This week we feature articles from Stefan Baack & Mozilla Insights, and Steve Jones.

Additional reading comes from Daniel Tunkelang, Kate Moran, Louis Rosenberg, and Matt Marshall.

News comes from Franz, Optimizely & Writer, Otter.ai, and Grammarly.

All previous issues are available at https://gilbane.com/gilbane-advisor-index

---

Opinion / Analysis

Training data for the price of a sandwich

It’s easy to think you know what Common Crawl is, but if you’re using some version of it as a key source for training data for generative AI, you need to do your own due diligence. This article from Stefan Baack and Mozilla Insights is a little bit repetitive so a quicker read than it first appears, and well worth it. (43 min)

https://foundation.mozilla.org/en/research/library/generative-ai-training-data/common-crawl/

Your data warehouse thinking is killing your AI ambitions

I’m not sure how many of you don’t already know the difference, but Steve Jones’ article is a useful explanation for reading, or sharing, with certain colleagues…

“The reality is that traditional data warehousing thinking and its idea of some mythical “perfect” post transactional data set only exists because historically data hasn’t been important, except for finance regulations… 

The future is different, AI is critical to that future, AI that works at operational speed and can be engaged in operational decisions. This is the participation of data in business, the day-to-day, minute-to-minute running of the business, not simply reporting after the fact what went on…” (7 min)

https://blog.metamirror.io/your-data-warehouse-thinking-is-killing-your-ai-ambitions-849356dd3679

More Reading

• Precision, Recall, and Desirability via Daniel Tunkelang
• The UX of AI: Lessons from Perplexity via Nielsen Norman Group
• Mixed Reality or Spatial Computing via Louis Rosenberg
• How enterprises are using open source LLMs: 16 examples via VentureBeat

All Gilbane Advisor issues

---

Content technology news

Otter.ai announces Meeting GenAI

AI Chat allow users to tap into the collective wisdom from past meetings no matter which platform.
https://otter.ai

Optimizely integration with Writer now live

Writer’s enterprise-grade, industry-specific generative AI capabilities within the Optimizely Content Marketing Platform.
https://www.optimizely.com ■ https://writer.com

Grammarly announces general availability of App Actions

New third-party app integrations and actions enable businesses and professionals to simplify workflows and reduce constant context-switching.
https://www.grammarly.com/app-actions

Franz announces Gruff 9

The web-based Knowledge Graph visualization tool offers LLM integration and unique RDF* (RDFStar) features for building AI applications.
https://allegrograph.com/products/gruff/

All content technology news

---

The Gilbane Advisor is authored by Frank Gilbane and is ad-free, cost-free, and curated for content, computing, web, data, and digital experience technology and information professionals. We publish recommended articles and content technology news weekly. We do not sell or share personal data.

Subscribe | View online | Editorial policy | Privacy policy | Contact

This week we feature articles from Sanjeev Mohan & Merv Adrian, and Simon Wardley.

Additional reading comes from Molly White, Cobus Greyling, Sissie Hsiao, Aleksandr Perevalov, Alan Morrison, and Michael Nuñez

News comes from Apple, Google, Digimarc, and Bluesky.

All previous issues are available at https://gilbane.com/gilbane-advisor-index

---

Opinion / Analysis

Dive deeper into Gartner’s 2023 cloud DBMS magic quadrant

Former Gartner analysts Sanjeev Mohan and Merv Adrian report on and analyze Gartner’s most recent cloud database management systems MQ. Their article is a useful short summary with some added insight into the challenges in creating MQs. And, of course you don’t need to be a client to read it. (12 min)

https://sanjmo.medium.com/dive-deeper-into-gartners-2023-cloud-dbms-magic-quadrant-key-takeaways-and-beyond-fcdd58406f8e

