Makers often look for computing devices that fit their daily routines rather than forcing routines to fit the devices. When Jankbu needed a new machine, he decided against buying a ready-made laptop. He spent months designing and assembling a cyberdeck around a Raspberry Pi 5 that delivers reliable power for browsing, running design software, and handling workshop tasks.

The main display here is a 10.1 IPS touchscreen that is mounted on a clever vertical sliding mechanism that folds down to cover the keyboard when you park the device. It’s lovely and firm, thanks to the use of steel linear rods and linear bearings; there’s no wobble in sight. The display connections are routed through a tiny cable chain repurposed from CNC machines. That way, they won’t be pinched or strained as the screen moves. The designers worked through multiple versions to get it perfect, and it shows in the smooth and consistent action you get.

Nintendo Switch 2 System

• The next evolution of Nintendo Switch
• One system, three play modes: TV, Tabletop, and Handheld
• Larger, vivid, 7.9” LCD touch screen with support for HDR and up to 120 fps

The base is a full-depth mechanical keyboard that is extremely comfortable to type on even after several hours of use. Large grab grips on the sides make it simple to move this thing about your desk. One side has bespoke scroll controls laid out in both directions, while the other has a trackball made from a Logitech Trackman Marble component that they hacked together. Buttons on the screen have a very simple arrangement, much like the industrial panels you see all the time, so you can get to what you need quickly without having to navigate menus.

This device runs on NP-F batteries, which are also used in camcorders. There’s no need to shut down to change them out, and you get a live voltage display right on the front so you can know how much runtime you have left at a look. The entire power module simply slides in and connects via the rail system, allowing for quick replacement in the field. Modularity is an important aspect of how this device is put together, since the entire chassis is lined with NATO rails, allowing you to clip on or clip off modules without getting out the tools. They also carry power and data connections, so the bits you add on communicate directly with the main board. Long story short, it’s incredibly simple to add more storage or switch in alternative sensors or ports depending on the job.

The printed parts are constructed of a particular form of polycarbonate combined with chopped up carbon fiber strands. It’s good enough to leave in a hot car without becoming soft and losing its shape, and it provides the stiffness required for a genuine chunky-feeling build. Some of the high-stress components, such as the handles and trackball housing, were machined from blocks of aluminum to provide additional strength where it is needed.

Project files are available on GitHub for anyone who wants to look at the design, print their own parts, or simply experiment with it to see if it works for them. There’s still a lot of work to do on the finishing touches, but as it stands, this device functions as a functional, repairable computer that is entirely yours. Every aspect demonstrates how intelligent choices can transform a Raspberry Pi into something that is ready for serious work, rather than just a flashy display.
[Source]

Patients who use mobile applications to manage medical conditions including depression and chronic pain might assume the apps have been evaluated by regulatory agencies to be safe and effective. But that isn’t necessarily the case.

Most of the more than 55,000 medical apps that claim to diagnose or treat a condition—or ones that provide clinical decision support, known as “therapeutic” apps—have never been assessed by any trusted neutral bodies or regulatory agencies to evaluate them for technical soundness, ethical design, or clinical benefit. The apps often don’t comply with regional data security and privacy laws to protect people’s sensitive health information.

Medical apps differ from traditional wellness apps, which provide users with insights into becoming healthier by, for example, tracking fitness activities, monitoring blood pressure, and analyzing sleep patterns.

There is no reliable way to verify that therapeutic apps deliver the results they indicate. To help ensure such apps are credible, the IEEE Standards Association (IEEE SA) recently launched the IEEE Global Medical Mobile App Assessment and Registry. The publicly searchable directory is designed to list apps that have been vetted by experts across several criteria including technical soundness, ethical design, compliance with data security and privacy regulations, and clinical efficacy, which is evidence of a clinical benefit for the patient.

“Patients, clinicians, payers, and health care systems often struggle to distinguish clinically meaningful therapeutic apps from those that are simply well-marketed,” says IEEE Senior Member Yuri Quintana, chair of the assessment and registry program. He is chief of the clinical informatics division at Beth Israel Deaconess Medical Center, in Boston. “Our goal is to establish a standardized review method using criteria developed by experts.”

Why regulation is lacking

Because the apps are intended for medical use without being part of a medical implement, they fall under the designation of software as a medical device (SaMD), according to the International Medical Device Regulators Forum. SaMD is supposed to be regulated by public health agencies such as the U.S. Food and Drug Administration, but the apps have developed and grown in popularity so quickly that regulators haven’t been able to keep up, Quintana says. Some companies have received approval, but most have not, he says.

Many users are unaware of the regulatory gap, he says.

“Seeing an app from a well-known company often creates the impression that it has been meaningfully vetted for safety and efficacy, even when that is not the case,” he says.

