“The New York Times begins publishing daily on the World Wide Web today, offering readers around the world immediate access to most of the daily newspaper’s contents,” stated the article, byPeter H. Lewis. “The electronic newspaper (address: http:/www.nytimes.com) is part of a strategy to extend the readership of The Times.”
Mr. Lewis had once owned that very URL.
In 1985, the Times editors A.M. Rosenthal and Arthur Gelb gathered a task force, which included Mr. Lewis, to work on a project called The New York Times in the Year 2000. Mr. Lewis this week shared the details of the project and his Times work in an email, from which much of this account is drawn.
Then an editor for the Science section and a personal computers columnist, Mr. Lewis recalled predicting that by the millennium, Times articles would be read on personal computer screens, in cyberspace.
“I recall Artie dismissing me with a wave,” Mr. Lewis wrote of Mr. Gelb.
Years later, the editor Bill Stockton, who Mr. Lewis said championed science and technology reporting, assigned Mr. Lewis to cover the “rise of the internet.”
At some point, “I asked permission to register a web domain for The Times, and was told no,” Mr. Lewis wrote in the email. “Several of us thought that was shortsighted.”
Another reporter, John Markoff, who had joined The Times to cover computer networking in 1988, had registered nyt.com some time after starting his role. (He used it for email; he did not set up a web page at the domain, so people got an error alert when they tried to visit it.) And Mr. Lewis scooped up nytimes.com around late 1993 or early 1994.
In mid-1995, Mr. Lewis got a call from Gordon Thompson, The Times’s manager of internet services, saying the paper wanted to go online as “The New York Times in Cyberspace” and needed the nytimes.com domain, which had won out in internal discussions over the shorter nyt.com URL registered by Mr. Markoff. (Per Mr. Markoff’s account, The Times thought the three-letter URL would be confused with the internet address of New York Telephone.)
In an email on Friday, Mr. Markoff said that he had registered the nyt.com domain before there were registration fees. But Mr. Lewis paid a $35 fee for nytimes.com. Mr. Lewis said he was happy to hand the domain over — as long as he was reimbursed. He transferred ownership of the URL to The Times, which activated the website on Jan. 19, 1996, from the Hippodrome office building in Manhattan.
A few days later, the website was live to the world. Mr. Lewis was not involved in the launch, though he covered the event for the newspaper.
As Mr. Markoff wrote in 2017, he eventually handed over nyt.com, on the condition that he get to keep his email, markoff@nyt.com, which he did until 2016. And today, both URLs send readers to The Times’s home page.
But there’s one problem: Mr. Lewis said he never received his $35 reimbursement.
Enterprise AI teams are discovering that purely agentic approaches (dynamically chaining LLM calls) don’t deliver the reliability needed for production systems.
The prompt-and-pray model—where business logic lives entirely in prompts—creates systems that are unreliable, inefficient, and impossible to maintain at scale.
A shift toward structured automation, which separates conversational ability from business logic execution, is needed for enterprise-grade reliability.
This approach delivers substantial benefits: consistent execution, lower costs, better security, and systems that can be maintained like traditional software.
Picture this: The current state of conversational AI is like a scene from Hieronymus Bosch’s Garden of Earthly Delights. At first glance, it’s mesmerizing—a paradise of potential. AI systems promise seamless conversations, intelligent agents, and effortless integration. But look closely and chaos emerges: a false paradise all along.
Your company’s AI assistant confidently tells a customer it’s processed their urgent withdrawal request—except it hasn’t, because it misinterpreted the API documentation. Or perhaps it cheerfully informs your CEO it’s archived those sensitive board documents—into entirely the wrong folder. These aren’t hypothetical scenarios; they’re the daily reality for organizations betting their operations on the prompt-and-pray approach to AI implementation.
Learn faster. Dig deeper. See farther.
The Evolution of Expectations
For years, the AI world was driven by scaling laws: the empirical observation that larger models and bigger datasets led to proportionally better performance. This fueled a belief that simply making models bigger would solve deeper issues like accuracy, understanding, and reasoning. However, there’s growing consensus that the era of scaling laws is coming to an end. Incremental gains are harder to achieve, and organizations betting on ever-more-powerful LLMs are beginning to see diminishing returns.