A good enough map

Simon Wardley is often asked about adding a new axis or legend to Wardley maps, and explains why he prefers not to. He makes a good case that adding more complexity would reduce Wardley maps usefulness as a planning tool. It is easy to see how doing so would confound participants in an offsite strategy meeting. I think his answer also serves as a good introduction to Wardley maps if you’re not familiar with them. (3 min)

https://swardley.medium.com/a-good-enough-map-eaed8a525bf4

More Reading

• Ouchy! Molly White’s Review: Chris Dixon’s Read Write Own via [citation needed]
• Comparing human, LLM & LLM-RAG responses via Cobus Greyling
• Lingua Franca — Entity-aware machine translation approach for question answering over knowledge graphs via Towards Data Science
• What data scientists overlook when it comes to knowledge graphs via Data Science Central

All Gilbane Advisor issues

---

Content technology news

Apple releases ‘MGIE’ AI model for instruction-based image editing

MGIE leverages multimodal large language models (MLLMs) to interpret natural language user commands and perform pixel-level manipulations.
https://venturebeat.com/ai/apple-releases-mgie-a-revolutionary-ai-model-for-instruction-based-image-editing/

Bard becomes Gemini: Try Ultra 1.0 and a new mobile app

Gemini Advanced is available today in more than 150 countries and territories in English, and via a new Android app. An iOS app is coming soon.
https://blog.google/products/gemini/bard-gemini-advanced-app/

Digimarc releases their next-generation digital watermarks

Updated Digimarc digital watermarks deliver greater security, adaptability, imperceptibility, performance, and control.
https://www.digimarc.com/press-releases/2024/02/07/digimarc-releases-next-generation-digital-watermarks

Join Bluesky today (bye, invites!)

Bluesky is building an open social network where anyone can contribute, while still providing an easy-to-use experience for users.
https://bsky.social/about/blog/02-06-2024-join-bluesky

All content technology news

---

The Gilbane Advisor is authored by Frank Gilbane and is ad-free, cost-free, and curated for content, computing, web, data, and digital experience technology and information professionals. We publish recommended articles and content technology news weekly. We do not sell or share personal data.

Subscribe | View online | Editorial policy | Privacy policy | Contact

This week we feature articles from Chia Jeng Yang, and Antoine Vidal.

Additional reading comes from AI2 Blog, Jarno van Driel, Anthony Alcaraz, and Manuel Faysse.

News comes from Fivetran & dbt Labs, Mindbreeze & Ariza, Box, and Quark.

All previous issues are available at https://gilbane.com/gilbane-advisor-index

---

Opinion / Analysis

Advanced RAG and the 3 types of recursive retrieval

Chia Jeng Yang…
“In this article, we dive into the different ways to use Recursive or Iterative Retrieval to help develop self-learning RAG systems that are able to learn over time, and to deeply explore unstructured data, either autonomously, or in directions that you point the knowledge graph to.” (7 min)

https://medium.com/enterprise-rag/advanced-rag-and-the-3-types-of-recursive-retrieval-cdd0fa52e1ba

Using AI for 3D rendering — a practical guide for designers

Antoine Vidal…
“The slow, meticulous process of creating high-quality renders is often incompatible with the fast pace required from early concept ideation. How do we find that balance?” (11 min)

https://uxdesign.cc/using-ai-for-3d-rendering-a-practical-guide-for-designers-a2a037ed1ad0

More Reading

• “A truly open large language model with pre-training data and training code…” OLMo: Open Language Model via AI2 Blog
• Does structured data markup influence EEAT? No? an… energetic expert explains. via The Invisible Graph
• DeepMind’s AlphaGeometry and beyond… The future is Neuro-Symbolic: How AI reasoning is evolving via Towards Data Science
• CroissantLLM: A truly bilingual French-English language model Bravo! Bonne idée via Hugging Face

All Gilbane Advisor issues

---

Content technology news

Fivetran introduces dbt Cloud orchestration integration

Integration enables joint customers to automate transformations post data load, reducing latency, cutting costs, and unlocking data’s potential.
https://www.fivetran.com/data-movement/transformations/dbt-cloud ■ https://www.getdbt.com/product/dbt-cloud

Mindbreeze and Ariza Content Solutions partner

The partnership expects to employ and enhance insights & search experiences and provide companies with better content management capabilities.
https://inspire.mindbreeze.com ■ https://www.arizacs.com

Box announces Box App for Apple Vision Pro

The uniquely immersive experiences and infinite desktop of visionOS transforms how people work with content in the secure Box Content Cloud.
https://blog.box.com/box-apple-vision-pro-brings-your-enterprise-content-life

Quark announces QPP NextGen v3.3

Newest release of Quark Publishing Platform NextGen includes more built-in AI functionality to transform content operations processes.
https://www.quark.com/whats-new-quark-publishing-platform

All content technology news

---

The Gilbane Advisor is authored by Frank Gilbane and is ad-free, cost-free, and curated for content, computing, web, data, and digital experience technology and information professionals. We publish recommended articles and content technology news weekly. We do not sell or share personal data.