Some companies are using deceptive advertising to sell their product, he adds. Marketing materials might claim that all of a company’s health apps are certified, even though only one app has been approved by a regulatory body to treat a particular condition. Or the verbiage might imply the company has clinical evidence proving its application works, even though the app has never been tested independently.

Another concern is that updated apps aren’t being vetted, says Maria Palombini, IEEE SA’s director of health care and life sciences global practice lead.

“The original app might have received approval from a regulatory agency, but not the updated version,” Palombini says. “There could have been significant changes from the original.”

“Not every medical-related app triggers the same regulatory classification or review across jurisdictions,” Quintana adds. “That leaves a large gray zone of clinically relevant but lower-risk apps that haven’t undergone an independent assessment. The IEEE registry was created to help fill these gaps.

“IEEE is the best organization to address this problem because this is fundamentally a standards, trust, interoperability, and conformity assessment challenge,” he says. IEEE “is the world’s largest technical professional organization, with deep expertise in developing globally recognized standards including in health care, cybersecurity, AI ethics, and interoperability.”

“Through the IEEE Conformity Assessment Program, we already run rigorous assessment and registry programs,” Palombini says. “Our neutral, consensus-driven, multidisciplinary approach—bringing together clinicians, regulators, developers, and ethicists without commercial bias—makes IEEE uniquely positioned to create trustworthy global guardrails that can scale across jurisdictions and support regulatory harmonization.”

How the registry works

The assessment framework was developed by a multidisciplinary group of 35 volunteer experts from 10 countries, Quintana says. The panel includes academics, AI experts, app developers, clinicians, ethicists, mental health experts, patient advocates, regulators, researchers, technologists, and those who assess safety in health care.

The registry is for any app used for clinical care or therapeutics that claims to demonstrate a medical benefit. That includes apps designed for cardiology, diabetes, mental health, neurology, oncology, rehabilitation, and respiratory diseases, Quintana says.

Initially, he says, the focus will be on apps that aim to treat mental health conditions, given the large number of offerings in that area and the registry committee’s expertise.

The submission of apps is voluntary. There is no government mandate that requires a company to use the IEEE registry.

The products will be evaluated against about 150 consensus-based criteria across three major areas:

• Clinical efficacy including therapeutic effectiveness, any sustained benefits, risk management, comparison to standard care, user engagement, and real clinical value.
• Technical soundness including accessibility, privacy and security, error handling, interoperability, AI governance, usability, and operational quality.
• Ethical design including bias prevention, patient consent, data governance, conflict-of-interest transparency, responsible use of AI and large language models, and prioritization of public health benefits.

IEEE charges a nonrefundable submission fee that covers the cost of the assessment plus the registry’s annual subscription for the first year.

Developers first must demonstrate they are a legally established entity before they can complete the app publisher registration form and then submit documentation and attestations about the product.

The IEEE review of an app is estimated to take six to eight weeks, Palombini says. The assessment results will be privately shared with the app publisher, she says, and to be listed in the registry, an app must achieve more than 85 percent compliance in each category.

Upgraded apps must be submitted and reassessed, Palombini says. Similar to how users are notified when an app on their smart devices has , the registry will be notified when listed apps have a new update available, she says.

Applicants who do not pass the assessment are to receive feedback explaining why. They will be given an opportunity to make changes or provide additional documentation, Palombini says.

“It’s a pretty methodological process, with checks and balances,” Quintana says. “We’re being very transparent about the process.”

Approved apps added to the registry receive an IEEE certification badge and submission identifier, which the company can display on its website, app store listings, and marketing materials.

“The badge serves as visible proof that the app has met the independent, consensus-based assessment for clinical value, technical robustness, and ethical design,” Quintana says.

The registry will be publicly available at no cost, he says.

Patients and families seeking safe, trustworthy apps—and payers and insurers evaluating reimbursement potential—will find the registry helpful, he says.

The application website is open. The public registry page does not yet list a specific count of approved apps because assessments are ongoing. Approved apps and their unique identifiers are to be published when the initial reviews are completed.

To learn more, you can watch a webinar recorded in March.

The assessment framework that underpins the registry is supporting the formal recognition of IEEE P3962 Standard for Criteria Assessment Framework f

from the I’m-sorry-I-can’t-do-that,-Dave dept

Late last year we wrote about a new startup that was flooding the internet with AI-generated podcast slop. Featuring fake hosts having fake discussions, the startup proudly stated it was creating about 3,000 new AI-generated podcasts every single week. The owners of the startup (who called critics of AI slop “Luddites,”) stated that because they cost so little to produce, even selling 30 episodes for a dollar nets them a tidy profit when scaled up appropriately.

That this results in an internet positively full of lazy mass-produced cack — and what that does to the public interest, authentic creators, and informed consensus — doesn’t really enter into it.