Against this backdrop, expectations for conversational AI have skyrocketed. Remember the simple chatbots of yesterday? They handled basic FAQs with preprogrammed responses. Today’s enterprises want AI systems that can:
Navigate complex workflows across multiple departments
Interface with hundreds of internal APIs and services
Handle sensitive operations with security and compliance in mind
Scale reliably across thousands of users and millions of interactions
However, it’s important to carve out what these systems are—and are not. When we talk about conversational AI, we’re referring to systems designed to have a conversation, orchestrate workflows, and make decisions in real time. These are systems that engage in conversations and integrate with APIs but don’t create stand-alone content like emails, presentations, or documents. Use cases like “write this email for me” and “create a deck for me” fall into content generation, which lies outside this scope. This distinction is critical because the challenges and solutions for conversational AI are unique to systems that operate in an interactive, real-time environment.
We’ve been told 2025 will be the Year of Agents, but at the same time there’s a growing consensus from the likes of Anthropic, Hugging Face, and other leading voices that complex workflows require more control than simply trusting an LLM to figure everything out.
The Prompt-and-Pray Problem
The standard playbook for many conversational AI implementations today looks something like this:
Collect relevant context and documentation
Craft a prompt explaining the task
Ask the LLM to generate a plan or response
Trust that it works as intended
This approach—which we call prompt and pray—seems attractive at first. It’s quick to implement and demos well. But it harbors serious issues that become apparent at scale:
Unreliability
Every interaction becomes a new opportunity for error. The same query can yield different results depending on how the model interprets the context that day. When dealing with enterprise workflows, this variability is unacceptable.
To get a sense of the unreliable nature of the prompt-and-pray approach, consider that Hugging Face reports the state of the art on function calling is well under 90% accurate. 90% accuracy for software will often be a deal-breaker, but the promise of agents rests on the ability to chain them together: Even five in a row will fail over 40% of the time!
Inefficiency
Dynamic generation of responses and plans is computationally expensive. Each interaction requires multiple API calls, token processing, and runtime decision-making. This translates to higher costs and slower response times.
Complexity
Debugging these systems is a nightmare. When an LLM doesn’t do what you want, your main recourse is to change the input. But the only way to know the impact that your change will have is trial and error. When your application comprises many steps, each of which uses the output from one LLM call as input for another, you are left sifting through chains of LLM reasoning, trying to understand why the model made certain decisions. Development velocity grinds to a halt.
Security
Letting LLMs make runtime decisions about business logic creates unnecessary risk. The OWASP AI Security & Privacy Guide specifically warns against “Excessive Agency”—giving AI systems too much autonomous decision-making power. Yet many current implementations do exactly that, exposing organizations to potential breaches and unintended outcomes.
A Better Way Forward: Structured Automation
The alternative isn’t to abandon AI’s capabilities but to harness them more intelligently through structured automation. Structured automation is a development approach that separates conversational AI’s natural language understanding from deterministic workflow execution. This means using LLMs to interpret user input and clarify what they want, while relying on predefined, testable workflows for critical operations. By separating these concerns, structured automation ensures that AI-powered systems are reliable, efficient, and maintainable.
This approach separates concerns that are often muddled in prompt-and-pray systems:
Understanding what the user wants: Use LLMs for their strength in understanding, manipulating, and producing natural language
Business logic execution: Rely on predefined, tested workflows for critical operations
State management: Maintain clear control over system state and transitions
The key principle is simple: Generate once, run reliably forever. Instead of having LLMs make runtime decisions about business logic, use them to help create robust, reusable workflows that can be tested, versioned, and maintained like traditional software.
By keeping the business logic separate from conversational capabilities, structured automation ensures that systems remain reliable, efficient, and secure. This approach also reinforces the boundary between generative conversational tasks (where the LLM thrives) and operational decision-making (which is best handled by deterministic, software-like processes).
By “predefined, tested workflows,” we mean creating workflows during the design phase, using AI to assist with ideas and patterns. These workflows are then implemented as traditional software, which can be tested, versioned, and maintained. This approach is well understood in software engineering and contrasts sharply with building agents that rely on runtime decisions—an inherently less reliable and harder-to-maintain model.
There’s a striking disconnect between the promise of fully autonomous agents and their presence in customer-facing deployments. This gap isn’t surprising when we examine the complexities involved. The reality is that successful deployments tend to favor a more constrained approach, and the reasons are illuminating.… Take Lindy.ai’s journey: they began with open-ended prompts, dreaming of fully autonomous agents. However, they discovered that reliability improved dramatically when they shifted to structured workflows. Similarly, Rexera found success by implementing decision trees for quality control, effectively constraining their agents’ decision space to improve predictability and reliability.
The prompt-and-pray approach is tempting because it demos well and feels fast. But beneath the surface, it’s a patchwork of brittle improvisation and runaway costs. The antidote isn’t abandoning the promise of AI—it’s designing systems with a clear separation of concerns: conversational fluency handled by LLMs, business logic powered by structured workflows.