Subscribe | View online | Editorial policy | Privacy policy | Contact

This week we feature articles from Matt Murphy, Tim Tully, Grace Ge, Derek Xiao & Katie Keller, and Gary Marcus.

Additional reading comes from Dharmesh Shah, Paul Cheesbrough & Melody Hildebrandt, Casey Newton, and Chris Ashby.

News comes from Uniform, ActionIQ, Pega, MadCap Software & Xyleme.

All previous issues are available at https://gilbane.com/gilbane-advisor-index

---

Opinion / Analysis

The modern AI stack: Design principles for the future of enterprise AI architectures

The AI technology ecosystem is difficult to keep up with, even for those whose full-time day job is doing just that, and especially for those tasked with recommending or applying it to enterprise software architectures and applications. Matt Murphy, Tim Tully, Grace Ge, Derek Xiao, and Katie Keller from Menlo Ventures bring you up-to-date. (10 min)

https://menlovc.com/perspective/the-modern-ai-stack-design-principles-for-the-future-of-enterprise-ai-architectures/

No, multimodal ChatGPT is not going to “trivially” solve Generative AI’s copyright problems

Gary Marcus takes issue with the AI Snake Oil guys over their claim (we pointed to it in last week’s issue) that “Output similarity is a distraction”. Both articles are worth a read. (6 min)

https://garymarcus.substack.com/p/no-multimodal-chatgpt-is-not-going

More Reading

• Delivers… A mere mortal’s visual guide to AI vector embeddings via Dharmesh Shah
• Fox blockchain-based protocol for content verification and traceability… The future of verified content via Tech Insiders
• His platform search continues… Why Platformer is leaving Substack via Platformer
• Ahhyupp, as we hear around the office… Why voice input is flawed as a primary interface input mode via UX Collective

All Gilbane Advisor issues

---

Content technology news

Uniform announces CMS capabilities for Visual Workspace

The new CMS and AI capabilities is meant to streamline the process of building digital experiences, using content from multiple sources.
https://www.uniform.dev

ActionIQ announces CXAI to integrate GenAI data & GenAI content

Built for customer experience (CX), marketing, & data teams, and includes Typeface, a GenAI AI platform for enterprise content creation.
https://www.actioniq.com/newsroom/actioniq-announces-cx-ai-genai/

Pega introduces Pega GenAI Knowledge Buddy

GenAI Knowledge Buddy synthesizes knowledge scattered across content libraries to quickly get answers to employee and customer inquiries.
https://www.pega.com/products/genai-knowledge-buddy

MadCap Software acquires Xyleme to expand suite of intelligent content management products

Organizations will have a single source of enterprise-class solutions for their content needs, from documentation to onboarding customers and their workforce.
https://www.madcapsoftware.com/press-release/madcap-software-acquires-xyleme/ ■ https://xyleme.com

All content technology news

---

The Gilbane Advisor is authored by Frank Gilbane and is ad-free, cost-free, and curated for content, computing, web, data, and digital experience technology and information professionals. We publish recommended articles and content technology news weekly. We do not sell or share personal data.

Subscribe | View online | Editorial policy | Privacy policy | Contact

This week we feature articles from Chua Chin Hon, and Arvind Narayanan & Sayash Kapoor.

Additional reading comes from Anna Kaley & Maria Rosala, Michael Parekh, Chenxin Li, and Sergey Levine & Karol Hausman.

News comes from Ontotext, DataStax, Pinecone, and Altova.

