Welcome to Issue #55

Coming Up:

Agent Orchestration: The Skill Every Professional Needs Next?

From Backyard to Breakthrough: Join the New Age of Citizen Science

Perplexity Labs: Your Next Project Manager Might Be AI

Plus, we discuss ChatGPT’s outage, Google’s hurricane hunter, the problem with our loud AI future, and all the news we found interesting this week.

It’s Sunday morning!

If you are a Dad, we hope you are getting a chance to relax and take in some amazing AI content in your favorite spot.

Happy Father’s Day!

The DAS Crew - Andy, Beth, Brian, Eran, Jyunmi, and Karl

Agent Orchestration: The Skill Every Professional Needs Next?

AI agents are moving from science fiction to business reality, and the new challenge is not building the smartest agent, but managing teams of agents to drive results. In this new era of “agent orchestration,” the most valuable skill may soon be knowing how to direct, evaluate, and coordinate dozens, or even hundreds, of AI workers, each specialized for different tasks.

The rise of agent orchestration means that soon, human domain expertise will no longer be a reliable moat. AI already democratizes access to knowledge that once required years of education or hard-won insider experience. And reasoning AI can put this into context for agent actions. In the past, experts protected their value by controlling scarce information. Now, the real differentiator is knowing how to turn that information into outcomes. Companies will need fewer traditional experts and more people who can structure tasks, manage workflows, and optimize the use of digital resources.

Orchestrators do not have to be technical experts in every domain. Instead, they look more like project managers, improvisational coaches, or air traffic controllers, the people who can coordinate complex systems and keep everything running smoothly. Experience across disciplines, resource management instincts, and the ability to see the big picture will become even more valuable. Some companies will struggle to adapt, weighed down by old processes and siloed teams. The fastest movers will empower employees to experiment with agents, track outcomes, and continuously improve workflows.

The stakes are high. First movers who build agent-native cultures can leap ahead, while laggards risk being left behind. In this landscape, your ability to manage AI teams could be the ultimate career advantage.

WHY IT MATTERS

Expertise Gets Democratized: AI unlocks access to specialized knowledge for everyone, shifting value from what you know to how well you can combine and apply knowledge and data to achieve results.

Project Management is Critical: Orchestrators who can direct AI teams, allocate digital resources, and coordinate across departments will drive the biggest gains.

Adapt or Fall Behind: Companies that fail to embrace agent-native processes risk becoming obsolete as more agile organizations leap ahead.

New Job Roles and Evaluation: Future hiring will favor people who bring both their own skills and a team of well-trained AI agents, shifting how talent is assessed and deployed.

Business Models Will Change: As agent networks become more common, businesses will reorganize around flexible, project-based teams, moving away from static hierarchies and job descriptions.

From Backyard to Breakthrough: Join the New Age of Citizen Science

AI is quietly transforming scientific research by inviting everyone into the process. Thanks to new AI-powered tools and platforms, people with little or no formal scientific training can now help collect, analyze, and even interpret data that drives major discoveries.

Platforms like Zooniverse have enabled over 1.6 million volunteers to work on projects from whale tracking to bird spotting, using AI to match human observations with global datasets. Other programs, like Mozilla’s Common Voice, let people contribute their own voices to improve speech recognition for underrepresented languages. AI’s pattern recognition and data crunching capabilities can now elevate simple observations, such as snapping a photo of a bird or sharing your tracked biometric data, into research-grade contributions.

When people collect and annotate data at scale, scientists gain access to observations and perspectives that would have been impossible before. AI makes it easy to process and verify these contributions, helping spot patterns, outliers, and opportunities for further research. Everyday individuals are already playing a role in personalized medicine, public health, and environmental science, with tools that empower anyone to ask new questions and push knowledge forward.

But this new age also comes with new responsibilities. While more eyes and minds make for better science, discernment and data quality matter more than ever. Platforms must help users understand what makes a meaningful contribution and guard against bad data or hasty conclusions.

WHY IT MATTERS

Science is Open to All: Anyone can now participate in research, regardless of location or academic background, thanks to accessible AI-powered tools.

Better Data, Faster Discovery: AI streamlines data collection and analysis, letting scientists crowdsource insights at unprecedented speed and scale.

Personal Impact Becomes Possible: Even small, personal data contributions like tracking your heart rate can add data to inform wider research, provided individuals keep agency over their data to protect their privacy when the data is used for research into wellness.

Grassroots Innovation: Local communities can use AI and shared data to solve problems that matter most to them, leading to more relevant and diverse scientific breakthroughs.

Discernment is Critical: The democratization of science raises the bar for data quality, transparency, and responsible interpretation. Platforms and individuals must stay vigilant.

Perplexity Labs: Your Next Project Manager Might Be AI

Perplexity Labs is making waves by promising more than just search or AI assistance. It acts like a “project operating system,” bringing together research, planning, design, code, and deliverables and it is all managed autonomously within one environment. Power users can create an entire workflow with just a prompt, then walk away as the system assembles research, writes reports, builds presentations, and organizes assets.