What Does Structured Automation Look Like in Practice?
Consider a typical customer support scenario: A customer messages your AI assistant saying, “Hey, you messed up my order!”
The LLM interprets the user’s message, asking clarifying questions like “What’s missing from your order?”
Having received the relevant details, the structured workflow queries backend data to determine the issue: Were items shipped separately? Are they still in transit? Were they out of stock?
Based on this information, the structured workflow determines the appropriate options: a refund, reshipment, or another resolution. If needed, it requests more information from the customer, leveraging the LLM to handle the conversation.
Here, the LLM excels at navigating the complexities of human language and dialogue. But the critical business logic—like querying databases, checking stock, and determining resolutions—lives in predefined workflows.
This approach ensures:
Reliability: The same logic applies consistently across all users.
Security: Sensitive operations are tightly controlled.
Efficiency: Developers can test, version, and improve workflows like traditional software.
Structured automation bridges the best of both worlds: conversational fluency powered by LLMs and dependable execution handled by workflows.
What About the Long Tail?
A common objection to structured automation is that it doesn’t scale to handle the “long tail” of tasks—those rare, unpredictable scenarios that seem impossible to predefine. But the truth is that structured automation simplifies edge-case management by making LLM improvisation safe and measurable.
Here’s how it works: Low-risk or rare tasks can be handled flexibly by LLMs in the short term. Each interaction is logged, patterns are analyzed, and workflows are created for tasks that become frequent or critical. Today’s LLMs are very capable of generating the code for a structured workflow given examples of successful conversations. This iterative approach turns the long tail into a manageable pipeline of new functionality, with the knowledge that by promoting these tasks into structured workflows we gain reliability, explainability, and efficiency.
From Runtime to Design Time
Let’s revisit the earlier example: A customer messages your AI assistant saying, “Hey, you messed up my order!”
The Prompt-and-Pray Approach
Dynamically interprets messages and generates responses
Makes real-time API calls to execute operations
Relies on improvisation to resolve issues
This approach leads to unpredictable results, security risks, and high debugging costs.
A Structured Automation Approach
Uses LLMs to interpret user input and gather details
Executes critical tasks through tested, versioned workflows
Relies on structured systems for consistent outcomes
Predictable execution: Workflows behave consistently every time.
Lower costs: Reduced token usage and processing overhead.
Better security: Clear boundaries around sensitive operations.
Easier maintenance: Standard software development practices apply.
The Role of Humans
For edge cases, the system escalates to a human with full context, ensuring sensitive scenarios are handled with care. This human-in-the-loop model combines AI efficiency with human oversight for a reliable and collaborative experience.
This methodology can be extended beyond expense reports to other domains like customer support, IT ticketing, and internal HR workflows—anywhere conversational AI needs to reliably integrate with backend systems.
Building for Scale
The future of enterprise conversational AI isn’t in giving models more runtime autonomy—it’s in using their capabilities more intelligently to create reliable, maintainable systems. This means:
Treating AI-powered systems with the same engineering rigor as traditional software
Using LLMs as tools for generation and understanding, not as runtime decision engines
Building systems that can be understood, maintained, and improved by normal engineering teams
The question isn’t how to automate everything at once but how to do so in a way that scales, works reliably, and delivers consistent value.
Taking Action
For technical leaders and decision makers, the path forward is clear:
Audit current implementations:
Identify areas where prompt-and-pray approaches create risk
Measure the cost and reliability impact of current systems
Look for opportunities to implement structured automation
2. Start small but think big:
Begin with pilot projects in well-understood domains
Build reusable components and patterns
Document successes and lessons learned
3. Invest in the right tools and practices:
Look for platforms that support structured automation
Build expertise in both LLM capabilities and traditional software engineering
Develop clear guidelines for when to use different approaches
The era of prompt and pray may just be beginning, but you can do better. As enterprises mature in their AI implementations, the focus must shift from impressive demos to reliable, scalable systems. Structured automation provides the framework for this transition, combining the power of AI with the reliability of traditional software engineering.
The future of enterprise AI isn’t just about having the latest models—it’s about using them wisely to build systems that work consistently, scale effectively, and deliver real value. The time to make this transition is now.
Each January for more than a decade, the editors of IEEE Spectrum have delivered an issue that tries to predict the future. Specifically, we bring you a selection of important technology projects that we think will reach a significant milestone in the coming year. Sometimes we get it right. Sometimes we miss the mark. Test your knowledge of how some tech milestones from recent years really turned out.