All previous issues are available at https://gilbane.com/gilbane-advisor-index

---

Opinion / Analysis

Fight fire with fire? Building and testing a custom AI-powered fact checker

Chua Chin Hon has been testing available fact checking software, as well as building his own since 2021 and found them all seriously lacking. His latest attempt is built with Custom GPT combined with Bing which he says significantly outperforms all of them. He explains how he did it without any coding. Especially interesting, not to mention timely, is his experience with newsrooms which he uses as a test case. (13 min)

https://chuachinhon.medium.com/fight-fire-with-fire-building-and-testing-a-custom-ai-powered-fact-checker-3eef32a5768e

Generative AI’s end-run around copyright won’t be resolved by the courts

Using the recent New York Times complaint about Open AI, Arvind Narayanan & Sayash Kapoor explain why the case is so complex, and why GenAI “Output similarity is a distraction. Training is the real problem.” (7 min)

https://www.aisnakeoil.com/p/generative-ais-end-run-around-copyright

More Reading

• Intranet usability guidelines: New findings from 57 intranets via Nielsen Norman Group
• Yes, it is problematic… AI: Designing ‘artificial emotions’ into ‘artificial intelligence’ via Michael Parekh
• Friends don’t let friends make bad graphs via Chenxin Li
• Robots need robot data to learn from… The global project to make a general robotic brain via IEEE Spectrum

All Gilbane Advisor issues

---

Content technology news

DataStax launches Data API to simplify GenAI application development

Also included is an updated developer experience for DataStax Astra DB, a vector database for building production-level AI applications.
https://www.datastax.com/blog/general-availability-data-api-for-enhanced-developer-experience

Ontotext’s GraphDB available on the Microsoft Azure Marketplace

The GraphDB listing  provides customers with a consolidated purchasing environment and integration with their existing Azure accounts.
https://www.ontotext.com/company/news/ontotexts-graphdb-solution-is-now-available-on-the-microsoft-azure-marketplace/

Pinecone announces Pinecone Serverless

Pinecone Serverless is launching with integrations to Anthropic, Anyscale, Cohere, Confluent, Langchain, Pulumi, and Vercel.
https://www.pinecone.io/blog/serverless/

Altova announces MobileTogether 9.0

This release introduces full support for MQTT allowing developers to build MQTT-enabled apps for communication among smart devices.
https://www.altova.com/whatsnew

All content technology news

---

The Gilbane Advisor is authored by Frank Gilbane and is ad-free, cost-free, and curated for content, computing, web, data, and digital experience technology and information professionals. We publish recommended articles and content technology news weekly. We do not sell or share personal data.

Subscribe | View online | Editorial policy | Privacy policy | Contact

This week we feature articles from Dhanshree Shripad Shenwai, and Stephanie Shen. 

Additional reading comes from Kyle Wiggers, Caleb Sponheim, Ben Waber & Nathanael J. Fast, and Raphaël Millière & Cameron Buckner.

News comes from Typeface, OpenAI, Argo Translation, and Digimarc.

All previous issues are available at https://gilbane.com/gilbane-advisor-index

---

Opinion / Analysis

A comprehensive study of knowledge editing for large language models

We’re still learning how knowledge graphs and large language models might be combined to reap the benefits of their individual capabilities. Knowledge graph embeddings in LLMs are one approach, but are typically not easily modified (as they are on their own) without re-training. Dhanshree Shripad Shenwai summarizes an interesting new paper on knowledge editing activity in general, and specifically on a new approach to knowledge editing in language models that doesn’t require re-training. (3 min)

Summary: https://www.marktechpost.com/2024/01/06/this-ai-paper-presents-a-comprehensive-study-of-knowledge-editing-for-large-language-models/
Paper: https://arxiv.org/abs/2401.01286

Bayesian Inference: A unified framework for perception, reasoning, and decision-making

Stephanie Shen provides an accessible and useful description of what Bayes’ Theorem is and how it works in English (minus the math), and explains why its applicability is much broader than appreciated. A good refresher for those of us not steeped in it, and a good intro for decision-makers. (14 min)

https://towardsdatascience.com/bayesian-inference-a-unified-framework-for-perception-reasoning-and-decision-making-bb9c47448f8f

