About
Activity
1K followers
-
Brian King reposted thisBrian King reposted thisMany people in Silicon Valley believe that artificial general intelligence — a technology that may never materialize — will create a “permanent underclass.” That belief is fueling a last chopper out of Saigon mentality where people see a limited window to build wealth before AI and robotics fully replace human labor. I believe this is a consensual hallucination of techno-narcissists over-indexing on the rapid advances in AI capabilities while completely ignoring … everything else.
-
Brian King reposted thisBrian King reposted thisToday we’re releasing 𝗘𝗠𝗢, a new mixture-of-experts (MoE) model trained so modular structure emerges from data without human-defined priors. 👇 Most LLMs are trained and deployed as one monolithic system, even when an application only needs a narrow capability like code or math. MoEs seem to break this pattern by using only a few experts per token. But across a full task, standard MoEs still rely on many experts. EMO’s key idea: use each training document as a weak signal for shared context. Instead of letting every token route independently, EMO restricts tokens from the same document to a shared expert pool, encouraging experts to organize around coherent domains. EMO’s expert clusters look very different from a traditional MoE—they organize around semantic domains like health, news, politics, & film/music. Traditional MoEs often cluster around surface patterns like prepositions and articles, making selective expert use tougher. EMO is a 1B-active, 14B-total MoE trained on 1T tokens with 8 of 128 experts active per token. Without any subsequent fine-tuning, EMO remains robust when only a subset of experts is kept: with 25% of experts, it loses ~1 percentage point in overall performance; with 12.5%, it drops ~3 points. Standard MoEs degrade sharply. We experiment on a smaller 130B token setting, where we show EMO subsets also match or outperform memory-matched models trained from scratch. Instead of training many separate small models for fixed memory budgets, one EMO model can provide many domain-specific expert subsets. We're releasing EMO, a matched standard-MoE baseline, and training code to help the community study modularity & expert selection: 🧠 Models: https://lnkd.in/e6Jd2ugi 📝 Blog: https://lnkd.in/e7kramun 📄 Tech report: https://lnkd.in/eWCww2jg 📊 Visualization: https://lnkd.in/e9p6Wy28
-
Brian King reposted thisBrian King reposted thisKnowledge base management is becoming one of the biggest hidden blockers to AI adoption. We’re seeing this across the organizations we work with. Teams are excited about AI. They want better assistants, better internal search, faster onboarding, smarter sales enablement, cleaner customer support, better content reuse. But then they run into the same issue: The knowledge is everywhere. It lives in Google Drive, SharePoint, Notion, Slack, Gong, Salesforce, Confluence, PDFs, decks, spreadsheets, call notes, and people’s heads. And AI is only as useful as the context it can access, understand, and trust. That’s why I was so interested to see the same pattern show up in Scott Brinker and Frans Riemersma’s latest State of Martech 2026 report. In the survey, “Knowledge & Documentation” had the highest activity across both buying AI tools and building internal AI solutions. It was the most bought use case, and also the most built use case. The report also makes the point that the stack is moving toward “context engineering,” where the work is not just prompting an AI tool better, but assembling the right data, content, tools, permissions, and instructions so AI can act effectively in a specific situation. Companies know this matters. This is why RAG, knowledge architecture, and AI governance are becoming connective tissue across the enterprise. Customer support, sales enablement, marketing, product, HR, finance, and operations may look like different AI use cases on the surface but they all need to organize, govern, and activate their knowledge. We are currently doing some research in this area with companies experiencing this pain point. If you are actively trying to manage this within your company we'd love to learn more and are happy to share what we're seeing work today.