Almost four years to the day since his father was taken into custody for his part in the January 6 Capitol riot, Jackson Reffitt watched in complete shock as President Donald Trump signed an executive order that pardoned and commuted sentences for his father and some 1,500 other insurrectionists.
Reffitt has spent most of the last four years in hiding, constantly on the move every few months. He was the person who tipped off the FBI about his father’s involvement in the insurrection. Jackson’s father, Guy Reffitt, was a member of the Texas Three Percenter group when he stormed the Capitol wearing body armor and carrying a pistol and zip ties. He was caught on camera urging other rioters to storm the Capitol building and told members of his militia group that he intended to drag House Speaker Nancy Pelosi out of the building by her ankles, “with her head hitting every step on the way down.”
“Trump himself has given him a presidential pardon to let him be free. That validation is a once-in-a-lifetime kind of experience that he’s never going to get again,” Reffitt tells WIRED. “I can’t imagine what he’s going to be willing to do now. It could get a whole hell of a lot worse.”
Reffitt is “terrified” about what is going to happen next, and has armed himself with a handgun and a rifle to protect himself and his boyfriend. Over the last few years, he has been targeted, harassed and threatened online.
Since Trump pardoned everyone, the threats are becoming even more intense.
“[In the last 24 hours] it’s gotten worse than ever,” Reffitt tells WIRED. “I think just because, yet again, the validation that Trump is bringing is just making people a whole lot more emboldened to just say some vile, disgusting shit.”
Reffitt is not the only family member of a released January 6 prisoner who is concerned about the fallout from Trump’s blanket pardons. Tasha Adams, the ex-wife of Oath Keeper leader Stewart Rhodes who had his 22-year sentence for seditious conspiracy commuted by Trump, is also worried about what might happen. “Stewart is out of prison now and, frankly, I could really use a bit of a run fund, in case it comes to that,” Adams wrote on her GoFundMe page on Tuesday, hours after her ex-husband was released from prison.
The investigation into the January 6, 2021 attack on the Capitol became the biggest in Justice Department history, and left many far-right militia groups in the country in ruins. But with a single pen stroke on Monday night, Trump has reinvigorated the militia movement, freeing their most prominent figures, including Rhodes and Proud Boy leader Enrique Tarrio.
“One of things I’m most concerned about is the risk of groups that were decimated after J6 coming back stronger, especially since many of them had their sentences commuted or were outright pardoned,” says Luke Baumgartner, a research fellow at George Washington University’s Program on Extremism. “I wouldn’t be shocked if the Oath Keepers began making more appearances, and seeing the Proud Boys accelerate their culture war tactics, especially against the LGBTQ community, like we’ve seen before. Their leaders are free, they have a lot to catch up on, and they are likely feeling vindicated.”
Got a Tip?
Are you a family member of a January 6 prisoner who is being released? We’d like to hear from you. Using a nonwork phone or computer, contact David Gilbert at david.gilbert@wired.com or securely on Signal on DavidGilbert.01
Guy Reffitt was the first rioter to go on trial for his actions on January 6 and initially received a sentence of seven years and three months, which was reduced by seven months in December after a Supreme Court ruling that led to the dismissal of an obstruction charge against him.
“I’m a very strong Patriot, with fabulous support from Patriot Warriors, as we navigate troubled waters,” Reffitt wrote to one acquaintance from jail in a text message submitted by the prosecution at his December resentencing.
Wildfires are raging across Los Angeles, turning the skies red, destroying homes and businesses, and blanketing the region with smoke and debris. The largest fire is in Pacific Palisades, which has grown to over 23,000 acres and forced hundreds of thousands of people to evacuate.
Two other major fires have engulfed Los Angeles County: The roughly 14,000-acre Eaton Fire in Altadena and the 799-acre Hurst Fire north of San Fernando, the latter of which has been largely contained. Several more small fires have also broken out throughout Southern California as powerful winds continue to sweep the region. At least 25 people have been killed and 12,000 structures have been destroyed.
This is a developing story. Follow here for the latest news, explainers, and analysis.
Best-selling science fiction and fantasy author Alan Dean Foster is moving into gaming in a multi-license deal with studio Pomme, starting with a game based on his classic Midworld novel.
Foster told me this in an exclusive interview for GamesBeat.
A New York Times best-selling author, Foster has written multiple book series, over 20 standalone novels, and novelizations of movie scripts including Star Wars, Alien, Terminator, Transformers and Star Trek. He has written for games before, but now he’s doing a big license of much of his work in the HumanX Commonwealth series.
