AI Models for Marketing
Multi-model architecture in production. We compare Claude, ChatGPT, Gemini, and others — and pick the right one for each marketing job. Most public model-comparison content is written by people who have never run these systems in production. We have. Every comparison on this hub is grounded in tasks we run live for clients on monthly sprints from £3,000 to £8,000 a month, no minimum — content production, ad copy at scale, attribution analysis, sales enablement, customer support routing, internal research. Different jobs, different best-fit models.
Model comparisons
Claude vs ChatGPT vs Gemini vs the open-source field — head-to-head on the marketing jobs we actually run. Output quality, latency, cost, failure modes.
Sub-clusterModel economics
What it costs to run AI marketing at production volume. Token economics, batch pricing, caching, fine-tuning thresholds, when smaller open models beat frontier API calls.
Sub-clusterUse-case selection
Job-to-model mapping. Long-form drafting, structured reasoning, retrieval, classification, ad copy variants, customer support routing — which model wins which job, and why.
Sub-clusterMulti-model architecture
How we orchestrate multiple models in one production stack. Routing layers, fallbacks, eval harnesses, graceful degradation — the operating layer that keeps the marketing function running when one model breaks.
Sub-clusterWe do not commit to one vendor. We run a multi-model stack and switch when the economics or output quality change. Sprint pricing covers the model bill — you do not get a separate API invoice.
— AI Models thesisAI Models
Claude Max gives you unlimited Opus at a flat rate. The API gives you granular control and programmatic access. Here is the decision framework for marketing teams choosing between them.
Read post:AI Models
We ran both for six months across content, operations, and client work. Claude won on every metric that matters for production marketing systems. Here is the full breakdown.
Read post:AI Models
Claude Max for marketing teams — when the subscription pays for itself and when it doesn't
When does Claude Max pay for itself for marketing teams? Token economics, break-even math, and the decision framework — from real practitioner data.
Read post: Claude Max for marketing teams — when the subscription pays for itself and when it doesn'tAI Models
Marketing MCP servers — connecting Claude to GA4, HubSpot, and ad platforms
Marketing MCP servers are small programs that connect Claude to the data systems where your marketing actually lives (Google Analytics 4, HubSpot, Meta Ads, and so on) using a shared protocol that Anthropic introduced in late 2024 . Once th...
Read post: Marketing MCP servers — connecting Claude to GA4, HubSpot, and ad platformsAI Models
Building brand-aligned Claude Skills so Claude stops defaulting to generic marketing templates
Most marketing-focused Claude Skills ship as dressed-up prompt templates and drift back to generic by paragraph five. The architectural fix: a Skill skeleton that reads brand references at runtime, validates output against your taxonomy, and enforces hard rules before returning. From running 139 custom Skills across 30 plugins for three brands.
Read post: Building brand-aligned Claude Skills so Claude stops defaulting to generic marketing templatesMarketing Operations
Why most AI marketing automation projects stall at month 3
Read Why most AI marketing automation projects stall at month 3Marketing Operations
The AI marketing operations playbook that's built for reality
Read The AI marketing operations playbook that's built for realityContent Strategy
ReadAttribution
ReadMarketing Operations
AI marketing automation and the central hub for unified system thinking.
Explore Marketing OperationsContent Strategy
AI content production at scale, with quality and brand voice as constraints.
Explore Content StrategyAttribution
Server-side tracking, multi-touch attribution, and the models that survive iOS and GA4 changes.
Explore AttributionAI Search
Showing up where buyers ask AI — Google AI Overviews, ChatGPT, Perplexity, Claude — plus the AI-era SEO that still feeds them.
Explore AI SearchBook a discovery call
Sixty minutes. No slide deck. We tell you which sprint tier fits and when the multi-model stack starts running.
Get in touch →