We monitor AI news \u2013 new models, libraries and tools \u2013 test them rapidly, and publish short reactions on our blog.<\/p>\n
The goal is speed and signal rather than exhaustive analysis: we get the thing up and running with minimal setup, test it against its main use cases, and capture first impressions, what worked, what did not, and whether it is worth a closer look.<\/p>\n
The latest reactions and technical notes are collected below as Research Outputs.<\/p>\n"},"related_publications":[{"type":"research_feed","content_type":"article","title":"What did we work on yesterday? Testing memory for coding agents","slug":"what-did-we-work-on-yesterday-testing-memory-for-coding-agents","date":"2026-10-01T13:54:15+00:00","quarter":"Q3 2026","authors":[{"id":42,"display_name":"George Konstantoulas","first_name":"George","last_name":"Konstantoulas","nickname":"george.konstantoulas","user_nicename":"george-konstantoulas","user_email":"george.konstantoulas@satalia.com","biographical_info":"Georgios Konstantoulas is an Electrical and Computer Engineering graduate currently working as a Junior Data Scientist at Satalia (WPP Group). His background is primarily on computer vision, with hands-on experience in designing and benchmarking deep learning architectures in representation learning and classification tasks. He is currently contributing to research in agentic AI systems, exploring design and deployment in real-world settings.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/09\/53333849015_e5884fc541_k.jpg","job_title":"Junior Data Scientist","is_lead":false,"display_as_researcher":true,"order_priority":null}]},{"type":"research_feed","content_type":"article","title":"Evaluating MinerU for PDF-to-Markdown Extraction","slug":"mineru-the-hard-part-isnt-the-parsing","date":"2026-09-29T08:30:39+00:00","quarter":"Q3 2026","authors":[{"id":43,"display_name":"Serafeim Papadias","first_name":"Serafeim","last_name":"Papadias","nickname":"serafeim.papadias","user_nicename":"serafeim-papadias","user_email":"serafeim.papadias@satalia.com","biographical_info":"Serafeim (Makis) Papadias is a Senior Data Scientist in Satalia's Research Lab, contributing to WPP Research's agenda on peer-to-peer communities of LLM agents. He holds a PhD from TU Berlin in scalable data systems and graph algorithms (VLDB, ICDE, EDBT) and previously built graph-based retrieval systems at Athena Research Center. He focuses on the network side of agent communities: how expertise is found, how teams form, and how information stays reliable as the population grows.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/09\/makis_onboarding_cropped.jpeg","job_title":"Senior Data Scientist","is_lead":false,"display_as_researcher":true,"order_priority":null}]},{"type":"research_feed","content_type":"article","title":"What Is a Promised Time Window Actually Worth?","slug":"what-is-a-promised-time-window-actually-worth","date":"2026-09-22T09:30:10+00:00","quarter":"","authors":[{"id":22,"display_name":"Panagiotis Repoussi","first_name":"Panagiotis","last_name":"Repoussis","nickname":"panos","user_nicename":"panos","user_email":"panos@satalia.com","biographical_info":null,"avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/04\/panos.jpg","job_title":"Head of Research","is_lead":false,"display_as_researcher":true,"order_priority":0}]},{"type":"research_feed","content_type":"article","title":"Are open-source embedding models good enough? A comparative study","slug":"are-open-source-embedding-models-good-enough-a-comparative-study","date":"2026-07-03T12:29:29+00:00","quarter":"","authors":[{"id":7,"display_name":"Eirini Kolimatsi","first_name":"Eirini","last_name":"Kolimatsi","nickname":"eirini.kolimatsi","user_nicename":"eirini-kolimatsi","user_email":"eirini.kolimatsi@satalia.com","biographical_info":"Eirini is a Data Scientist at Satalia with a multidisciplinary background in Management Science and Computer Science. She specialises in architecting end-to-end data science solutions, leveraging a deep technical toolkit to solve complex industrial challenges across diverse sectors. Known for bridging the gap between theoretical research and scalable application, she focuses on delivering high-impact models that translate abstract data patterns into actionable strategic intelligence.\r\n\r\nHer current research focuses on sophisticated campaign performance multimodal modelling and the development of data enrichment frameworks to maximise predictive accuracy.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/03\/DSC_0100.jpg","job_title":"Data Scientist","is_lead":false,"display_as_researcher":true,"order_priority":2},{"id":8,"display_name":"Jael Freixanet","first_name":"Jael","last_name":"Freixanet","nickname":"jael.freixanet","user_nicename":"jael-freixanet","user_email":"jael.freixanet@satalia.com","biographical_info":"Jael is a Data Scientist at Satalia, leveraging her physics background for a deep analytical foundation in complex systems analysis and modelling. Her experience, spanning foundational research in computational physics and a proven track record in data science consultancy, provides a unique perspective for architecting robust, scalable models in intricate environments. In Satalia's Research Lab, she bridges scientific methodology with industrial innovation to address WPP\u2019s most sophisticated data challenges. Her current research focuses on multimodal fusion models, aiming to improve campaign performance and pioneer state-of-the-art machine learning.