Pomme is going to start with Midworld, the first novel in that series where humans encounter an alien race on a jungle-like planet. Some say this world inspired James Cameron’s Avatar films, but we won’t go there just yet. Suffice to day it’s a creatively imagined world that Foster told me in an interview lends itself to a gaming universe.
Pomme founder Darryl Still said in an interview with GamesBeat that Foster will work closely with the game team in an advisory role. Pomme, a consulting firm with many games under its belt, will assemble a team to make the first game and set up the plans for multiple games.
Alan Dean Foster has written 80 sci-fi books, story story collections, fantasy novels and novelizations of films.
Under the agreement, Pomme could make at least 14 of Foster’s works into games, including previously published novels and eight unpublished works, released in a series called “Alan Dean Foster Presents….”
As mentioned, the first will be an adaptation of Midworld – the book that launched Humanx Commonwealth, which will be published by Pomme’s publishing arm Sunset Sugar Studios on PC via Steam in 2026.
Foster said, “I look forward with great pleasure to working with the wonderful folks at Pomme as they prepare to develop and release a series of games based on my stories, the first of which is planned to be Midworld.”
Pomme was founded by industry veteran Darryl Still, formerly an executive at Atari, EA, Nvidia and Kiss. He has more than 40 years of experience in games in the industry. He is joined by Canadian industry exec Jillian Mood, who has worked with Canada Game Expo, Ottawa Game Jam, Bendy & the Ink Machine and on over 30 games. She will head up marketing and HR for the company.
Other Pomme executives include James Deputy who has project managed many games for Kiss over the 12 years while Still was CEO for that company; and Mateo Młodowski, developer of the hugely successful Pixel Puzzles franchise and development director at Sunset Sugar Studios, Pomme’s publishing arm.
“I have been working with Alan and the Pomme team for around a year now fleshing out this project and it is an absolute pleasure to be able to push the green button as we enter 2025 on what is one of the most exciting developments I have ever worked on,” Still said.
And Mood said, “Alan Dean Foster is an absolute genius and the fact he wants to be personally involved in developing these games makes me very proud.”
Pomme is collaborating with key members of the Cuphead art team, including respected animator Tina Nawrocki, to lead the art on Midworld. Pomme’s studio head Mateo Młodowski said, “As soon as we started talking to them, it became obvious they were perfect to create Börn and the other characters in the universe. The creativity they bring to the art and animation is stunning.”
Meanwhile, Pomme has established an advisory board of entertainment giants to help steer the Alan Dean Foster projects. It includes Alan Wilson, cofounder of Tripwire Interactive, publisher of Killing Floor and Red Orchestra, and Jon Radoff, CEO of Beamable, who has worked on Game of Thrones, Star Trek and Walking Dead games.
Origins of the deal
A flavor for the imagery of Midworld.
Foster, 78, has had 80 of hi sown novels published, with seven short story collections and over 40 novelizations of film scripts. (I read Foster’s novelizations of Star Wars, including Splinter of the Mind’s Eye, back when I was young).
Still founded Pomme in the United Kingdom in April 2020 as a consultancy to serve the global game businesses. Key contributors include James Deputy, Jillian Mood, and Matteo Mlodowski. Pomme (the French word for Apple) is named after Still’s love for Pomeranian dogs.
Still said that he was talking with a friend who was making a documentary about Foster’s life. The friend connected them to look into the idea of publishing games based on Foster’s unpublished works. Foster suggested the HumanX Commonwealth universe, and the team decided to make a triple-A game focused on Midworld, the first in the series, Still said.
Midworld has some crazy stuff in it.
“Midworld ideally suits our art team, which comes from Cuphead and 2D action adventure properties. We also have an Unreal Engine 3D shooter team in place that can do great things with the trilogy,” Still said. “We have licensed 14 different titles from Alan and are looking at the right teams to make each game. Some of these will be episodic with Alan narrating them.”
Still added, “We want to do the right games on the right properties and do him proud. He has a retro fan base and we can introduce it to a new fan base of gamers, and we have an ideal set of people to do that.”
Sunset Sugar Studios is the publishing arm with a publisher account on Steam. Via Mood, Still is looking for funding from the Canada Media Fund. Another available team is based in Turkiye. It’s a small team, but with big ambitions, Still said.
“This is because of the enthusiasm of people who are crazy about doing this,” Foster said. “And there are certain books like Midworld, which they’re going to start with, that are more easily adaptable into games. And fortunately, there’s enough material. I’m happy to be a part of that.”
Writing for games?
A Grazer in Midworld.
