Kole Sam @Kole_Sam

@eliana_jordan Facts

This week I’m attending the AI4 Conference in Las Vegas, and one session really got me thinking: Rewriting Student Records Through Blockchain Innovation. As AI gets better at creating realistic documents, including transcripts with official looking seals and signatures, how will we know what’s authentic? One answer might be blockchain. The demo showed how students own their official transcript in a secure digital wallet and can instantly share it with employers or other institutions. Instead of trusting a PDF, the recipient verifies it against the blockchain. If someone edits or tampers with the document, verification fails immediately. Beyond preventing fraud, I like that it also puts students in control of their own records instead of making them wait on an institution every time they need to share a transcript. It made me wonder whether this is where higher education is headed. Not just for transcripts, but for diplomas, certificates, and other academic credentials. What do you think? Is blockchain the missing piece for verifying documents in the age of AI?

AI has become part of my daily workflow. Today, this is what I had it do. I used Claude as an AI coworker to generate a usage report with graphs, tables, financial projections, and analysis. The raw data lived across multiple analytics screens inside a web application. Normally, that would mean clicking through pages, pulling numbers, exporting data, cleaning it up, building charts, formatting a document, and saving it somewhere useful. Instead, I asked Claude, using the Chrome extension, to navigate the web app, find the right usage data, aggregate it, create the projections, build the report, apply our organization’s brand colors and logo, and save the finished document into a folder on my computer. While Claude gathered the data, built the report, and saved the final document, I spent that time working on other higher priority work. When it was done, I opened the document, made a few light edits, and moved on. That is the shift I keep seeing with AI at work. It is no longer just answering questions. It is becoming a coworker that can work across tools, gather context, produce real deliverables, and give time back to people. This is what gets me excited about where work is headed. Not AI replacing human judgment, but AI taking on the manual execution layer so we can spend more time making decisions, thinking strategically, and driving outcomes.

This is getting interesting. A lot of the robot’s movement looks human-like.

ABC News @ABC

2 weeks ago

A San Francisco company says it is offering a new home-cleaning service using humanoid robots for a rate of $30 per hour. abcnews.link/ml2TgwG

254 375 1K 610K 504

@sumitroy2 Let’s see if Anthtopic signs

Who’s up next?

Sundar Pichai @sundarpichai

2 weeks ago

Very happy to support this on behalf of Google. We have long benefited from open source, are big contributors to open source and in fact have consistently made open weights models with Gemma available from @GoogleDeepMind @demishassabis . Onwards!

756 2K 23K 4.3M 2K

Let’s see if Anthropic signs too 👀

tae kim @firstadopter

3 weeks ago

OpenAI tells me they signed the Nvidia–Microsoft open-source/open-weights AI model letter today. What an amazing turn of events. Your move, Anthropic!

keyboard_arrow_left Previous keyboard_arrow_right Next

115 239 3K 540K 350

Jeez, things didn’t go as planned

Culture Crave 🍿 @CultureCrave

3 weeks ago

Robot powered by Qualcomm’s new AI chip dies mid-presentation 😭

2K 2K 15K 1.7M 3K

@matthew_d_green 👀

@sumitroy2 Very interesting

Sounds about right

@mkratsios47 Shocker

@attrc That’s even worse than I thought

@ClementDelangue @CommonSenseMars @OpenAI Interesting

The recent news coming from OpenAI about two of its AI models has really got me thinking. As AI becomes more capable, it’s becoming clear that it can also be used in harmful ways if we are not paying attention. This is a reminder that we need to stay aware of how quickly this technology is changing and what it could mean for cybersecurity. I’ll try to simplify what happened so it’s easy to understand. OpenAI recently carried out a security test by giving one of its newest AI models a cybersecurity challenge to see how good it was at hacking. The AI was placed inside a computer with no internet connection and was expected to solve the task without leaving that environment. Instead of solving the challenge the normal way, the AI searched for another option. It found a weakness in the system that allowed it to escape the restricted environment and connect to the internet. Once it was online, it looked for places that might have the information it needed. It identified Hugging Face, a platform used to store and share AI models and related files, as a possible source. The AI then found stolen credentials and combined them with another security flaw to gain access to Hugging Face’s servers. Once inside, it retrieved the information it needed to complete the challenge. Rather than solving the test as intended, it found a way to cheat. Hugging Face’s security team detected the activity, stopped the intrusion, and began investigating. OpenAI is also reviewing what happened. The part that stands out to me is that the AI was not directly instructed to do any of this. It chose its own path, found multiple ways around security, and carried out the entire sequence without a human telling it what to do. Another point that caught my attention is that Hugging Face said they initially used their own open source models to help respond to the threat. This was AI versus AI. That raises an important question. What happens if an organization does not have access to capable AI models that can defend against attacks like this? You have to fight fire with fire. Every organization that takes cybersecurity seriously should already be thinking about how they will defend against AI powered threats. That means equipping themselves with capable AI models, whether they are open or closed, and making them part of their security strategy. Now is the time to start!