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/03\/jael_img.jpeg","job_title":"Data Scientist","is_lead":false,"display_as_researcher":true,"order_priority":2},{"id":36,"display_name":"Simon Philp","first_name":"Simon","last_name":"Philp","nickname":"simon.philp","user_nicename":"simon-philp","user_email":"simon.philp@satalia.com","biographical_info":"Simon is a Data Scientist with experience applying Bayesian methods and predictive modeling to design innovative, data-driven solutions for complex operational challenges. His background spans machine learning, deep learning, and advanced statistical analysis, alongside leading impactful AI initiatives that drive organizational efficiency. During his time as a Research Assistant with the Met Office, he utilized Bayesian techniques to model environmental patterns, bringing together rigorous research and practical, real-world application.\r\n\r\nSimon holds a Master's degree with Distinction in Data Science and Analytics from the University of Leeds and a First-Class Honours degree in Mathematics from Durham University. This strong academic and research foundation supports his work in translating complex mathematical concepts into robust, scalable technology that automates processes and delivers measurable value.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/06\/author.jpeg","job_title":"Data Scientist","is_lead":false,"display_as_researcher":true,"order_priority":null}]},{"type":"research_feed","content_type":"article","title":"Evaluating Synthetic Audiences as Proxies for Human Users","slug":"synthetic-audiences-as-proxies-for-human-users-a-systematic-literature-review","date":"2026-06-24T10:00:46+00:00","quarter":"","authors":[{"id":5,"display_name":"Ted Lappas","first_name":"Ted","last_name":"Lappas","nickname":"theodoros.lappas","user_nicename":"theodoros-lappas","user_email":"Theodoros.Lappas@wpp.com","biographical_info":"Ted co-leads WPP Research and serves as Head of Data Science at Satalia. He is an Assistant Professor in the Department of Marketing and Communication at the Athens University of Economics and Business. His research spans scalable algorithms for multimodal data, synthetic data generation, simulation-based verification for AI agents, and information diffusion and collective intelligence in expert networks.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/04\/pic.png","job_title":"Head of Data Science","is_lead":false,"display_as_researcher":true,"order_priority":1}]},{"type":"research_feed","content_type":"article","title":"Learning through experience: teaching the viral Hermes agent to automate our work","slug":"learning-through-experience-teaching-the-viral-hermes-agent-to-automate-our-work","date":"2026-06-10T16:07:38+00:00","quarter":"","authors":[{"id":5,"display_name":"Ted Lappas","first_name":"Ted","last_name":"Lappas","nickname":"theodoros.lappas","user_nicename":"theodoros-lappas","user_email":"Theodoros.Lappas@wpp.com","biographical_info":"Ted co-leads WPP Research and serves as Head of Data Science at Satalia. He is an Assistant Professor in the Department of Marketing and Communication at the Athens University of Economics and Business. His research spans scalable algorithms for multimodal data, synthetic data generation, simulation-based verification for AI agents, and information diffusion and collective intelligence in expert networks.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/04\/pic.png","job_title":"Head of Data Science","is_lead":false,"display_as_researcher":true,"order_priority":1},{"id":15,"display_name":"Nikos Gkikizas","first_name":"Nikos","last_name":"Gkikizas","nickname":"nikos.gkikizas","user_nicename":"nikos-gkikizas","user_email":"nikos.gkikizas@satalia.com","biographical_info":"Nikos is a Staff Data Scientist at Satalia. He started off as a mining engineer and later specialized in explosives and blasting engineering. Through signal processing he was introduced to data science, with which he gradually fell in love with. Over the last decade he's been working in Data Science, having developed various simulation, predictive and optimization solutions.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/04\/signal-2026-04-01-111218.jpeg","job_title":"Staff Data Scientist","is_lead":false,"display_as_researcher":false,"order_priority":null},{"id":16,"display_name":"Andreas Stavrou","first_name":"Andreas","last_name":"Stavrou","nickname":"andreas.stavrou","user_nicename":"andreas-stavrou","user_email":"andreas.stavrou@satalia.com","biographical_info":"Andreas is a Lead Data Scientist at Satalia and part of the WPP Research team. With over a decade of hands-on experience, he has delivered machine learning solutions across retail, online gambling, and credit risk, and now builds AI systems at scale in the global advertising industry.