-
Brian King reposted thisBrian King reposted thisScott Brinker's "The State of Martech 2026" report dropped yesterday. A great many recaps this week will lead with the same point: 15,505 products in the landscape, effectively flat for the first time in 15 years. That's the peak martech story, and it's worth reading. The number more should be talking about is on page 34. 91% of marketing organizations are producing content with AI, and only 37% are doing anything to verify it. In healthcare marketing, where the content touches disease claims and patient populations, it isn't an operational inconvenience. It's a trust problem, and governance isn't optional when patients are on the other end of that content. Talked more about this in the blog post this morning... Thanks as always for the fascinating and insightful read, Scott. https://lnkd.in/gc-nbcUY
-
Brian King reposted thisBrian King reposted thisI meet 4–5 customers every day, and almost all of them assume: 𝗔𝗜-𝗻𝗮𝘁𝗶𝘃𝗲 = 𝗲𝘅𝗶𝘀𝘁𝗶𝗻𝗴 𝘀𝗼𝗳𝘁𝘄𝗮𝗿𝗲 𝘄𝗶𝘁𝗵 𝗔𝗜 𝗮𝗱𝗱𝗲𝗱 𝗼𝗻 𝘁𝗼𝗽. It’s not. That framing keeps you stuck evaluating the wrong thing. The question is no longer: “𝗗𝗼𝗲𝘀 𝘁𝗵𝗶𝘀 𝗽𝗿𝗼𝗱𝘂𝗰𝘁 𝗵𝗮𝘃𝗲 𝗔𝗜?” It’s: “𝗪𝗵𝗮𝘁 𝘄𝗼𝗿𝗸 𝗱𝗼𝗲𝘀 𝘁𝗵𝗶𝘀 𝘀𝘆𝘀𝘁𝗲𝗺 𝘁𝗮𝗸𝗲 𝗼𝗳𝗳 𝗺𝘆 𝘁𝗲𝗮𝗺—𝗮𝗻𝗱 𝘄𝗵𝗮𝘁 𝗱𝗼𝗲𝘀 𝘁𝗵𝗮𝘁 𝗱𝗼 𝘁𝗼 𝗰𝗼𝘀𝘁, 𝘀𝗽𝗲𝗲𝗱, 𝗮𝗻𝗱 𝗿𝗶𝘀𝗸?” Most enterprise software we buy is still a system of record: capture → store → report Even with AI layered in, it rarely moves the metrics that matter: • cost to operate • cycle times • financial/control risk That’s the gap. What’s emerging instead are 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 𝗼𝗳 𝗮𝗰𝘁𝗶𝗼𝗻 - software that executes work: • Timekeeping that improves revenue capture without manual entry • OOO catch-up that reduces decision latency • Procurement flows that enforce spend controls in real time When you evaluate through that lens, the conversation changes: • How many FTE hours does this remove? • Where does it reduce leakage or enforce controls? • What decisions can it make autonomously and how are they governed? • Is there an audit trail that stands up to scrutiny? • Does it simplify the stack or add another layer? This isn’t a feature discussion. It’s an operating model decision: 𝗼𝗽𝘁𝗶𝗺𝗶𝘇𝗲 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 𝗼𝗳 𝗿𝗲𝗰𝗼𝗿𝗱 𝗼𝗿 𝗶𝗻𝘃𝗲𝘀𝘁 𝗶𝗻 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 𝘁𝗵𝗮𝘁 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗲𝘅𝗲𝗰𝘂𝘁𝗲 𝘄𝗼𝗿𝗸. That distinction shows up directly in margin, speed, and control. I wrote a deeper breakdown here: https://lnkd.in/gwPacmkz #DigitalTransformation #SystemsOfAction #FinOps #AITransformation #AI #AIAgents #EnterpriseAI #Bytemethod #Dexian
-
Brian King reposted thisBrian King reposted thisMAHESH YADAV went from PM at Microsoft to $1.4M AI PM at Google in 8 years. His definition of builder PM is the cleanest one I've heard. "A builder PM talks to customers, figures out what needs to be built, and builds the first version to 10 customers, without talking to any developer at all." That's the bar from his episode of The Growth Podcast. Not a PM who codes. Not a PM who knows Claude Code. A PM who removes the handoff between insight and shipped product. His comp arc tracks the bar exactly. $120K starting at Microsoft. $360-400K as Senior PM there. $600-700K as AI PM at Meta. $1.3-1.4M as AI PM at Google. He says his peers in similar AI PM roles at Nvidia are at $2-2.5M today. The trajectory roughly doubles every two years because the work compresses. Traditional PM time to first prototype runs 3-6 months. Builder PM time to first 10 paying users runs 3-4 weeks. The handoff is the bottleneck the market is repricing. There are two paths inside the same skill set, and they go to different places. Path 1 builds customer-facing product without engineering handoffs. Talks to customers, builds in Claude Code, ships to 10 paying users, raises a seed. This is the founding PM at an AI startup. Mahesh is one example. Small group, real outcomes. Path 2 builds internal agents that automate the PM's own work. PRD reviewers, competitive intel runners, learner loops on top of each. You stay in your current org and ship three times the output of every other PM in your skip-level review. The promotion comes from the inside. The skills overlap. The trajectory does not. Most readers' actual lives map to Path 2. You have a mortgage. Your spouse's job anchors your geography. You can't take an 80% pay cut for 18 months on the chance the AI startup you join hits product-market fit. The same builder skills that get a Path 1 PM the founding role get a Path 2 PM promoted on the inside. The first ten weeks are identical for both. n8n for 2-3 weeks to learn the four components of an agent. Claude Code for 4-6 weeks building skills that automate weekly PM work. Then enterprise agent tooling for delegating complete jobs. The fork comes around month four. Path 1 ships prototypes to external customers. Path 2 ships internal agents. If you're choosing this quarter, the question worth answering is which path your actual life looks like. Here's how to start either one: 1. Practical AI agents for PMs: https://lnkd.in/eyamsW7a 2. Claude Code for PMs: https://lnkd.in/gwhgmSfw 3. PM Operating System: https://lnkd.in/gsH9RSfJ 4. AI prototyping tutorial: https://lnkd.in/eJujDhBV 5. Mahesh's episode: https://lnkd.in/geWmmZCF Pick one path and run. The market pays both. Picking neither is what the market stops paying. **** I’m teaching a live Builder PM course starting Monday. Join us: www.landpmjob.com