Foster said he wrote a narrative for The Moaning Words, a Lovecraft-like collectible card game with puzzles that debuted in 2014. He ssaid it taught him the complexity of writing for games and how you have to back and fix things in the narrative based on when the player counters them in the game.
“It got very complicated, but it was a lot of fun. I enjoyed it. My history with games goes way back. I did the novelization of an original computer game that was called Shadowkeep for Trillium,” he said. “That was a thousand years ago. I also did a version of LucasArts’ The Dig game. So I’ve been around the gaming industry without being directly involved with it for a long time.”
Foster said he believes that games capture actions like movement so well, and he has to write such action scenes over and over again in his various works. Early on, however, game technology couldn’t really keep up with the imagination. Now that problem has gone away as gaming tech has moved forward quickly.
HumanX Commonwealth games
Born in Midworld
“Some projects didn’t live up to the promise, but they were interesting experiments. With games, you have more options and I love the possibilities,” he said. “With Midworld, we actually have a story with different levels.
The Commonwealth series has numerous characters like Philip Lynx, who appears in 15 of the books in the series. There are dozens of books and they’re all part of the same sci-fi universe, with some fantasy elements. The HumanX Commonwealth is kind of like the United Federation of Planets in Star Trek, where in the future the humans of Earth have allied with other alien races like the insectoid Thranx from Hivehom.
“Thank goodness for computers and fans, or I’d never be able to keep this all straight,” Foster said.
He said he is glad to see so many transmedia successes, where games have been turned into movies like The Super Mario Bros. Movie and TV shows like The Last of Us.
“Technology advances along with storytelling,” he said. “You started out scratching pictures on cave walls. They didn’t move, and they weren’t very exciting. And now we just move along, and each time the technology advances, somebody finds a way to adapt it for entertainment. And that’s what’s happened with games. And once it gets sufficiently sophisticated, then you can take the game and expand it into film or TV. It would be interesting to contemplate, for example, if Tolkien were alive today, 22 years old, would he have started by writing books or would he have started it as a game?”
Ruumahum in Midworld.
Foster added, “Everything changes so fast. I love it. I love things. And the newest thing that comes along, I’m there. I want to be a part of it.”
Regarding some recent films, Foster said he was a big fan of Flow, which has no dialogue. As for the tech entrepreneurs who are always trying to turn science fiction into reality, Foster has some opinions.
“They get so wrapped up in stuff, I think, and they’re so distant from the people that they actually are supposed to be serving that they kind of forget what that’s like,” Foster said. “I haven’t met many of them and I’d be happy to sit down and chat. I don’t know. You wonder if you can have influence on people. In my books, I try to do it gently. I don’t believe in socializing or shouting.”
Daily insights on business use cases with VB Daily
If you want to impress your boss, VB Daily has you covered. We give you the inside scoop on what companies are doing with generative AI, from regulatory shifts to practical deployments, so you can share insights for maximum ROI.
Read our Privacy Policy
Thanks for subscribing. Check out more VB newsletters here.
Although genetic engineers have ways to direct evolution of molecules in the lab, they can usually test only so many possibilities. And even a protein of typical length can be changed in nearly infinite ways (since they’re built from hundreds of amino acids, and each acid comes in 20 possible varieties).
OpenAI’s model, however, often spits out suggestions in which a third of the amino acids in the proteins were changed.
OPENAI
“We threw this model into the lab immediately and we got real-world results,” says Retro’s CEO, Joe Betts-Lacroix. He says the model’s ideas were unusually good, leading to improvements over the original Yamanaka factors in a substantial fraction of cases.
Vadim Gladyshev, a Harvard University aging researcher who consults with Retro, says better ways of making stem cells are needed. “For us, it would be extremely useful. [Skin cells] are easy to reprogram, but other cells are not,” he says. “And to do it in a new species—it’s often extremely different, and you don’t get anything.”
How exactly the GPT-4b arrives at its guesses is still not clear—as is often the case with AI models. “It’s like when AlphaGo crushed the best human at Go, but it took a long time to find out why,” says Betts-Lacroix. “We are still figuring out what it does, and we think the way we apply this is only scratching the surface.”
OpenAI says no money changed hands in the collaboration. But because the work could benefit Retro—whose biggest investor is Altman—the announcement may add to questions swirling around the OpenAI CEO’s side projects.
Last year, the Wall Street Journal said Altman’s wide-ranging investments in private tech startups amount to an “opaque investment empire” that is “creating a mounting list of potential conflicts,” since some of these companies also do business with OpenAI.
