Delphium Labs
Delphium LabsApplied AI Research . London . 2026
Get in touch
Drop us a line
Delphium Labs

Frontier technology
applied to organic behaviour.

We observe how industries actually adopt AI, then build the vertical tools that fit.

Our research
§1 . Thesis“Real adoption reveals real opportunity.”
1. Thesis

Three things we
know to be true

01

People use AI differently than predicted

Travellers describe moods to ChatGPT, not keywords. Brokers draft relationship notes, not reports. Real adoption patterns diverge sharply from how tools are built - and that gap is where the opportunity lives.

02

Every industry has its own AI behaviour

A hotelier uses AI nothing like a yacht broker. Horizontal tools - built for everyone - miss the specific workflows, relationships, and discovery patterns that define each industry's actual use.

03

The right vertical tools do not exist yet

Most AI tools are optimised for volume, not fit. We study specific industries, identify the gaps that horizontal tools structurally miss, then build what fits - from the behaviour up, not from the technology down.

Delphium Labs . Applied AI Research02§1 - Thesis
§2 . Projects“Built from the behaviour up.”
2. Projects

What we are
building

FindingFin
Hospitality
Live

Travellers ask AI for hotel recommendations. Most independent properties are invisible to those answers. FindingFin monitors how AI engines see your property and optimises your presence across ChatGPT, Gemini, Perplexity, and Claude.

6AI engines tracked
24hrAudit turnaround
3.2xAvg. visibility lift
Explore FindingFin →
Meridian
Brokerage
Research

Yacht and luxury brokers spend hours drafting relationship notes, client briefs, and market comparisons. Meridian studies how brokers actually use AI in their workflows and builds tools that match those patterns - not generic CRM automation.

40+Brokers studied
3Behaviour patterns
Q3Pilot launch
Research in progress
Protocol
Procurement
Pilot

Procurement teams are adopting AI for supplier evaluation, but the tools they reach for are built for sales teams. Protocol studies procurement-specific AI behaviour and builds workflow tools shaped to how buyers actually evaluate and compare.

12Pilot partners
5Workflows mapped
Q4Early access
Pilot in progress
Delphium Labs . Active Projects03§2 - Projects
§3 . Research“We publish what we find.”
3. Research

From the lab
Recent research

Playbook

The Independent Hotelier's AI Visibility Playbook

Mar 2026

A five-phase plan for independent hotels to become visible in AI travel recommendations. Combines technical foundations, content strategy, and ongoing optimisation into a practical roadmap.

Read →
Guide / Content

How to Write Website Content That AI Engines Actually Cite

Mar 2026

AI engines extract and cite specific types of content. This guide covers the content structures, formats, and writing patterns that increase your chances of being recommended.

Read →
Perspective

Why AI Visibility Monitoring Is Not Enough

Mar 2026

Knowing your AI visibility score is useful. Knowing what to change is what actually matters. At Delphium Labs, we think the industry is focused on the wrong problem.

Read →
Market / Report

The State of AI Visibility in UK Hospitality: 2026

Mar 2026

Where the UK hospitality industry stands on AI visibility in early 2026. Most businesses remain invisible to AI recommendations, creating a significant first-mover advantage for those who act now.

Read →
GEO / Hotels

Why Your Boutique Hotel Disappears in AI Travel Answers

Mar 2026

Most boutique hotels are invisible to AI travel recommendations. Delphium Labs analysis identifies the five technical gaps that cause this and the specific fixes that restore visibility.

Read →
Market / Comparison

ChatGPT vs Perplexity vs Gemini: Which Matters Most for Hospitality?

Mar 2026

Each AI engine recommends hotels and restaurants differently. Delphium Labs tested 300 queries across all three to map where each engine sources its answers and which matters most for your business.

Read →
Market / Analysis

Why OTAs Are Investing in AI and What It Means for Direct Bookings

Mar 2026

Booking.com, Expedia, and TripAdvisor are building AI-powered travel assistants. For independent properties relying on direct bookings, this creates a new competitive threat that traditional SEO cannot address.

Read →
Guide / Technical

Schema Markup for Hospitality: A Technical Guide

Mar 2026

How to implement Hotel, Restaurant, LocalBusiness, and Event schema markup so AI engines understand your hospitality business. Includes JSON-LD examples ready to use.

Read →
View all research →
Delphium Labs . Lab Notes04§3 - Research
§4 . Approach“Observe. Test. Build.”
4. Approach

How we work - from behaviour
to product

01
Observe
Study real behaviour

We embed with businesses in specific industries and watch how people are actually using AI - not what they say they do, not predictions, not surveys. Real workflows, real queries, real friction.

02
Test
Validate the hypothesis

Behavioural patterns become hypotheses. We run controlled interventions with real businesses and measure what actually changes. Ideas that don't survive testing don't become products.

03
Build
Ship what fits

Validated patterns become vertical tools - built from the behaviour up, not the technology down. Shaped to how people in each industry actually use AI, not how we think they should.

§4 - Approach05Delphium Labs . 2026
§5 . Work with us
§ 5.0 - Collaboration & CorrespondenceDelphium Labs . London, UK

Built with industry,
not for it.

We partner with independent operators, domain experts, and research teams who want to understand how AI is actually changing their industry.

For businesses
Explore our projects

Hotels, brokerages, clinics. If you're living the problem our research covers, we'd like to work with you directly.

For researchers
Our research

We publish findings from our behavioural studies and share our methodology openly. Collaboration with domain researchers is welcome.

Read our research →
For investors & partners
Work with us

We're building vertical AI infrastructure for industries being left behind by horizontal tools. If you're interested in the thesis, let's talk.

Research Updates . New Projects
Follow the research
© 2026 Delphium Labs06hello@delphiumlabs.com