-
Brian King reposted thisBrian King reposted thisAt Glean, we reduced our own agent system prompt by 45%+. Not by doing less, but by moving capability into skills that load progressively rather than front-loading everything into the prompt. That's the idea behind search-first skill discovery. And it's one of the shifts we're seeing in harness design as context management becomes the central engineering challenge: - Programmatic Tool Calling (PTC): Instead of the model deciding which tool to call turn-by-turn through conversation, the agent writes and executes code to orchestrate tool calls directly. Loops, conditionals, and branching logic live in the code, not in the model's reasoning chain. - Sub-agents for context execution: Rather than one agent loop handling an entire task, work gets decomposed into separate agents, each running in its own isolated execution context with its own context window. Each sub-agent handles a scoped piece of the problem independently. - Compaction: Keep load-bearing conversation state, intent, decisions made, approaches that failed, and move raw intermediate outputs to summaries or the file system. - Search-first skill discovery: Instead of loading every tool's full schema upfront into the context window, the agent first searches to find which skills are relevant, loads the descriptions of the ones it needs, then fetches the full schemas only at execution time. As long-running agents take off, context scales. The job of the harness becomes keeping the agent focused on the right context at the right time. That's why the harness has become a distributed context management system. https://lnkd.in/eWjxe4-3
-
Brian King reposted thisBrian King reposted thisOne of the questions I often get from executives is: Can agents reliably get real work done, end to end? A lot of that answer depends on the harness around the model—the system that decides how to break a request into steps, which tools to call, what to remember, and when to stop. Models have a fixed attention span and the harness decides how it gets filled. Agents are now taking on longer-running, more complex work. To do that reliably and to completion, the harness itself has to be built to scale context. Giving agents access to more tools is what lets them act on enterprise context. But cramming more instructions and schemas into the system prompt isn’t the solution. A better approach is to treat the harness as a distributed context management system: a lightweight orchestrator agent that decides what should happen next, with scoped subagents around it, each only seeing the context it needs for its part of the job. In order to take real action across enterprise systems, agents will need to operate as a coordinated team of specialists rather than one loop trying to do everything. Context is the problem we've been solving at Glean since day one. We wrote up some of what we've learned from our 3rd iteration of our harness here: https://lnkd.in/guAukbwp
-
Brian King liked thisBrian King liked thisI’m on Planet Money today! I chatted with them about the predictability of job destruction, based on a paper where I evaluated employment forecasts in the Occupational Outlook Handbook. These were career guides put out by the BLS and distributed to colleges and career centers throughout the country since the 1940s. Most interesting to me, each issue contained employment predictions: how much will the demand for dentists, carpenters, and teachers change over the next 5-10 years? I found that the predictions were broadly accurate. Job changes have been foreseeable, at least historically. Throughout the project, I was really enchanted by the reports themselves. We have a lot of government researchers dedicated to demystifying the world for the common good and we’re lucky to have them. https://lnkd.in/ghF-_cX7Which jobs are future-proofed? : The Indicator from Planet MoneyWhich jobs are future-proofed? : The Indicator from Planet Money
-