More Reading

• OpenAI launches a store for custom AI-powered chatbots via TechCrunch
• Sycophancy in Generative-AI Chatbots, via Nielsen Norman Group
• Skeptics… Is GenAI’s impact on productivity overblown?via HBR
• Excellent… A philosophical introduction to language models – Part I: Continuity with classic debates via arXiv

All Gilbane Advisor issues

---

Content technology news

Typeface announces integration within Microsoft Dynamics 365

Dynamics 365 Customer Insights customers get access to Typeface’s brand and audience-personalized AI seamlessly within their environments.
https://www.typeface.ai/blog/typeface-announces-integration-within-microsoft-dynamics-365-customer-insights-to-help-redefine-marketer-experiences

OpenAI introduces ChatGPT Team

OpenAI introduces ChatGPT Team — ChatGPT Team offers access to GPT-4 and DALL·E 3, tools like Advanced Data Analysis, a dedicated collaborative workspace, and admin tools.
https://openai.com/blog/introducing-chatgpt-team

Argo Translation launches CMSConnect

Support for multiple CMS, PIM, DAM, and CRM platforms expands their ability to help businesses make content available in numerous languages.
https://www.argotrans.com/argo-translation-launches-cmsconnect

Digimarc offers free digital watermark embedding and detection tools to device and chip manufacturers and content creation platforms

The Secure, Automated, Fair, and Efficient (SAFE) digital watermark tools are designed to protect digital content owners and consumers.
https://www.digimarc.com/press-releases/2024/01/04/digimarc-offers-free-digital-watermark-embedding-and-detection-tools

All content technology news

---

The Gilbane Advisor is authored by Frank Gilbane and is ad-free, cost-free, and curated for content, computing, web, data, and digital experience technology and information professionals. We publish recommended articles and content technology news weekly. We do not sell or share personal data.

Subscribe | View online | Editorial policy | Privacy policy | Contact

This week we feature articles from Hanane Dupouy, and Ilia Teimouri. 

Additional reading comes from Dries Buytaert, Thomas Claburn, Scott Brinker, and Hugo Lu.

News comes from Figma & Adobe, Axel Springer & OpenAI, Franz, and Bloomreach & Spryker.

All previous issues are available at https://gilbane.com/gilbane-advisor-index

---

Opinion / Analysis

Discover DocLLM: The new LLM from JPMorgan for working with complex documents

Lots going on in this area these days including similar approaches. Hanane Dupouy provides a useful summary, but you’ll also want to check out the paper and at least read the introduction. (5 min)

Summary: https://medium.com/@basics.machinelearning/discover-docllm-the-new-llm-from-jpmorgan-for-working-with-complex-documents-5f54ea287d52
Paper: https://arxiv.org/pdf/2401.00908.pdf

Illuminating insights: GPT extracts meaning from charts and tables

Ilia Teimouri covers the same topic but describes an application focused on using data from images in financial documents. Scroll down to see his impressive-looking results and explanation. If you’re technical there is a link to all his code on Github. (7 min)

https://towardsdatascience.com/illuminating-insights-gpt-extracts-meaning-from-charts-and-tables-a0b71c991d34

More Reading

• Good take… The new old: Jamstack and MACH’s journey towards traditional CMS concepts via Dries Buytaert
• The amazing turning point when martech systems complexity and martech UX complexity diverge Looks promising for users but of course depends largely on GenAI. via chiefmartec
• Finally… Google to start third-party cookie cull for 30 million Chrome users via The Register
• Where to process unstructured data in the pipeline and why… The unstructured data funnel via Towards Data Science

All Gilbane Advisor issues

---

Content technology news

Figma and Adobe abandon proposed merger

Fifteen months into the regulatory review process, Figma & Adobe no longer see a path toward regulatory approval of proposed acquisition.
https://www.figma.com/blog/figma-adobe-abandon-proposed-merger/

Franz Unveils AllegroGraph 8.0

Neuro-Symbolic AI Platform incorporates LLM components into SPARQL with vector generation and vector storage for Knowledge Graph solutions. 
https://allegrograph.com/new-allegrograph-v8-neuro-symbolic-ai-platform/