Unlike traditional AI chat tools that focus on single tasks, Perplexity Labs orchestrates complex, multi-step workflows. It now supports memory, tool use, and multi-agent collaboration, letting users spin up an “AI team” that acts on project goals. Users can upload their own data or select sources, and the platform generates everything from dashboards to slide decks to code. New integrations, including financial data and SEC filings, make it even more useful for real-world business analysis and planning.

The platform’s user interface and voice controls win praise for their simplicity and flexibility, and everything you generate such as slides, CSVs, code, and reports, lives in one organized space. That means less manual back-and-forth and more “set it and forget it” project management. Still, limitations remain: Perplexity can’t yet handle every kind of third-party integration, nor can it make phone calls or execute every workflow as robustly as some competitors.

As the AI “project OS” trend accelerates, major vendors will compete to add more features and integrations. Perplexity Labs shows what’s possible when you blend deep research, dynamic workflow management, and user-friendly design. Whether it will remain independent or be snapped up by a tech giant, its approach points to a future where project work is less about juggling apps and more about orchestrating intelligent systems that get the job done.

WHY IT MATTERS

Workflows Get Streamlined: Perplexity Labs lets users complete multi-step projects in one place, reducing the friction of switching between apps or tools.

AI as a Team, Not Just an Assistant: The platform treats projects as team efforts, running multiple agents to manage research, planning, analysis, and reporting.

Better Data, Better Results: Support for custom uploads and real-time data sources means users can build reports and dashboards grounded in current, relevant information.

Customization and Integration: While not perfect, Perplexity’s expanding integrations and export options point to a world where users expect AI to “just work” with the tools and data they already use.

Shifting Role of Human Oversight: As platforms get more autonomous, users will spend less time on manual tasks and more time checking and fine-tuning AI-generated output.

Just Jokes

A little Father’s Day humor about this week’s ChatGPT outage

Did you know?

Google DeepMind and Google Research have launched an experimental AI system that can generate 50 potential scenarios for tropical cyclone paths, sizes, and intensities up to 15 days in advance. That’s far more than typical weather models offer. This AI tool is being evaluated by the National Hurricane Center (NHC) to improve forecast precision and give communities more lead time before storms hit .

Early tests show it can predict cyclone tracks an average of 87 miles closer to the actual path than the European ECMWF model for storms in 2023 and 2024 .

The model delivers quicker results too: what used to take hours now takes about a minute on a single AI chip, making real-time forecasts more feasible

This innovation doesn’t aim to replace traditional forecasting methods. It enhances them. Forecasters gain a powerful new tool they can compare with physics-based models to improve warning systems. The collaboration marks the first time NHC has incorporated experimental AI guidance into its operations. Over the next hurricane season, we'll see how this AI assistant performs in real-world conditions.  Brian, who lives in Tampa, has his fingers crossed.

Source

This Week’s Conundrum
A difficult problem or question that doesn't have a clear or easy solution.

The Public Voice-AI Conundrum

Voice assistants already whisper through earbuds. Next they will speak back through lapel pins, car dashboards, café table speakers and everywhere a microphone can listen. Commutes may fill with overlapping requests for playlists, medical advice, or private confessions transcribed aloud by synthetic voices.

For some people, especially those who cannot type or read easily, this new layer of audible AI is liberation. Real-time help appears without screens or keyboards. But the same technology converts parks, trains, and waiting rooms into arenas of constant, half-private dialogue. Strangers involuntarily overhear health updates, passwords murmured too loudly, or intimate arguments with an algorithm that cannot blush.

Two opposing instincts surface:

• Accessibility and agency
  When a spoken interface removes barriers for the blind, the injured, the multitasking parent, it feels unjust to restrict it. A public ban on voice AI could silence the very people who most need it.

• Shared atmosphere and privacy
  Public life depends on a fragile agreement: we occupy the same air without hijacking each other’s attention. If every moment is filled with machine-mediated talk, public space becomes an involuntary feed of other people’s data, noise, and anxieties.

Neither instinct prevails without cost. Encouraging open voice AI risks eroding quiet, privacy, and the subtle social glue of respectful distance. Restricting it risks denying access, spontaneity, and the human right to be heard on equal footing.

The conundrum
As voice AI spills from headphones into the open, do we recalibrate public life to accept constant audible exchanges with machines even knowing it may fray the quiet fabric that lets strangers coexist, or do we safeguard shared silence and boundaries, knowing we are also muffling a technology that grants freedom to many who were previously unheard?

There is no stable compromise: whichever norm hardens will set the tone of every street, train, and café.

How should a society decide which kind of public space it wants to inhabit?

Want to go deeper on this conundrum?
Listen/watch our AI hosted episode

SAG-AFTRA Reaches Tentative Deal on AI Use in Video Games

The actors’ union reached a tentative agreement with major video game publishers after a lengthy strike. The deal reportedly puts strong AI guardrails in place to protect performers’ jobs in the AI era, though details are still emerging.