Brian King liked thisBrian King liked thisMeet your new teammates. Lytho AI Teammates help you build compliance into every asset, not chase it at the end. - Automated compliance checks - On-brand content from the start - Less manual work, more control See it in action in our upcoming live webinar. Save your spot: https://hubs.li/Q04dJ7bW0
-
Brian King liked thisBrian King liked thisWhat if every asset in your DAM was accurate, compliant, and ready to use? Without the right structure, metadata, and governance, even the best systems fall short. Lytho’s AI for DAM Bundle changes that. Powered by AI Teammates, it applies brand and regulatory standards, surfaces issues early, and ensures every asset is governed before it ever enters your system — so your team can trust what they find and move faster with confidence. Explore the DAM Bundle: https://hubs.li/Q04cTplF0 #DigitalAssetManagement #ContentOperations #BrandGovernance #Compliance #AI #CreativeOps #MarTech
-
Brian King liked thisBrian King liked thisOpenAI finally confirmed it... You need bidirectional token classification models for efficient NER. The bidirectional lobby is growing. Back to the main news — OpenAI released a new open-source PII redaction model. It supports the identification of: • account number • private address • private email • private person • private phone • private url • private date • secret The most interesting part for me is its backbone, a 1.5B encoder with 50M active parameters: Architecturally, it's essentially a scaled-down gpt-oss. 8 pre-norm transformer blocks, d_model=640, grouped-query attention (14 query heads over 2 KV heads, so 7 queries per KV head), and a sparse MoE feed-forward with 128 experts and top-4 routing per token — that's where the ~30x gap between total and active params comes from. It was pretrained autoregressively, then converted into a bidirectional classifier by swapping the LM head for a token-classification head over a BIOES label scheme, and switching to bidirectional banded attention with band size 128 (effective 257-token window). At inference, it runs constrained Viterbi decoding instead of per-token argmax, so span boundaries stay coherent in messy text. The pattern right now is clear — distill decoders into smaller bidirectional encoders and point them at tasks that make sense on the edge: PII redaction, classification, information extraction. One caveat: this OpenAI model isn't zero-shot. If you need to extend entity types to your custom ones, try these: https://lnkd.in/dbuAN-U4 OpenAI model: https://lnkd.in/dYsBP8yU
-
Brian King liked thisBrian King liked thisIf you want employees to stick around and give their best, make sure they have meaningful work, managers and colleagues who value and trust them, and opportunities to advance their careers. https://s.hbr.org/4sv6rfJ
-
Brian King liked thisBrian King liked this💡🤖 Transforming marketing at Microsoft with AI‑powered agents 💡🤖 Our Azure AI marketing team is using Microsoft Foundry to build agents that apply intelligence on tap, boosting their creativity, accelerating execution, and amplifying human impact across Microsoft. Our teams have been deeply involved in shaping how these agents are designed, deployed, and integrated into real marketing workflows. The story highlights how this work is helping marketers automate repetitive tasks, generate insights faster, improve content quality, and free up time for higher‑value strategic work. Here’s what we’re learning as we build and scale these agents: ⚡ Accelerate content workflows: Agents help teams move from idea to execution with greater speed and consistency. 🧠 Unlock deeper insights: AI‑driven analysis surfaces patterns and opportunities humans can act on quickly. 🤝 Strengthen collaboration: Shared agent tools help teams coordinate, align, and deliver more effectively. 🔁 Iterate with real‑world feedback: Continuous refinement ensures agents stay relevant, safe, and high‑impact. 📈 Amplify human creativity: Agents handle the heavy lift so people can focus on strategy, storytelling, and innovation. Why this matters for IT leaders Marketing is just one example, but it shows how enterprise teams can use AI agents to modernize operations, improve decision‑making, and scale impact across the business. Read the full Inside Track story: https://msft.it/6042v6rd0 #MicrosoftDigital #CustomerZero #AzureAI #Marketing #MicrosoftFoundry #Foundry #MicrosoftTransforming the marketing function at Microsoft with AI - Inside Track BlogTransforming the marketing function at Microsoft with AI - Inside Track Blog
-
Brian King liked thisBrian King liked thisThere’s no place like Knowledge. 💚 See you next year! #Know26
-
