AI in the Ad Tech Ecosystem

Azerion — Intelligence AI Platform

Company: Azerion (Euronext: AZRN) | Product: Azerion Intelligence / AI Marketplace | Website: https://www.azerion.com

Azerion promotes "AI for Everyone," offering pay-per-use, enterprise-grade AI compute for advertisers, agencies, and publishers. It democratises access to heavy machine learning infrastructure.

Core Capabilities

• Multi-Cloud Infrastructure: An open, independent AI architecture offering transparent pay-per-use pricing with automatic scaling so users require zero upfront infrastructure investment.
• Generative Automation: Native capabilities spanning text, image, audio, and video synthesis localized across multiple languages.

Key Features

• AI Marketplace: A centralized hub of AI applications and agents where third-party developers can contribute serverless endpoints for campaign creation.
• Smart Bidding: Advanced algorithms that predict clearing prices and demand fluctuations to dynamically improve yield in real-time open market auctions.
• Smart Content: Combines a massive proprietary video library with ad tech to generate highly relevant, brand-safe video clips on the fly.

Key Results

• Total Democratisation: Successfully allows mid-market agencies and independent publishers to deploy the same caliber of Generative AI and ML models as walled gardens.
• Extreme Workflow Efficiency: Capable of deploying complex, multi-language, multi-format generative campaigns (e.g., auto-translating video, audio, and display banners) in minutes rather than weeks.
• Zero Capital Expenditure: Pre-negotiated multi-cloud rates and serverless architecture eliminate all infrastructure overhead for platform users.

"The internet is no longer dominated by humans. As humans, bots and AI agents increasingly act on behalf of legitimate consumers and fraudsters alike, initial and persistent trust must be earned, enforced and continuously interrogated."

Cape.io — Intelligent Campaign Automation

Company: Cape.io (formerly Peach) | Product: Intelligent Campaign Automation Platform | Website: cape.io

Cape.io unifies creative production, regulatory compliance, technical QA, and cross-channel distribution into a single workflow. It closes the industry's creative execution gap using expert-trained models.

Core Capabilities

• Expert-Trained Models: Distinct from generic LLMs, Cape.io's validation models are trained by industry specialists (ex-Clearcast, Ofcom, ARPP) to interpret highly nuanced regulatory guidelines.
• Global Spec Database: Maintains thousands of living technical specifications for every broadcaster, streamer, and social platform globally.
• Streamline Multi-Market Campaigns: Cape.io enables users to design a single master creative and automatically scale it across hundreds of formats and channels while allowing local teams to customise content.
• Compliance at Speed: Automation tools significantly accelerate speed to market with 3x faster launches, while technical compliance and reducing production rework by 70%.

Key Features

• Technical QA: Auto-detects asset types and checks bitrates, safe zones, and file formats across linear TV, CTV, DSPs, social, and DOOH instantly.
• AI Compliance Check: Reviews creatives against market and publisher guidelines / codes, including complex rules (e.g., LHF, HFSS for the UK market).
• Creative Analysis & AI Studio: Leverages synthetic audiences to predict visual attention, while generative tools allow fast adjustments to CTAs, images, and localized voice-overs.

Key Results

• 5x Faster QA: Technical quality assurance processing is executed 5 times faster than traditional manual workflows.
• Global Regulatory Precision: Reduced risk of creative compliance issues and costly rework across major global markets and publishers  simultaneously.
• Zero Platform Rejections: Ensures creative assets meet the completely divergent specifications of Linear TV, CTV, and social media from a single upload point, virtually eliminating failed uploads and missed SLAs due to creative .

"Without expert training, you're dependent on whatever you get out of a large language model. Our competitive advantage is leveraging 30 years of direct regulatory expertise to close the creative execution gap."

 Joe Hollywood, Head of Product

Microsoft Advertising — AI & Agentic Commerce

Company: Microsoft | Product: Microsoft Advertising Platform | Website: https://ads.microsoft.com/

Microsoft Advertising is leveraging the full breadth of its AI infrastructure to reshape how advertising is created, managed, and transacted. Built on a consumer ecosystem of 100M+ Copilot users, it introduces a seamless agentic commerce layer.

Core Capabilities

• Generative Foundation: Utilises Microsoft's deep integration with OpenAI to bring enterprise-grade LLMs directly into the advertising workflow.
• Agentic Commerce Layer: Removes the traditional redirect from AI conversation to external website, closing the loop entirely within the chat interface.

Key Features

• Copilot in Advertising Platform: Natural-language campaign management for fast diagnostics, real-time troubleshooting, comparative reporting, and optimisation recommendations.
• Ads Studio: A generative AI creative suite built for advertising, featuring background generation, image-to-video animation, and API access for massive scale customization.
• Brand Agents & Copilot Checkout: Custom AI shopping assistants deployed on merchant websites in hours, processing payments directly within Copilot.