Deeper Insight:
The agreement could set a precedent for how AI is used across creative industries, particularly in protecting the rights and livelihoods of human performers as digital cloning and synthetic voices become more common.

OpenAI Launches o3 Pro, Slashes Price of o3, and Doubles Plus Plan Limits

OpenAI released o3 Pro, its new $200/month premium model, and cut the price of the standard o3 model by 80 percent. The Plus plan now allows for 200 o3 conversations per week, doubling its previous limit. API access is now more affordable, and o3’s performance is comparable to GPT-4.1.

Deeper Insight:
This pricing move makes advanced AI more accessible for developers and power users, but also signals a new, more tiered era for premium model access. The price drop could fuel wider adoption and experimentation across the AI ecosystem.

OpenAI Delays Its First Open-Source Model

OpenAI announced it will delay the release of its much-anticipated open model, originally set for June. The company says it needs more time to adjust some aspects of the release, likely involving additional safety guardrails.

Deeper Insight:
Open-source AI models allow for broader transparency, security, and community innovation. The delay points to the ongoing tension between speed and safety as model capabilities grow.

Mistral Unveils Magistral, a New Open-Source Reasoning Model

Mistral released Magistral, a new family of open-source models focused on reasoning. Available in “small” and “medium” versions (24 billion parameters and up), they are licensed under Apache 2.0 and accessible on Hugging Face and Mistral’s Le Chat platform.

Deeper Insight:
Magistral gives researchers and builders new options for running advanced models locally or integrating them into custom apps, while pushing the open-source ecosystem forward.

Meta Buys 49% Stake in Scale AI for Nearly $15 Billion

Meta acquired a 49 percent stake in Scale AI, the leading data labeling company, for $14.8 billion. Scale’s CEO will join Meta to run a new division, while Scale looks for a new leader. The deal values Scale at $30 billion.

Deeper Insight:
The move gives Meta a fast track to top-tier training data for AI while raising questions about competition and data access for Scale’s other customers, who may now find themselves sharing the same data pipeline source with a direct rival.

TSMC Reports Record Revenue Despite Tariff Uncertainty

Taiwan Semiconductor Manufacturing Company (TSMC) reported $11.6 billion in April revenue, a 48 percent year-over-year jump, despite looming tariffs on Taiwanese chip imports to the U.S. Much of the surge is attributed to companies stockpiling chips in advance of possible trade disruptions.

Deeper Insight:
TSMC remains a bottleneck for global AI and chip supply chains. While U.S. plants are ramping up, they cannot yet match the precision of Taiwan’s factories, leaving much of the industry exposed to geopolitical risk.

Apple’s Live Translation and On-Device AI Get Developer Upgrades

Apple previewed new live translation features and a foundational models framework for running AI on devices at WWDC. The upgraded AI runs locally using a quantized 3 billion-parameter model and supports use case-specific guided generation for apps.

Deeper Insight:
This shift toward more local AI models improves privacy, speed, and reliability for end users, while giving developers powerful new tools for Apple’s ecosystem.

Google Search Labs Expands AI Features With “Ask For Me”

Google’s new “Ask For Me” feature can call local businesses, ask about prices and availability, and deliver responses to the user via text or email. This is part of Google’s ongoing rollout of AI-powered search tools that directly answer user queries, potentially impacting publisher traffic and the traditional search ad business.

Deeper Insight:
Google’s AI push changes how users get information and how businesses reach customers. It may also further reduce traffic to publishers as AI summaries take center stage.

Google and UK Government Roll Out Gemini-Powered Extract Tool for Urban Planning

Google and the UK government launched Extract, a Gemini-powered AI tool that digitizes millions of planning documents, streamlining processes that used to take hours down to seconds. It converts handwritten notes, blurry maps, and other documents for faster infrastructure decision-making.

Deeper Insight:
AI’s ability to process and interpret messy real-world data can speed up housing and infrastructure projects, helping governments hit ambitious development goals.

Hugging Face Launches MCP Connector for Developers

Hugging Face announced a new MCP connector, allowing developers to integrate its model hub directly with popular coding tools. This simplifies access to models, datasets, and services across the AI ecosystem.

Deeper Insight:
Easy integration speeds up research and deployment, making advanced AI accessible to more developers and teams.

Univ. of Hong Kong Builds Super Agile Micro-Drone With AI-Assisted Flight

Researchers at the University of Hong Kong built a MAV (micro aerial vehicle) that can fly at 45 mph through dense forests, day or night, using a combination of lidar sensors and AI. The system generates both fastest and safest routes, helping the drone avoid obstacles as thin as a twig.

Deeper Insight:
This advance could improve search-and-rescue operations and pave the way for smarter, safer autonomous vehicles in complex environments.

Sam Altman Publishes “The Gentle Singularity”

OpenAI CEO Sam Altman published a new essay, “The Gentle Singularity,” where he predicts personalized AI will become as common as personal computers and shift value to people with ideas, not just technical skills.

Deeper Insight:
Altman’s vision points to a future where AI is easy to use and helps anyone build or create, emphasizing the growing importance of creativity and problem-solving over traditional coding.