Brian King liked thisBrian King liked thisComputer-use agents need a Docker moment, not another demo 🖥️ Cua is one of the more interesting open-source projects I’ve seen in this space because it is working on the bit everyone quietly needs before desktop agents become useful: safe, repeatable environments. Everyone likes the demo where an agent opens an app, clicks around, tests a UI, fills in a form, fixes a bug and carries on. But demos hide the awkward questions. Where does the agent run? Can it destroy local files? Can you replay what it did? Can you run the same task twice? Can you run ten agents at once without losing control of your own machine? Cua is trying to package that layer properly. → Sandboxes for agents across Linux, macOS, Windows and Android → A Python SDK for shell, screenshots, mouse, keyboard and mobile gestures → CuaBot for running coding agents like Claude Code, Codex, Gemini CLI, OpenClaw and Aider inside sandboxed desktops → Cua-Bench for evaluating agents against OSWorld, ScreenSpot, Windows Arena and custom tasks → Cua Driver for background macOS control, so an agent can drive real apps without stealing your cursor or focus That last one is the part that made me stop. Most "computer use" tools still assume the agent takes over the machine like a remote support session. Fine for a demo. Annoying for actual work. Cua's bet seems different: agents need their own desktops, their own traces, their own sandboxes, and sometimes their own background control plane. That is less flashy than a model launch, but probably more important for real adoption. The repo is already around 14.5k stars, and it is MIT licensed. There is a cloud product too, but I like that the core direction is open infrastructure rather than another closed "operator" wrapper. Still early. Still risky. Still very easy to imagine agents clicking the wrong thing with total confidence. That is the less glamorous part of the stack, but it may decide whether computer-use agents become useful outside demos. Better agents will matter, but better agent operating environments might matter more. Are you building anything with computer-use agents yet? Would you let one operate your real desktop in the background, or keep it boxed inside a sandbox? https://lnkd.in/eVXTVf6k #AI #AIAgents #OpenSource #DeveloperTools #ComputerUseGitHub - trycua/cua: Open-source infrastructure for Computer-Use Agents. Sandboxes, SDKs, and benchmarks to train and evaluate AI agents that can control full desktops (macOS, Linux, Windows).GitHub - trycua/cua: Open-source infrastructure for Computer-Use Agents. Sandboxes, SDKs, and benchmarks to train and evaluate AI agents that can control full desktops (macOS, Linux, Windows).
-
Brian King liked thisBrian King liked thisWhat if every asset was compliant before it ever reached review? With Lytho AI Reviewers, it is. By evaluating content against brand and regulatory standards during creation, AI catches issues early, applies rules consistently, and ensures every asset is governed before it moves forward. Less rework. Fewer delays. Compliance at scale. Learn more: https://hubs.li/Q04dJbgd0 #ContentOperations #Compliance #BrandGovernance #AI #CreativeOps #MarTech
Experience
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Education
-
Bowling Green State University
-
-
Activities and Societies: Digital Media Research Group [DMRG], Semester abroad in Europe (Salzburg, Austria) completing R&D, Graduate Assistantship (Researcher)
- 3.62 cumulative GPA
- http://bit.ly/dM1Ks - Thesis Document - Commercial Media Viewing Habits: Digital Immigrants Versus Digital Natives -
-
-
Volunteer Experience
Projects
-
Product Owner/Product Manager DockDogs Event Management System
-
See projectChallenges: DockDogs (a canine aquatics event company where dogs jump off a dock into a large pool compete for the distance of their jump, speed of their swim, or height of their jump) was handling event registrations via manual submissions via PDFs sent in via email or via fax. This data was entered manually into a centralized, proprietary, offline and deprecated system. DockDogs was growing in popularity and was often conducting 5 or more events in different cities/states or even countries…
Challenges: DockDogs (a canine aquatics event company where dogs jump off a dock into a large pool compete for the distance of their jump, speed of their swim, or height of their jump) was handling event registrations via manual submissions via PDFs sent in via email or via fax. This data was entered manually into a centralized, proprietary, offline and deprecated system. DockDogs was growing in popularity and was often conducting 5 or more events in different cities/states or even countries during a given weekend. The database where event registrations were being entered also handled all event scores, rankings, and titles earned by participants in the DockDogs competitions.