Not to be outdone, Amazon appears poised to join the AI slop podcast race. The company announced this week that it had begun mass producing AI-generated podcasts featuring two fake experts having conversations about all sorts of stuff. More specifically, Amazon is reformatting Alexa+’s extended answers on different topics and turning them into “podcasts.”

During this process, Jeff Bezos owned software will express manufactured opinions on all sorts of things, from the death of monoculture to the health of the U.S. recording industry:

“In an example clip shared by Amazon of the new Alexa Podcasts feature, the two AI-generated hosts discuss “the latest music releases.” A male Alexa+ narrator says more than 50% of music listening now comes from unsigned artists. “The monoculture is just gone,” a female-voiced Alexa+ narrator chimes in. The male Alexa+ host says there has been “stoner metal,” indie pop and experimental hip-hop music “all dropping on the same Friday,” and adds, “That’s not chaos — that’s the healthiest the music ecosystem has ever been.”

Cool.

For some reason the Variety story didn’t quote the best part of the shared Amazon example clip; namely where software in a female voice informs you that there’s no gatekeeping anymore and authenticity rules the day:

“There’s no gatekeeping anymore. If you make something real people are going to find it, and the algorithm is working for artists in a way it wasn’t five years ago.”

Clearly concerned that people would accuse them of creating yet more lazy and quickly automated engagement slop in the era of AI obituary scams, Amazon is pinky swearing that journalists will play a central role in fact-checking the content:

“Seemingly to dispel the notion that these “podcasts” will be AI audio slop, Amazon emphasized that it has deals with major news organizations to ensure “accurate, real-time news and information.” Those include the Associated Press, Reuters, the Washington Post, Time magazine, Forbes, Business Insider, Politico and USA Today; publications from Condé Nast, Hearst and Vox Media; and more than 200 local newspapers across the U.S.”

All that extra journalistic manpower just laying around from places like the Jeff Bezos owned Washington Post (which just fired 300 journalists and shitcanned its last black female opinion columnist). Or Business Insider, one of the cornerstones of what I call “CEO said a thing!” pseudo journalism. Or Forbes, which now just lets any random yahoo contribute as a “regular columnist.” Or Vox, which is about to be sold off to Rupert Murdoch’s kid. Or Politico, the website owned by a rich German Trump fan.

You know, all the places that have been hollowed out by layoffs and mismanaged into the ground by incompetent billionaires who have no idea how anything works and are keen to produce a giant badly automated engagement ouroborus that shits money without needing to pay human beings a living wage (or health insurance).

In effect they’re using software automation to algorithmically hijack and repackage the informed expertise of other people, then reselling it to you as something new. With some lip service to the idea that there are enough journalists left to maintain factual quality control over large language models prone to errors, plagiarism, and all sorts of disastrous fuckery at scale.

I desperately want to believe that as we accelerate into the era of badly automated mass engagement slop, there will be a value premium placed on authentic expertise. That the bland homogenized vibe coded half-assed sameness being plattered up at impossible new scale will usher forth a renaissance for real connection, genuine skill, actual talent, and human expertise.

But then I remember what most people buy at the grocery store. And the kind of people dictating the contours of both large language models and our increasingly consolidated, authoritarian-friendly media gatekeeping systems. And I quickly have my doubts that authentic expertise and connection has any meaningful chance of being heard above the din.

Filed Under: ai, ai slop, authenticity, engagement, expertise, jeff bezos, media, podcasts

Companies: amazon

Biometric authentication is no longer limited to fingerprints and face unlock. Researchers are now exploring whether your earbuds can recognize you simply by listening to the tiny vibrations created by your heartbeat.

A new study published on the arXiv preprint server introduces “AccLock,” a passive authentication system that uses standard earphone hardware to verify a user’s identity. Instead of relying on microphones or voice prompts, the system works through built-in accelerometers already found in many modern earbuds.

Your heartbeat may become your next password

The technology captures heartbeat-induced vibrations inside the ear canal, known as ballistocardiography (BCG) signals. These signals travel through bones and tissues, creating patterns unique to each person. That uniqueness is what makes the system interesting. Once the earbuds register a user’s BCG signal, they can continuously check whether the same person is still wearing them. If another user puts on the earbuds, the authentication fails automatically.

Unlike older earphone-based authentication systems, AccLock does not require users to actively interact with the device. The entire process runs quietly in the background, which could eventually make tasks like unlocking devices, approving payments, or entering smart homes feel almost invisible.

It works well — until too much movement is involved

To improve reliability, the researchers used a deep learning model and a multi-stage denoising system to separate user-specific heartbeat patterns from environmental noise and general body movement. In tests involving 33 participants, the system achieved false acceptance and false rejection rates of 3.13% and 2.99%, respectively, which is fairly promising for an experimental prototype.