Is this the future we’ll get used to, app interfaces replaced by CLI? What’s next, uber?

Andy Fang @andyfang

4 weeks ago

Today we're opening up the DoorDash CLI in limited beta. `dd-cli` lets you order DoorDash directly from your agent: search stores, find the best deals, check out, and more. Early access for US/Canadian macOS developers by waitlist. Excited to see what folks build!

1K 554 13K 5.3M 4K

@cdngdev Good stuff, this is pretty awesome

Interesting… great use of technology. I was wondering how they figured out that the Croatian player had a slight touch of the ball.

PFT Commenter @PFTCommenter

a month ago

This is a wild new feature- the chip in the ball graph that shows when it gets touched. the NBA is gona have this asap

682 903 18K 8.0M 3K

I have started using AI as my junior project manager. Claude, specifically. As a project and product portfolio manager, it is important for me to always have a pulse on my ongoing projects and products. I have already shared some of the ways I use AI for product-related work, and I will continue covering those in future posts. But for this one, I want to focus on the project management side. The shift is bigger than just saving time. I am now assigning AI some of the work I would have assigned to a project manager in the past. Because Claude has access to my project management tools, team communication, and project context, I can give it work in natural language and have it execute across the systems I already use. That changes the role of the project manager. Project managers will still need judgment, leadership, communication, prioritization, and stakeholder management. But the manual layer of project management is going to shrink. Taking notes. Transcribing meetings. Summarizing decisions. Creating action items. Updating project plans. Writing status reports. Flagging blockers. Identifying slipped tasks. Monitoring critical path risk. Surfacing resource conflicts before they become real problems. Those are all areas where AI can already help. In my own workflow, I invite AI into project activities from end to end. During meetings, AI captures the discussion, creates the summary, identifies decisions, pulls out action items, and flags blockers. After the meeting, if the project plan needs to be adjusted, I can ask Claude to update it. If new work needs to be added, I can ask Claude to create the tasks. When team members provide updates, AI can add those updates to the right work items in Jira or Asana. It can also generate daily and weekly progress reports, detect slipped tasks, and warn me when a resource looks overutilized before I assign more work to them. The part I find most interesting is the interface shift. I am using one AI chat interface instead of constantly jumping between multiple tools, screens, boards, filters, menus, and reports. Less clicking. Less searching. Less scrolling. Less copying information from one place to another. I do not think AI removes the need for project managers. I think it removes a lot of the low-value manual work that has kept project managers from spending more time on the work that actually matters. If I had this back in the day when I was a project manager, it would have freed me up to focus more on people, risks, decisions, alignment, and outcomes. That is where project management should have been focused all along.

An important part of UI/UX design is accessibility: making sure digital products are usable by everyone, including people with disabilities, people who are visually impaired, and users who rely on assistive technologies. Accessibility is not just a “nice to have.” When products do not meet accessibility standards, the result can be exclusionary design, poor user experience, and potential legal risk. More importantly, it means real people may be blocked from using something that should have been built with them in mind from the start. As we build more products with the help of AI, we should also be using AI to help us improve accessibility. There are so many nuances to consider: color contrast, keyboard navigation, screen reader behavior, focus states, semantic structure, WCAG guidelines, error messaging, form labels, and more. It is easy to miss issues when you are only reviewing a product visually. After completing a product recently, I asked Claude to run an accessibility review. It was able to spin up the site, navigate through the experience, click around, inspect different flows, and evaluate whether the product aligned with accessibility guidelines, including WCAG. The output was an extensive accessibility report that broke findings into clear categories: • Critical issues that needed to be fixed • Recommended improvements • Items that were already working well • Notes related to usability and accessibility best practices In the past, this type of review often required hiring a specialized accessibility consultant or relying on a third-party audit tool. Those are still valuable, especially for formal compliance work, but AI can now help teams catch accessibility issues much earlier in the product development process. That is a big deal. AI should not replace inclusive design thinking or expert accessibility audits, but it can make accessibility testing faster, more frequent, and easier to build into our normal workflow. The real opportunity is this: use AI not just to build faster, but to build better products for more people.