In Retro’s case, simply being associated with Altman, OpenAI, and the race toward AGI could boost its profile and increase its ability to hire staff and raise funds. Betts-Lacroix did not answer questions about whether the early-stage company is currently in fundraising mode.
OpenAI says Altman was not directly involved in the work and that it never makes decisions based on Altman’s other investments.
A bad mattress can have serious health implications that go far beyond relentless tossing and turning at night. It can cause severe pain and easily exacerbate existing health conditions. That’s why the IRS allows you to use your pre-tax dollars in your health savings account (HSA) or flexible spending account (FSA) towards health-forward items, such as a smart ring or bedding. An HSA or FSA can be a great way to afford that new mattress you need, but it isn’t available to just anybody. First, you have to qualify.
Every mattress manufacturer has a different process. For example, with Sleep Number, you purchase your mattress first and then complete a survey for qualification. With Avocado, you fill out your survey first and then check out. If you’re considering purchasing a new bed, this is how to buy your next bed with HSA or FSA funds.
Can I use my FSA or HSA for a mattress?
Andrey Popov/Getty Images
Mattresses are not generally covered by HSA and FSA funds, but there are exceptions. According to the HSA Store, “A mattress expense is only eligible when its components and construction costs exceed those of a standard mattress for general health.” To use your funds for a new mattress, you’ll need to provide a letter of medical necessity (LMN) from a medical professional stating you have a medical need for a new HSA- or FSA-eligible mattress.
Common symptoms for an LMN include but are not limited to the following:
Mattress eligibility includes beds for both adults and children. For 2024, the annual pre-tax contributions for a health savings account are capped at $4,150 individually or $8,300 per family, with an additional $1,000 allotment for those 55 years and older. You have all year to use these funds, but any unused funds are forfeited after December 31 of each year. After you submit your claim and LMN, it typically only takes days to receive expense approval.
Mattress brands that may be FSA- or HSA-eligible
These are some of the most popular mattress brands that may be FSA- or HSA-eligible and their respective processes for buying a new mattress.
Sleep Number
Video producer and mattress expert Owen tests the adjustable bed frame on Sleep Number’s Climate360 smart bed.
Jon Gomez/CNET
Sleep Number is one of the most well-known manufacturers of mattresses and has an established process for buying new FSA- and HSA-eligible mattresses. Its line of smart beds includes the c2, p5 and i8 smart beds.
Buy your mattress:Once you choose the mattress you want to buy, complete the purchase as you normally would.
Complete the Truemed survey: Your receipt or order confirmation will hold a link to a quick reimbursement survey from Sleep Number’s health advocacy partner, Truemed. Results will automatically be submitted to a licensed provider so you can access a free evaluation.
Submit paperwork:If you’re approved, a LMN will be emailed within days. It’ll include instructions on how to submit your reimbursement claim to your HSA or FSA provider.
Nectar
Sleep writer Aly Lopez testing the Nectar Premier Hybrid as a stomach sleeper.
Dillon Lopez/CNET
Nectar is a popular mattress manufacturer, with popular models such as the Nectar Luxe and Nectar Premier Hybrid.
Nectar uses a simple process, much like its competitors, to apply HSA or FSA funds toward a new mattress purchase.
Go to checkout:Add your chosen mattress to your cart, and under Payment Method, choose “Pay with HSA/FSA.”
Complete the questionnaire: Nectar requires you to complete a quick survey regarding your health to confirm eligibility.
Submit your claim: Use your HSA or FSA funds directly to pay for your mattress, or you can use a personal payment method to submit a claim later to your HSA or FSA administrator.
Avocado
Video producer and mattress expert Dillon Lopez testing the Avocado Green mattress.
Jon Gomez/CNET
With Avocado, you have a wide choice of FSA- and HSA-eligible mattresses available for both adults and kids. All mattresses are eligible and have an exclusive HSA/FSA callout with a green check icon where the “Buy Now” box is located.
Check your eligibility: Complete a quick survey from Sika to determine eligibility.
Purchase your mattress: Buy your selected mattress using Sika to checkout with your HSA or FSA funds.
Submit your claim: Instructions for submitting your reimbursement claim will be on your Sika receipt. Your letter of medical necessity will then follow if you’re approved.
DreamCloud
Video producer and mattress expert Wes Ott testing the edge support of the DreamCloud mattress.
Checkout: Add your DreamCloud mattress to your cart and choose “Pay with HSA/FSA” as your payment method.
Complete the survey:DreamCloud will send you an eligibility survey to complete to determine if you can use your HSA or FSA funds for a new mattress.
Pay for your mattress: You can use your HSA or FSA card like a normal credit card to make payment. You can also use a personal form of payment and submit a claim through your HSA or FSA administrator.