After each and every event the database had to be manually manipulated and merged and hours of manual tasks had to take place to produce updated event results, rankings, and titles for posting on the dockdogs.com website.
Result: I primarily served as a Product Owner/Product Manager for DockDogs with a team of overseas developers to streamline the event discovery, registration, management, and execution of events into a single web application that integrated into the WordPress website for dockdogs.com which I had built. While working as a Product Owner on this project I participated in an Agile (SCRUM) team and was responsible for writing user stories, performing user acceptance testing, QA/QC testing, and writing documentation for participants of DockDogs events as well as event staff entering scores and processing the events live. -
Campfire Distance Learning & Reflective Teaching Application
-
Challenge: Non-traditional students were participating in a distance learning Master of Education program and needed a way to receive constructive feedback and criticism of their teaching styles. We needed to produce a tool that allowed non-traditional students to present their lesson plans in an interactive and asynchronous environment where their peers and instructors could provide feedback from recorded video sessions. The goal of this was to enhance the students teaching skills, especially…
Challenge: Non-traditional students were participating in a distance learning Master of Education program and needed a way to receive constructive feedback and criticism of their teaching styles. We needed to produce a tool that allowed non-traditional students to present their lesson plans in an interactive and asynchronous environment where their peers and instructors could provide feedback from recorded video sessions. The goal of this was to enhance the students teaching skills, especially for online distance learning.
Result: I served primarily as the IT Project Manager for my team as well as contributing to the codebase and QA/QC of the application we developed. The application we developed allowed for students in the Master of Education program to load Microsoft PowerPoint presentations and video files into the web application along with their webcam and record their lessons. Their peers and Professor would receive notifications via their Moodle account and be able to provide text, audio, and video annotations at any point during the recorded (or live) lesson being presented. The viewers were also provided with a UI overlay that they could designate positive or negative sections of the lesson. The presenter, their peers, and Professor could go back and view these comments and subsequently better tailor their lessons for increased engagement and better delivery.
Languages
-
English
Native or bilingual proficiency
-
German
Elementary proficiency
Recommendations received
10 people have recommended Brian
Join now to viewOther similar profiles
Explore more posts
-
Luigi Bai
For those of you using #Drupal (bespoke site or DLAW/Urban Insight's OpenAdvocate), there's a new initiative to pull together the various integrations with generative AI tools: https://lnkd.in/epRP7njP. I'm not sure how well this maps onto how folks are using Drupal to deliver #legalInformation to the public, but there may be tools there that may already implement ideas you might have around chatbots, navigation, and even content generation. Apparently the integrations do now or will let you use whatever model you prefer. I'll be honest, I have mixed feelings about promoting generative AI use in our justice space. GenAI models are trained and used in data centers that use fresh water for cooling (often in places where it's scarce)[1][2][13], require megawatts of "always-on" electricity (which is already raising prices for consumers[3][4]), typically have on-site generators that generate noise[5] and particulate[6] pollution, and which are sited in locations that disproportionately impact our marginalized communities. For example, it would be helpful to know the actual impact of a proposed data center in north Houston. In addition, the use of the training data, when obtained without permission, arguably violates creator's rights[7]. The input data need to be curated by humans using extractive labor practices[8][9][10]; they tend to be inadequately compensated for this skilled work and don't have access to mental health support they need because they spend all day filtering out works that include CSAM and other violence. After a model is trained, "reinforcement learning through human feedback" (RLHF) has similar issues[11]. It's hard for me to see using generative AI tools as a "net positive" in terms of justice for our communities[12], at least in the general case. There may be limited uses where the benefits outweigh the costs; maybe some of the Drupal modules above implement these. And it may be that the costs are reduced over time - in which case I would wait until they actually are, and not proceed on the assumption they will. In the meantime, I recommend exploring and implementing other automation that doesn't include genAI. If you're not part of a case-handling organization, I recommend you reach out to a local legal aid organization and ask if they're seeing (and helping clients with) data center or related labor practice issues in their service area. Citations in the first two comments. Thanks to The Drop Times for the information about the Drupal AI project. #genAI #legalTech #justice #accessToJustice