Key Results

• +194% Purchase Intent: Copilot-engaged journeys deliver an extraordinary 194% uplift in purchase intent.
• +16% Conversion Rates: Brands utilizing Microsoft's AI tools see significantly higher conversion rates, with some pilot partners reporting up to 3x higher conversions on agent-assisted sessions.
• -33% Journey Time: The agentic commerce flow shortens the total consumer discovery-to-purchase journey by 33%.

"Consumer behaviour and expectations are changing rapidly, and that requires new engagement approaches. At Microsoft, we're focused on being the partner for the AI era who can deliver performance-driven advertising, commerce and monetisation solutions that protect brand value, earn preference and drive results." 

Tracey Pilon, EMEA AI Acceleration Director

OneTag - The Value Exchange

Company: OneTag | Product: The Value Exchange (Media Quality + Creative Intelligence + AI Decisioning) | Website: https://www.onetag.com/

OneTag operates before and beyond the bidstream, focusing on how media value is assessed and improved. It combines media quality controls, creative intelligence and AI-driven decisioning across more than 170 billion daily auctions and 50,000 placements, with the aim of supporting measurable business outcomes rather than impression volume alone.

Core Capabilities

• Supply-Side Decisioning: Applies AI to supply before impressions enter the auction, filtering lower-quality inventory, prioritising placements more likely to support outcomes, and enriching them with additional data signals.
• Unified Media & Creative Optimisation: Connects media quality with Aryel creative formats and engagement data to inform real-time optimisation.
• Outcomes-Driven Optimisation: OneTag Dynamo uses AI to optimise towards business objectives such as CPA, CPC and CPL, rather than efficiency metrics alone.

Key Features

• AI-Powered Curation: Shapes inventory using performance signals, semantic analysis and contextual relevance to identify higher-quality opportunities for buyers.
• Creative Intelligence (Aryel): Interactive creative formats designed to generate engagement signals such as interaction, attention and behavioural indicators.
• Dynamic Outcomes Loop (Dynamo): Feeds performance data, including CPA, CPC and engagement, back into supply decisioning for impression-level optimisation.
• Media Quality & Attention Signals: Uses curation controls such as time-in-view, ads-to-content ratio and MFA (Made for Advertising) filtering.
• Cookieless Data Integration: Uses contextual, attention and first-party signals to support performance in a privacy-first environment.

Key Results

• 20–50% Performance Uplift: Outcomes-based optimisation can improve campaign performance across engagement and conversion metrics.
• Higher Engagement & CTR: Interactive creative can generate stronger engagement than standard display formats.
• Increased Yield & Demand Quality: Supply-side optimisation can improve inventory value and attract more outcome-focused demand.
• 170 Billion Daily Auctions: Cloud infrastructure supports large-scale data processing and delivery across campaigns.

"The industry doesn’t need more complexity, it needs a better value exchange, where media, creative, and data work together to deliver real outcomes."

Filippo Gramigna, co-CEO, OneTag

PubMatic — AgenticOS

Company: PubMatic (Nasdaq: PUBM) | Product: AgenticOS / PubMatic Assistant | Website: https://pubmatic.com/solutions/agents

PubMatic launched AgenticOS at CES in January 2026. It is the first operating system built for agentic advertising, where AI agents plan, transact, and optimise media across premium digital supply while people set the strategy. AgenticOS spans more than 20 agents and connects directly to PubMatic's Activate buying platform, so it works with existing buying models rather than replacing them.

Buyers choose their level of AI integration. PubMatic Assistant offers a chat-based interface for discovering audiences and inventory. Proprietary or white-label buyer agents can run fully autonomous, end-to-end campaigns. In between, buyers set objectives, guardrails, brand-safety requirements, and creative parameters in their preferred LLM interface and let agents handle setup, activation, and optimisation within those limits.

Core Capabilities

• Natural-language campaign planning. Buyers describe objectives, audiences, and guardrails in plain language. AgenticOS interprets that intent and turns it into a live campaign, with no manual form-filling and no complex UIs.
• Custom and white-label buyer agents. Agencies and trading desks can deploy their own proprietary agents inside AgenticOS, including white-label agents that run agentic buying under their own brand.
• Human control through clear guardrails. People set the objectives, constraints, and creative intent. Agents operate within those limits, with real-time transparency over delivery and performance.
• Continuous monitoring. Agents track pacing, fill, and performance, surface clear recommendations, and report where working media goes, with fee transparency built in.

