Skip to content
Valtair
Selected work

Systems you can use, and the kind of work we build

Two demonstration builds are live and open to anyone: real working systems, built around fictional businesses so we can show the engineering without exposing a client. Below them, representative examples of client work. Named, detailed case studies are published as engagements complete and with client permission.

Live demos

Working systems, open to use right now

Each one is a complete build rather than a recorded walkthrough. The businesses are fictional; the systems are not.

The Aurelia Properties demo homepage: a Dubai property search, with the Ask Ayla assistant in the corner.
Live demoReal estate

Aurelia Properties

Conversational property advisor

A demo site for a fictional Dubai estate agency, with an assistant, Ayla, that answers questions against the live listing set and qualifies an enquiry before a human picks it up.

Channels

Web chat, embedded in the site and in each listing

Open the demoHow it is built
The Cedarline Home Services demo homepage: "One call. Every trade.", with the next available arrival window and a chat button.
Live demoHome services

Cedarline Home Services

Multi-channel AI agent suite

A fictional seven-trade contractor with agents on web chat, WhatsApp, voice, and email, all answering from the same coverage, scheduling, and pricing rules the website publishes.

Channels

Web chat, WhatsApp, voice, email

Open the demoHow it is built
Case studies

Examples of client work

What gets built, the problem it solves, and the kind of outcome it drives. Illustrative until named engagements are published.

Permissions at retrievalDocumentation, support tickets and product data are ingested into hybrid retrieval. Permissions are enforced before content reaches the model; grounded answers with citation checking are embedded in the existing product.DocumentationSupport ticketsProduct dataIngestion & hybrid retrievalPermissions before modelExisting productGrounded answerCitation check
Revenue operationsIllustrativeConfidential

A grounded knowledge assistant for a revenue platform

Support and sales teams could not find answers across scattered docs, tickets, and product data.

Result

Cited answers in place of searching across scattered sources.

Distinctive. Graph and vector retrieval with strict permissions and inline citations for every claim.

View the case study
Connected to the scheduling stackAn inbound call reaches the voice agent. Availability and account context come from the scheduling and CRM stack at call time, bookings connect back to scheduling, and calls outside routine scheduling go to staff with context.Scheduling systemAvailabilityCRMAccount contextInbound callVoice agentBookingStaff with context
OperationsIllustrativeConfidential

An inbound voice agent for high-volume scheduling

Missed and abandoned calls were losing bookings during peak hours.

Result

Routine inbound scheduling calls handled by the agent, with clean handoff to staff.

Distinctive. Low-latency speech orchestration wired directly into the existing scheduling and CRM stack.

View the case study
Evaluation before featuresA prototype becomes a production system in stages. An evaluation harness is built first, cost controls are included from the start, and monitoring and guardrails are added for a production-ready product.Evaluation harness firstMonitoringGuardrailsPrototypeStaged releasesProduction-readyCost controls from the start
Vertical AIIllustrative

From MVP to a production-ready product

A promising prototype had no evaluation, no monitoring, and no path to reliable scale.

Result

Shipped a monitored, evaluated production system ready for the first paying cohort.

Distinctive. Evaluation harness and cost controls built in from the first architecture decision.

View the case study

Client ideas stay confidential

We publish case studies only with permission, and anonymise where required. Your product brief and data are treated as confidential.

Discuss a project

Book a discovery call