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/03\/avatar.jpg","job_title":"Lead Data Scientist","is_lead":false,"display_as_researcher":true,"order_priority":null}]},{"type":"research_feed","content_type":"article","title":"A 50-line python function outperformed every frontier LLM – With 100% accuracy","slug":"a-50-line-python-function-outperformed-every-frontier-llm-with-100-accuracy","date":"2026-05-27T08:42:38+00:00","quarter":"","authors":[{"id":15,"display_name":"Nikos Gkikizas","first_name":"Nikos","last_name":"Gkikizas","nickname":"nikos.gkikizas","user_nicename":"nikos-gkikizas","user_email":"nikos.gkikizas@satalia.com","biographical_info":"Nikos is a Staff Data Scientist at Satalia. He started off as a mining engineer and later specialized in explosives and blasting engineering. Through signal processing he was introduced to data science, with which he gradually fell in love with. Over the last decade he's been working in Data Science, having developed various simulation, predictive and optimization solutions.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/04\/signal-2026-04-01-111218.jpeg","job_title":"Staff Data Scientist","is_lead":false,"display_as_researcher":false,"order_priority":null}]},{"type":"research_feed","content_type":"article","title":"MiroFish: Is swarm intelligence worth the cloud bill?","slug":"mirofish-is-swarm-intelligence-worth-the-cloud-bill","date":"2026-05-19T14:52:36+00:00","quarter":"","authors":[{"id":15,"display_name":"Nikos Gkikizas","first_name":"Nikos","last_name":"Gkikizas","nickname":"nikos.gkikizas","user_nicename":"nikos-gkikizas","user_email":"nikos.gkikizas@satalia.com","biographical_info":"Nikos is a Staff Data Scientist at Satalia. He started off as a mining engineer and later specialized in explosives and blasting engineering. Through signal processing he was introduced to data science, with which he gradually fell in love with. Over the last decade he's been working in Data Science, having developed various simulation, predictive and optimization solutions.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/04\/signal-2026-04-01-111218.jpeg","job_title":"Staff Data Scientist","is_lead":false,"display_as_researcher":false,"order_priority":null}]},{"type":"research_feed","content_type":"article","title":"Using DeepSeek v4 Pro as an Agentic Brain","slug":"using-deepseek-v4-pro-as-an-agentic-brain","date":"2026-05-14T19:33:53+00:00","quarter":"","authors":[{"id":5,"display_name":"Ted Lappas","first_name":"Ted","last_name":"Lappas","nickname":"theodoros.lappas","user_nicename":"theodoros-lappas","user_email":"Theodoros.Lappas@wpp.com","biographical_info":"Ted co-leads WPP Research and serves as Head of Data Science at Satalia. He is an Assistant Professor in the Department of Marketing and Communication at the Athens University of Economics and Business. His research spans scalable algorithms for multimodal data, synthetic data generation, simulation-based verification for AI agents, and information diffusion and collective intelligence in expert networks.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/04\/pic.png","job_title":"Head of Data Science","is_lead":false,"display_as_researcher":true,"order_priority":1},{"id":16,"display_name":"Andreas Stavrou","first_name":"Andreas","last_name":"Stavrou","nickname":"andreas.stavrou","user_nicename":"andreas-stavrou","user_email":"andreas.stavrou@satalia.com","biographical_info":"Andreas is a Lead Data Scientist at Satalia and part of the WPP Research team. With over a decade of hands-on experience, he has delivered machine learning solutions across retail, online gambling, and credit risk, and now builds AI systems at scale in the global advertising industry.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/03\/avatar.jpg","job_title":"Lead Data Scientist","is_lead":false,"display_as_researcher":true,"order_priority":null},{"id":17,"display_name":"Ilias Papastratis","first_name":"Ilias","last_name":"Papastratis","nickname":"ilias.papastratis","user_nicename":"ilias-papastratis","user_email":"ilias.papastratis@satalia.com","biographical_info":"Ilias Papastratis is a Senior Data Scientist based in United Kingdom. He specializes in LLM-based agentic architectures, generative models, and computer vision for large-scale enterprise applications. Currently working at Satalia (WPP Group) , he designs and deploys production-grade AI solutions\u2014including custom multi-agent LLM systems and diffusion-based image generation pipelines \u2014for globally recognized clients.\r\n\r\nPreviously, he spent five years as a Deep Learning Researcher and developed computer vision models for real-time sign language translation and industrial object recognition. He holds a combined Bachelor\u2019s and Master\u2019s degree in Electrical Engineering and Computer Science from the University of Patras , alongside an MSc in Digital Media and Computational Intelligence from Aristotle University of Thessaloniki. In addition to being the author of peer-reviewed publications, he actively contributes to open-source deep learning projects.