Key Features

• An operating system of more than 20 agents. Agents cover audience discovery, inventory marketplace and curation, media activation with brand safety and budget controls, deal management, and insights. Buy-side agents generate ranked audience recommendations from a plain-English or a brand URL, turn a prompt into curated deal templates, and monitor pacing, performance, and CPMs.
• A three-layer architecture. The infrastructure layer runs on PubMatic's owned NVIDIA-accelerated computing, supporting up to 5x faster decisioning and sub-millisecond response times. The application layer interprets intent through open protocols, including the Ad Context Protocol (AdCP) and Model Context Protocol (MCP). The transaction layer connects decisioning directly to Activate across Programmatic Guaranteed and Private Marketplace deals.
• Premium supply at scale. AgenticOS draws on more than 2,000 publishers and 100,000 sites and apps, including 28 of the top 30 global streamers, plus more than 200 data and commerce media partners, across premium CTV, video, live sports, mobile, and the open internet.
• Open, with no technical lift. Buyers connect once through an MCP or AdCP-enabled connection, with no intermediary integrations and no development work, and reach billions of premium daily impressions.

Key Results

• Efficient Early Results: has delivered 40% more impressions at 30% lower effective CPM in early campaigns.
• Workflow Efficiency: Campaign setup time was slashed by 87% through natural-language briefs that automate planning and activation.

• Buyside Cost Reduction: 80% reduction in buyside costs from early results.

  By April 2026, around 30 end-to-end agentic campaigns were running globally, supported by more than 1,000 AI-enabled deals.

“Agentic advertising should give buyers more control, not less, and should take cost out of the supply chain rather than simply automating it. The goal is a faster, more transparent, and more efficient open internet, where buyers spend their time on strategy and performance rather than setup and troubleshooting. With agentic systems, you can describe the audience you want in plain language and choose from a set of proposals, rather than picking and choosing manually.”

Mohsin Pervez, Associate Vice President, Technical Account Management, EMEA

Particular Audience — AI-Native Retail Media & Search Platform

Company: Particular Audience | Product: Unified Customer AI (Retail Media + AI Search & Personalisation) | Website: https://particularaudience.com/

Particular Audience helps retailers turn search, recommendations and sponsored placements into a single AI decisioning layer. Rather than relying on manual keyword targeting or broad audience segments, the platform uses real-time shopper signals and commercial priorities to decide which product, ad or recommendation should appear for each customer in each moment.

Core Capabilities

• Signal-Led Decisioning: Captures explicit signals such as search terms and page navigation alongside implicit behaviour such as product views, hovers, basket adds and sequence patterns, then adapts the on-site experience as the journey unfolds.
• AI Search & Personalisation: Combines collaborative filtering, NLP, computer vision and vector/transformer-based search models to power recommendations, search relevance and sponsored placements from one customer-intent engine.

Key Features

• Retail Media Optimisation: Lets retailers run relevant campaigns across sponsored slots, banners, search results, listing pages and recommenders, balancing advertiser reach with relevance and shopper experience.
• Search Model A/B Testing: Allows teams to compare model configurations, data fields and embedding structures, with analytics around performance rather than treating AI decisioning as a black box.
• Real-Time Recommendation Logic: Updates recommendations as shoppers view products, add to basket and move through pages, with retailer-controlled tactics such as most recent item, first basket item or product relationship.
• Viewability & Incrementality Measurement: Tracks true slot impressions, widget views, organic versus sponsored share of voice and post-impression test cohorts so retailers can understand attribution and lift.

Key Results

• 10.4% CTR vs 1.97% Fallback: A relevant cart-page recommendation tactic outperformed a generic popular-items fallback, with 10.4% click-through and 12.5% click-conversion versus 1.97% and 8%.
• Ads Live in Five Weeks: For retailers with long API roadmaps, the JavaScript implementation can get ads up and running quickly, while leaving a path to deeper API integration once proven.
• Incremental Share of Voice: Gradient-boosted campaigns for Bosch and TCL on an appliance retailer showed clear incremental gains in share of voice, share of click and share of sales.

"Sponsored slots, banners and placements should be as tailored to each customer's journey as possible: right ad, right customer, right moment."

James Taylor, CEO

RAAS LAB - Agentic Relevance Platform

Company: RAAS LAB (Relevant Advertising At Scale) | Product: Agentic Relevance Platform & Relevant Marketplace (RMP) | Website: https://raaslab.com/

RAAS LAB is the only agentic platform that understands and activates on the relevance between creative and media placement. Unlike traditional contextual tools that only classify pages, RAAS LAB computes a semantic relevance score between the creative asset and the page at the individual impression level.

Core Capabilities

• Semantic Understanding: Comprehends the meaning of both creative and webpage content, building dynamic 'content webs' that discover relevant placements human planners would miss.
• Proprietary Ad Server: Classifies URLs within 1 hour vs. the 48-hour industry standard. Built from scratch for agentic capabilities.