Ghostbed
Video producer and mattress expert Jon Gomez testing the edge support of the GhostBed mattress.
Owen Poole/CNET
With Ghostbed, you have mattress options like the Ghostbed Oasis or the Ghostbed Luxe. Ghostbed offers a straightforward process that works in conjunction with Truemed.
Go to checkout:Add your chosen Ghostbed mattress to your cart, and select “Pay with HSA/FSA funds” at checkout.
Complete the Truemed survey:After you buy your mattress, you’ll be redirected to the Truemed website, where you’ll need to complete a quick healthcare survey.
Finish checkout: After the survey, you can complete checkout directly through Truemed. If you prefer to pay with a personal payment method, you’ll need to file a claim afterward for reimbursement.
Bottom line
If your sleep is being sabotaged by your mattress, you could receive pre-tax dollars to put towards a new one. However, mattresses are commonly excluded from HSA/FSA funds unless a medical professional deems them medically necessary. Keep in mind that it can be challenging to become approved, so you may not receive the funds for a new bed.
If budget is a concern, consider a more affordable mattress like the Brooklyn Bedding Signature Hybrid, which is a cost-effective choice with great value. Our guide to the best mattresses of 2024 can help you find the right mattress for your needs at a price you can afford.
Still looking for a New Year’s resolution for self-improvement? Consider keeping a journal, which studies have shown might help with one’s mental well-being and anxiety issues, while also providing a creative outlet for personal expression.
Handsome paper-based diaries and notebooks are available if you want to go the screen-free sensory route, but if you prefer a more multimedia approach to journaling, wake up your phone. Free apps that come with Apple’s iOS software and Google’s Android system allow you to add photos, audio clips and more to corral your thoughts — and set up electronic reminders to write regularly.
Here’s an overview.
Getting Started
Keeping a digital diary requires a few basic steps: picking an app, writing an entry and adding new posts on a regular basis. And don’t let the fear of typing long contemplative dispatches on a small screen dissuade you. Just dictate your thoughts to your iPhone or Android phone with its transcription tools, although check its privacy policy if you’re nervous about your data.
Don’t feel like pecking away on a small keyboard? You can dictate your journal entry by tapping the microphone icon on the keyboard.Credit…Google
To compose a journal entry, tap the plus icon (+) at the bottom of the screen and select the New Entry button at the top of the next screen or under a suggested topic. Go to the text field to title your entry and start writing — or tap the microphone icon at the bottom corner of the keyboard to dictate.
In the row of icons above the keyboard, you can format the text with bold, italic or other styles; get more topic suggestions; add photos from the library or the camera; add an audio recording; and note your location. You can describe your current mood with the State of Mind screen, which can be shared with the Health app (if you allow it).
With your permission, the app shows you a list of topic suggestions drawn from your photos, locations and activities. You can turn off the suggestions by opening the iPhone’s Settings icon, selecting Apps, choosing Journal and tapping the button next to Skip Journaling Suggestions.
You can bookmark and edit your compositions by tapping the three-dot menu icon in each entry’s lower-right corner. The Journal app has a search function for looking up older entries if you don’t feel like scrolling back in time.
Using Google Keep
Google has yet to release a similar dedicated journaling app, but its 12-year-old Google Keep can do the job, organizing notes, audio clips, web pages, photos and drawings. To use it, you need a Google account and the Keep app. The app is available for Android and iOS (including the iPad), and Keep content is backed up online, where it can be viewed in a web browser.
Once you’ve installed the Keep app, open it and tap the plus button (+) in the bottom-right corner to start an entry. Using the icons at the bottom of the text-entry screen allows you to do things like add a photo or give the entry a background color.
Creating and adding a “journal” label filters your posts from other notes or lists you may use within the app. And while Keep, unlike Apple’s Journal, can’t pepper you with suggestions, you can ask Google’s Gemini or your favorite artificial intelligence assistant for topic ideas.
Other Options
Samsung Galaxy users have the Samsung Notes app as another diary option, and keeping a journal on one of the company’s pen-based tablets recreates the pen-to-paper vibe for the electronic age.
If you want a journal app with additional features (like automatically adding the day’s weather conditions), you have plenty of other choices, but you’ll probably need to pay for the premium product. Among the many apps that work on most platforms are Day One (about $3 a month), Diarium ($10 to buy) and the ambitious, A.I.-powered Reflectary (about $7 a month).
Journal apps make it easier to write about your life without the performative aspect of social media. And paying less attention to what everyone else is doing gives you more time to spend on yourself.
