What we build and operate: five products, one standard

ReplyOS is revenue infrastructure for UK dental and aesthetics practices, built dental-first and extending to aesthetics, optometry and physiotherapy. It captures enquiries across SMS, WhatsApp, Instagram, Facebook, email and web forms; responds with AI assistance under operator-grade controls with a human handover; runs booking flows inside the conversation; and shows the chain from enquiry to revenue in one place.

It comes in three plan tiers, Core, Pro and Scale, from essential channels to advanced reporting and governance. ReplyOS is a registered trade mark and is live at replyos.co.uk. It is also an ALPHA startup at Web Summit Lisbon, 9 to 12 November 2026.

ReplyBench scores a business on six dimensions: enquiry speed and first response, clarity of calls to action, follow-up readiness, competitor positioning, trust signals and revenue tracking. It turns them into one ReplyBench Score with an indicative estimate of revenue at risk. It is not an SEO audit: it measures how you handle the enquiries you already get.

The Free Snapshot costs nothing. The human-reviewed Revenue Leakage Audit is £495, at a launch price of £249, and the fee is credited in full against ReplyOS setup.

When patients ask an AI assistant for the best dentist near them, someone gets named. ReplyScore crawls a practice's site, puts real patient questions to AI engines, and scores its visibility against audited UK peer practices, with prioritised fixes ready to paste in. A teaser score is free; the full report is £49 at launch pricing.

Slipmark makes AI agents prove what they checked by binding each verdict to a hash of the files the agent read and refusing to ship if they change. It is free and open source under Apache-2.0, released on 17 October 2026.

Weft is a runtime where a model assembles each screen from a closed catalogue of components you define, with every screen checked before it is shown and recorded so it can be reproduced. It is in early access and not yet generally available.

The same rules run through every product and every client project: record what the AI did, keep a person in charge where a mistake would matter, and test the failures on purpose. None of the products is required to hire us for a service.