Key Features

• Impression Decisioning: Selects BOTH creative and placement in unison for hyper-personalised messaging at the individual impression level.
• Relevant Marketplace: Programmatic integration into existing DSP or media partner tech stacks with 100% transparency.
• Automated Tagging: Auto-generates robust meta-tags for creatives using computer vision and LLMs.

Key Results

• +21% Brand Lift: Average brand uplift for campaigns maintaining a 70%+ semantic relevance score.
• +35% Attention: Significant increase in user attention metrics due to hyper-relevant contextual alignment.
• -30% CPA: Decreased cost-per-acquisition driven by precise agentic matching, executing what would take 9.5 years of manual human effort in just one second.

"In just one second, RAAS LAB achieves what takes other programmatic tools 9.5 years of manual effort. It needs to go back to these basic principles of advertising: use relevance to unlock outcomes."

Thomas Ives, CEO

Seedtag — Neuro-Contextual AI

Company: Seedtag | Product: Liz (Neuro-Contextual AI) | Website: https://www.seedtag.com/

Seedtag's Liz takes contextual advertising beyond URL categorisation, using AI to understand how content is likely to make people feel, what interests it signals and what purchase intent it creates. The platform applies this neuro-contextual intelligence across open web and CTV to build audiences, insights and dynamic creative at scale.

Core Capabilities

• Transformer Embeddings: Understands words in the context of the sentence and the whole document, then fine-tunes that general language intelligence for Seedtag's URLs, campaign data and advertising-specific tasks.
• Neuro-Contextual AI: Moves beyond keywords and page categories to model emotion, interest and intent, with academic validation from a Columbia University study measuring brain-wave responses to neuro-contextual ads.

Key Features

• Liz Agent: An internal conversational workflow that lets Seedtag teams turn a client brief into a custom audience in seconds, save it and activate a campaign, with an ambition to externalise the tool for agencies and brands.
• CTV Integration: Extends neuro-contextual intelligence into connected TV by combining bid-request and show-level signals where available with open-web behaviour through a household contextual graph.
• Dynamic Creative Optimisation: Uses generative AI to turn a single creative into multiple variants that can change dynamically depending on the emotion, interest or intent detected around the content.

Key Results

• +350% Emotional Engagement: A Columbia University study found emotion-driven engagement 350% higher than non-contextual advertising, with some KPIs also outperforming basic contextual approaches.
• Audience Creation in Seconds: Custom audience creation that once required months of model training can now be prompted through Liz from a client brief and converted into an activatable segment.
• Global Scale: Transformer embedding architecture allows Seedtag to add a new language in a single day, compared with the year-long market-opening process required by older manual tagging workflows.

“Seedtag doesn’t just understand what content is about; Liz understands the receptive state it creates, from the open web to CTV. By connecting context, emotion, interest and intent across channels, we help brands reach people in the moments when their attention is most meaningful and their message is most likely to resonate.”

Grego Martinez, VP of Product

Smartology — SmartMatch™ Contextual Platform

Company: Smartology | Product: SmartMatch™ SaaS Platform | Website: https://smartology.net/

Smartology is a specialist in AI-driven contextual advertising, matching branded adverts with contextually relevant articles across a curated network of over 300 premium publishers. Entirely cookie-free since 2010.

SmartMatch™'s fine-tuned multilingual LLM reads both the publisher article and the advertiser landing page, converts each into a high-dimensional semantic vector, and calculates cosine similarity to surface a highly relevant, contextual ad placement - entirely without cookies.

Core Capabilities

• Contextual Similarity: Keyword, topic and category tools result in shallow matches, whereas SmarMatch’s approach of generating deep semantic vectors from the full body of both advertiser content and publisher articles, creates highly relevant and surprisingly intuitive matches.
• Multilingual LLM: Fine-tuned pre-trained multilingual model.

Key Features

• AI Brand Safety: Dedicated deep-learning model trained on curated examples. SmartMatch uses semantic category exclusions rather than blunt keyword blocklists, preventing accidental over-blocking.
• Proprietary DSP: End-to-end control over bidding, optimisation and delivery through RTB programmatic infrastructure via self activation or managed service.
• Creative Flexibility: A diverse selection of ad formats, including display, native, video, and rich media, is available to align with specific campaign objectives. Creative executions can be refined to effectively address both B2B and B2C activation requirements.

Key Results

• 100% Cookie-Free: Has delivered targeted, performant campaigns entirely without cookies since inception in 2010.
• Premium Execution: Matches complex messaging across an exclusive, curated network of over 300 top-tier global publishers.
• Rapid Activation: Intuitive platform and AI enables campaigns to go live in under 24 hours.

"We’ve been training and fine-tuning models since 2015, but I'm still amazed at how fast things have progressed… For the time being, you still need that human element."

Marcus, Chief Product & Technology Officer