RESEARCH // WHILTER LABS
The lab notebook.
Playbooks, benchmarks, and architecture for AI-native marketing — written from the engine, not about it.
What does Whilter research?
Builder-facing research on AI-native marketing — the operating model, keeping generative models on-brand, and why the loop compounds. Our POV essays sit alongside.
How Marketing Teams Prove ROI From AI-Generated Creative
Marketing teams prove ROI from AI-generated creative with head-to-head tests against existing creative, matched spend, and the right metric at each layer.
roimeasurementai-generated-creativeperformance-marketingincrementality R-02One Guest Conversation Across WhatsApp, Web Chat and Voice
How hospitality brands keep one guest conversation across WhatsApp, Instagram, web chat and voice, with AI and human agents sharing one context.
hospitalitycustomer-engagementomnichannelwhatsapphuman-handoffkonne-cx R-03How Brands Scale Personalized Video With First-Party Data
How brands scale personalized video with first-party customer data: one template, a customer table, and delivery on WhatsApp, email, apps and RCS.
personalized-videofirst-party-datapersonalizationsnypeomnichannel R-04How Fashion Retailers Offer Virtual Try-On on WhatsApp
How fashion retailers run virtual try-on inside WhatsApp, the flow step by step, and what W for Woman got: a 4× engagement uplift and 65% WhatsApp engagement.
virtual-try-onwhatsappfashion-retailconversational-commercesnype R-05Personalized, Compliant Marketing at Scale
How brands run personalization across channels and languages without losing brand control or regulatory compliance — the operating model, the failure points, and what it looks like in production.
personalizationcomplianceoperating-modelmultilingual R-07Strategic Thinking in Vision Models
What multimodal models can reason about in creative and competitor analysis — and where they still need a human.
vision-modelsmultimodalcreative-analysis R-08Monolith vs Agentic Systems
Why marketing software is moving from one monolithic platform to a system of specialised agents — and what changes for the buyer.
architectureagentssystems R-09How to Make an LLM Focus
Generative models drift without constraints. The context-engineering and brand-DNA techniques that keep output on-task and on-brand.
llmcontext-engineeringbrand-dna R-10Benchmarking Brand Compliance
How to measure whether AI output is actually on-brand — a benchmark and a method, not a vibe.
brand-complianceevaluation R-11The AI-Native Marketing Playbook
What marketing looks like when AI is the substrate, not a bolt-on — the operating model, not the tool list.
ai-nativeoperating-modelplaybook