2 Comments -
The Drop Times
Tutorial highlight 💡 In Part 2 of his series, "Unmanaged Files in Drupal: Building a Random File Handler" Jeff G. shows how to build a lightweight service that fetches random unmanaged files—no file entities, just pure flexibility. Perfect for rotating images, quick demos, or handling large asset sets. Read now Check it now : https://lnkd.in/g4aQRmXx
-
DrupalFit
In Conversation with DrupalFit – Featuring Mike Gifford, CPWA! We had the pleasure of sitting down with Mike Gifford to discuss key topics around accessibility. As the Open Standards & Practices Lead at CivicActions, Drupal Core Accessibility Maintainer, and a W3C Invited Expert, he shared valuable insights on why accessibility must be a core priority. He emphasized that accessible websites benefit everyone and should never be treated as a checklist—it must be an essential part of the process. Watch the video for detailed insights!👇 #accessibility #websiteaccessibility #experts #session #insights #ai Harsh Phartyal Pritam Prasun Pritish Kumar Rohit Singh
-
CivicActions
Mike Gifford joined DrupalFit to talk about what accessibility really means in practice. As CivicActions’ Open Standards & Practices Lead, Drupal Core Accessibility Maintainer, and W3C Invited Expert, Mike makes one thing clear: accessibility isn’t a checklist — it’s foundational to building effective digital services. Watch the conversation below. 👇 #a11y #accessibility #Drupal #DigitalCommons #FOSS #AI Harsh Phartyal
-
Morpht
DrupalSouth has launched a new blog series exploring how Drupal continues to innovate in the open-source ecosystem. In the first article, Morpht's co-founder, Murray Woodman shares insights on AI, integrations, and innovation, and how Drupal is keeping pace. 👉 Read more: https://lnkd.in/gfcs-cPj 🔗 DrupalSouth feature: https://lnkd.in/gEAzyBMY #DrupalSouth #Drupal #DrupalAI #Innovation #OpenSource #Morpht
-
Salsa Digital
Drupal isn’t just open source, it’s enterprise ready. The latest article from DrupalSouth challenges the old assumption that enterprise means proprietary. Instead it shows Drupal powering high-stakes government platforms and major global brands with reliability, scalability and security. As our own Kurt Foster, CTO at Salsa Digital, put it: “During COVID-19 — a period of intense, daily public updates — we didn’t have a single outage. The cache management system meant we could handle massive traffic spikes without overwhelming the infrastructure.” That track record matters for decision makers choosing technology for complex ecosystems. Drupal’s flexibility, composable architecture and global community make it a strong contender in enterprise RFPs. Read the full article on DrupalSouth for more insights into how Drupal meets enterprise requirements in live environments. 👉 https://lnkd.in/g9KdfVri #Drupal #OpenSource #EnterpriseTech #GovTech #DigitalTransformation
-
Metadrop
Thrilled to see our 'Content First' module, featured as 'Module of the Week' on the Talking Drupal podcast! This module is a perfect example of our core mission at Metadrop: creating powerful, open-source tools that solve real-world problems and empower content teams. You'll find a sharp AI/SEO tip. #Drupal #OpenSource #ContentStrategy #UX
-
Numble
Choosing a CMS? The wrong one might already be draining your budget. We see it all the time: companies stuck in overly complex setups, heavy maintenance cycles, and frustrated editors just because “Drupal felt like the safe option.” Truth is: it depends on your context. If you’re working in a .NET environment, want smooth editing, and need real integration power, Umbraco is probably the smarter fit. We broke it down slide by slide. Swipe through → and let us know: what’s your CMS story? #Umbraco #Drupal #CMSChoice #CustomDevelopment #WebPlatforms #SystemIntegration #DigitalExperience #Numble
2 Comments -
Eric Jordan
Copilot isn’t just for writing emails. It’s your shortcut to operational clarity. One of our clients used it to run internal surveys—and uncovered: → Duplicate software spend → Tools no one was using → Broken internal processes End result? 17% savings in one quarter. That’s just one use case. We’ve got six more for you here - https://bit.ly/4kXDapA #GenerativeAI #AIInnovation #Copilot
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content