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/04\/ilias-1.jpg","job_title":"Senior Data Scientist","is_lead":null,"display_as_researcher":null,"order_priority":null}]},{"type":"research_feed","content_type":"article","title":"Autoresearch: a closer look at the agent that runs its own experiments","slug":"autoresearch","date":"2026-05-11T13:27:00+00:00","quarter":"","authors":[{"id":16,"display_name":"Andreas Stavrou","first_name":"Andreas","last_name":"Stavrou","nickname":"andreas.stavrou","user_nicename":"andreas-stavrou","user_email":"andreas.stavrou@satalia.com","biographical_info":"Andreas is a Lead Data Scientist at Satalia and part of the WPP Research team. With over a decade of hands-on experience, he has delivered machine learning solutions across retail, online gambling, and credit risk, and now builds AI systems at scale in the global advertising industry.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/03\/avatar.jpg","job_title":"Lead Data Scientist","is_lead":false,"display_as_researcher":true,"order_priority":null}]},{"type":"research_feed","content_type":"article","title":"Inside the MemPalace: Does the structure earn its keep?","slug":"inside-the-mempalace-does-the-structure-earn-its-keep","date":"2026-05-06T10:01:29+00:00","quarter":"","authors":[{"id":15,"display_name":"Nikos Gkikizas","first_name":"Nikos","last_name":"Gkikizas","nickname":"nikos.gkikizas","user_nicename":"nikos-gkikizas","user_email":"nikos.gkikizas@satalia.com","biographical_info":"Nikos is a Staff Data Scientist at Satalia. He started off as a mining engineer and later specialized in explosives and blasting engineering. Through signal processing he was introduced to data science, with which he gradually fell in love with. Over the last decade he's been working in Data Science, having developed various simulation, predictive and optimization solutions.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/04\/signal-2026-04-01-111218.jpeg","job_title":"Staff Data Scientist","is_lead":false,"display_as_researcher":false,"order_priority":null}]},{"type":"research_feed","content_type":"article","title":"OpenClaw for messaging: a closer look at WhatsApp and Telegram","slug":"openclaw-for-messaging","date":"2026-05-04T08:28:53+00:00","quarter":"","authors":[{"id":16,"display_name":"Andreas Stavrou","first_name":"Andreas","last_name":"Stavrou","nickname":"andreas.stavrou","user_nicename":"andreas-stavrou","user_email":"andreas.stavrou@satalia.com","biographical_info":"Andreas is a Lead Data Scientist at Satalia and part of the WPP Research team. With over a decade of hands-on experience, he has delivered machine learning solutions across retail, online gambling, and credit risk, and now builds AI systems at scale in the global advertising industry.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/03\/avatar.jpg","job_title":"Lead Data Scientist","is_lead":false,"display_as_researcher":true,"order_priority":null}]},{"type":"research_feed","content_type":"technical-walkthrough","title":"Tech Pulse Pod: Technical walkthrough","slug":"tech-pulse-pod-technical-walkthrough","date":"2026-05-04T08:26:53+00:00","quarter":"Q1 2026","authors":[{"id":16,"display_name":"Andreas Stavrou","first_name":"Andreas","last_name":"Stavrou","nickname":"andreas.stavrou","user_nicename":"andreas-stavrou","user_email":"andreas.stavrou@satalia.com","biographical_info":"Andreas is a Lead Data Scientist at Satalia and part of the WPP Research team. With over a decade of hands-on experience, he has delivered machine learning solutions across retail, online gambling, and credit risk, and now builds AI systems at scale in the global advertising industry.","avatar_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/03\/avatar.jpg","job_title":"Lead Data Scientist","is_lead":false,"display_as_researcher":true,"order_priority":null}]}],"featured_image_url":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/05\/image-8.png","featured_image_sizes":{"thumbnail":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/05\/image-8-150x150.png","medium":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/05\/image-8-300x300.png","large":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/05\/image-8.png","full":"https:\/\/cms.research.wpp.com\/wp-content\/uploads\/2026\/05\/image-8.png"},"_links":{"self":[{"href":"https:\/\/cms.research.wpp.com\/index.php?rest_route=\/wp\/v2\/research_pods\/1362","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/cms.research.wpp.com\/index.php?rest_route=\/wp\/v2\/research_pods"}],"about":[{"href":"https:\/\/cms.research.wpp.com\/index.php?rest_route=\/wp\/v2\/types\/research_pod"}],"author":[{"embeddable":true,"href":"https:\/\/cms.research.wpp.com\/index.php?rest_route=\/wp\/v2\/users\/16"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/cms.research.wpp.com\/index.php?rest_route=\/wp\/v2\/media\/1364"}],"wp:attachment":[{"href":"https:\/\/cms.research.wpp.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=1362"}],"wp:term":[{"taxonomy":"pod_status","embeddable":true,"href":"https:\/\/cms.research.wpp.com\/index.php?rest_route=%2Fwp%2Fv2%2Fpod_status&post=1362"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/cms.research.wpp.com\/index.php?rest_route=%2Fwp%2Fv2%2Fppma_author&post=1362"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}