Axel Springer and OpenAI partner to deepen beneficial use of AI in journalism

ChatGPT users will get summaries of selected global news from Axel Springer’s media brands including POLITICO, BUSINESS INSIDER, BILD & WELT.
https://www.axelspringer.com/en/ax-press-release/axel-springer-and-openai-partner-to-deepen-beneficial-use-of-ai-in-journalism

Bloomreach partners with Spryker to expand composable commerce

Bloomreach users can leverage Spryker’s composable, API based functionalities to constantly improve their digital and marketplace experience.
https://www.bloomreach.com/en/news/2023/bloomreach-partners-with-spryker-to-unlock-a-new-level-of-composable-commerce ■ https://spryker.com

All content technology news

---

The Gilbane Advisor is authored by Frank Gilbane and is ad-free, cost-free, and curated for content, computing, web, data, and digital experience technology and information professionals. We publish recommended articles and content technology news weekly. We do not sell or share personal data.

Subscribe | View online | Editorial policy | Privacy policy | Contact

This week we feature articles from Kurt Cagle, and Mike Dillinger.

Additional reading comes from Michael Parekh, Jeremy Arancio, Sam Shedden, and Mike Loukides.

News comes from Neoj4, Acquia, Ontotext, and Fivetran.

Reminder: We don’t usually publish in December. We’ll be back in January. Hope you all have healthy and happy holidays! 

All previous issues are available at https://gilbane.com/gilbane-advisor-index

---

Opinion / Analysis

Do you really NEED that LLM?

Kurt Cagle explains why large language models (LLMs), specialized or not, may be overkill for many applications, describes some of the characteristics of those applications, and suggests knowledge graphs and other existing technologies that may be preferable. (6 min)

https://metaphoricalweb.substack.com/p/do-you-really-need-that-llm?

Knowledge graphs or plain ol’ snake oil: How would you know?

Mike Dillinger addresses the question of what a knowledge graph is, is not, and how it is different from related technologies and resources.

“The overarching goal of knowledge graphs is to store and reproduce conceptual information in a machine-accessible way, as verifiable facts linked to objective data — not only as statements or opinions…” (11 min)

https://medium.com/@mike.dillinger/knowledge-graphs-or-plain-ol-snake-oil-how-would-you-know-0e810a868d28

More Reading

• Why are AI products doomed to fail? The mistakes businesses have been making. via Towards AI
• Q*… AI: Doing the reasoning math reliably via Michael Parekh
• Model Collapse: An experiment showing that collapse is not unique to AI via OºReilly Radar
• OpenAI offers ChatGPT voice chat to all free users on mobile Not recommended while driving. via readwrite

All Gilbane Advisor issues

---

Content technology news

Neo4j collaborating with AWS to enhance generative AI results

The agreement includes integration with Amazon Bedrock for enterprise generative AI outcomes that are more accurate, transparent, and explainable.
https://neo4j.com/press-releases/neo4j-aws-bedrock-integration/

Acquia Site Studio introduces capabilities to create Drupal and headless applications from a single platform

Now developers and marketers can create and manage digital experiences for traditional Drupal websites and headless applications using the same content.
https://www.acquia.com/newsroom/press-releases/site-studio-drupal-headless

Ontotext and TopQuadrant partner to accelerate adoption of graph and semantic technologies

Ontotext clients gain a knowledge graph creation and curation tool, while TopQuadrant clients get improved scalability, usability, performance.
https://www.ontotext.com/company/news/ontotext-and-topquadrant-a-powerful-partnership/

Fivetran deepens relationship with Microsoft; adds new data lake destinations

Fivetran supports Microsoft OneLake through integration with Microsoft Fabric and Delta Lake on Azure Data Lake Storage Gen2.
https://www.fivetran.com/blog/fivetran-supports-microsoft-onelake-as-a-destination-through-integration-with-microsoft-fabric

All content technology news

---

The Gilbane Advisor is authored by Frank Gilbane and is ad-free, cost-free, and curated for content, computing, web, data, and digital experience technology and information professionals. We publish recommended articles and content technology news weekly. We do not sell or share personal data.

Subscribe | View online | Editorial policy | Privacy policy | Contact