel policy, pricing, negotiation, technical, safety, clinical, legal, regulatory and professional decisions stay with a named person.

Do you use open-source models, and does it matter where they're from?

Open models are part of the toolkit, and where one was developed isn't by itself a reason to pick or reject it: licence, security, data flow and task performance decide.

Also asked as: Would you use a Chinese model?

How does a public market signal become a B2B lead?

The agent detects the defined event, verifies the organisation and ICP fit, selects the approved business contact route and starts a tailored conversation. It then follows the agreed sequence, classifies the reply and hands genuine interest or a complex question to the named salesperson.

Would you ever tell us not to automate something?

Yes, and we'd rather say it early: sometimes the honest answer is fix the process first, keep the platform you have, or that a step needs a person.

Also asked as: Do you ever turn work down?

Which AI do you use?

Whichever suits the task: the architecture is built so models are replaceable components, and material choices are agreed with you rather than assumed.

Also asked as: Do you use ChatGPT?

What does the AI Opportunity and Vendor Review involve?

One workflow reviewed, your current platform checked for features you already pay for, alternatives benchmarked, ending in a written recommendation, not a pitch.

Why not just use ChatGPT or a boxed AI agent product ourselves?

Those are legitimate tools and may end up inside what we build. The difference is designing a system whose value survives a change of model, price or product: that's the part that's ours.

Also asked as: What do you do that we couldn't do in-house?

Why not just hire a receptionist instead?

Often you should, and we'd say so. This covers the moments when nobody is free to pick up rather than replacing the person who is, and every workflow still hands to a human at a set point.

Also asked as: Wouldn't a person be better than a bot? · Why not just employ someone?

Can you find and nurture B2B leads automatically?

Yes: we can configure an agent to monitor approved public B2B sources, verify the company and ICP fit, use the agreed business email or messaging route, run the approved follow-up sequence, classify replies and hand qualified interest to sales. Consumer campaigns are outside this offer.

Do you work with businesses like mine? I run a busy dental practice.

Yes: dental is a named segment, and clinical questions are always kept outside automated routines.

We sell software or services to fleet operators. Can you help with lead generation?

Yes, if an operator-licence change is relevant to a precise ICP: the agent verifies the company, selects the approved business contact route, prepares or sends the tailored opening, follows the agreed nurture sequence and hands a positive or complex reply to your salesperson.

Do you work with businesses like mine?

Our named offers serve fleet suppliers, trade distributors and hire firms, inspection and maintenance providers, transport-compliance consultancies and appointment-based businesses. Outside those, fit depends on a repeatable workflow, usable source data and a named owner.

We're a solicitors' firm, a gym, something else. Is this relevant?

Possibly, but we would need one repeated case, authorised source data, a clear output and a named human owner. A broad request to automate the business is not enough to establish fit.

Can you find service renewals or remedial quotes in our maintenance system?

Potentially: we first check the existing field-service features, then surface only the residual due-service or unresolved-quote cases that need a named owner's review.

Can you help us follow up old trade quotes or lapsed accounts?

Yes: we first audit the data, select one useful group of records and configure a specific follow-up sequence. The agent can then run the agreed messages and hand positive or complex replies to the account owner.

Can the system prepare our monthly operator compliance reviews?

It can prepare a source-linked evidence and exception pack for the named Transport Manager; it cannot decide compliance, roadworthiness, corrective action or sign off the review.

Would this suit a veterinary clinic?

Yes, within Front Desk & Enquiry Handling, with clinical and urgent questions always kept outside the automated routine. Want the free 20-minute call to see what is worth checking?

Can the assistant book appointments into our calendar?

Depends on the project: the messaging assistant hands off to your booking link, the full front-desk build writes directly where the API allows. Worth 20 minutes to work out which you need.

What happens when you cannot answer a question?

The conversation goes straight to a person rather than receiving a guessed answer. The same principle is built into every client assistant.

Which messaging channels do you support?

Up to six on the full build: phone, web chat, WhatsApp, SMS, Telegram, Instagram. Which ones actually matter for you is a good first question for the free call.

Also asked as: Do you do WhatsApp? · Instagram DMs?

Is this basically just ChatGPT with a different skin?

No: it answers only from a knowledge base you approve, escalates anything uncertain by design, and you hold your own API key.

Are you a UK company? Where are you based?

Durward is not yet a registered company; registration is deliberately still ahead of us, and the details will be published here as soon as they exist. What protects clients meanwhile is a written scope, a contract, and accounts that stay in the client's own name; the client protection page sets it out in full.

What do you need from me to get started?

Nothing is required in advance. The first step is a free 20-minute call to understand the current setup.

Also asked as: What do I have to prepare? · How do we begin?

Do we have to buy the whole thing at once?

No: work is presented as defined modules with their own boundaries, so you can see what's required, what's optional and what can wait.

What if we want to change scope halfway through?

Every project has a written scope; the £700–£900 day rate applies only to the published appointment-business catalogue. New-portfolio changes are priced in the scoped proposal.

Also asked as: What's your day rate?

How long does a project take, start to live?

Delivery time depends on the existing setup and the workflow being changed. A fit call is needed before a responsible timeframe can be given.