However, heavy movement remains a major problem. Walking, talking, or shaking the head significantly increased error rates, showing that the system still struggles in real-world conditions. The researchers also tested the technology on Apple AirPods and found that it remained functional despite hardware limitations. While AccLock is far from becoming a commercial feature today, it offers a glimpse of a future where your earbuds quietly recognize you before you even unlock your phone.

I recently wrote an article on why I am excited about the upcoming MacBook Pro. One of the reasons mentioned there was the expected display upgrades, which will include the move to an OLED panel and quite possibly the addition of a touch screen. 

It seems that the OLED rumor is almost confirmed. According to TheElec, the OLED panels for MacBook Pro have already entered the mass production phase. Samsung Display has crossed a major manufacturing milestone, achieving a yield of over 90% for its 8.6th-generation OLED panels. 

What does yield even mean?

In simple terms, yield refers to the percentage of panels that come out of production without defects. The display industry considers anything around 90% to be the sweet spot for mass production. Samsung has not only hit that number but pushed some individual processes to 95%, which the company calls a “golden yield.”

The process had only a 80% yield last month, so getting here in just over a month is impressive. Higher yields mean lower production costs, which can eventually translate into more affordable products for consumers. Although if Apple’s past record is anything to go by, the cost benefit will not be passed on to its users. 

Samsung Display is currently running one production line at half its total capacity, producing 7,500 sheets per month. The panels are headed straight for Apple’s 14-inch and 16-inch MacBook Pro models, with shipments expected to begin as early as next month and a supply volume of around 2 million units estimated for this year.

Why is making laptop OLEDs so hard?

Laptop OLED displays are significantly harder to manufacture than smartphone panels. They are larger, stay on for longer periods, and require higher brightness, better lifespan, and brightness uniformity across a larger surface area. 

Samsung’s panel uses a two-stack tandem structure that layers two light-emitting layers on top of each other, which is one of the reasons securing high yield has been such a challenge. It’s the same tandem-OLED technology that Apple uses in its iPad Pro lineup, which packs some of the best OLED displays on the market. 

If sales of OLED MacBook Pros go well, Samsung is ready to activate its second production line, which would double output almost immediately. MacBook Neo has already become a runaway success for Apple, and Apple might be looking for a repeat. 

However, the upcoming MacBook Pros are also rumored to receive a significant price hike, which might curtail some enthusiasm.

It’s been 17 years since I sat in on the iconic weekly search quality meeting in the Ouagadougou conference room at Google’s Mountain View campus. That Thursday morning, around three dozen engineers, product managers, and executives sat at a table or sprawled on the floor to discuss why certain search queries or categories didn’t yield a perfect result and to suggest fixes. In 2010 those meetings led Google to make 550 changes to its search algorithm, a number that seemed impressive at the time.

That memory seems like a tintype. At Google’s I/O developer conference this week, a keynote speaker—head of search Liz Reid—officially down-ranked good old-fashioned search to virtual oblivion. This was a continuation of a process that began two years ago, when Google introduced “AI Overview,” its summaries that sit at the top of its search results page and literally lurk over the famous “10 blue links.” By then those links had already been degraded, so that all too often the most relevant ones were buried beneath aggregators, spam, and Google’s own shopping results and maps. Now, in what Reid described as the most significant change to the search box in the company’s history, users are in direct communication with the latest version of Google’s Gemini. Even the term “query” seems outdated, as human inputs are conversation starters for the AI to collaborate. The process can also incorporate personal information Google knows about you, which can be a lot. The answer to a query could be a bespoke presentation, maybe bolstered by AI agents that forage digital backroads to root out information. The transformation is complete. Onstage, Google said it out loud: “Google Search is AI Search.”

The search box used to be a portal to the web. The new “intelligent” box is an invitation to order up a Gemini-powered, customized response to a user’s queries, sometimes even creating on the fly a bespoke mini-publication with charts, bullet points, and even animations. Google used to pride itself on interpreting cryptic search terms to divine user intent. Now it encourages searchers to engage with Gemini in a conversational prompt-a-thon. To emphasize the change, Google representatives at the conference wore T-shirts saying “Ask Me Anything,” reflecting the prompt that Gemini offers. Just as with the computerized version, if you asked for directions from these smiling aides, the answer did not result in a click to a website.

Our digital life these days is perched at an uncomfortable transition point. AI seems to be driving every business model, and giants like Google are weaving AI into all their products and operations. At the same time, there’s rising resistance and even disgust as this powerful and scary technology worms its way into our lives. Just note the boos when commencement speakers mention AI. But as Google sees it, AI search—if you still want to call it that—is an inevitability that even AI haters will embrace.

I was among those who recoiled at the introduction of AI Overview in 2024. Now I acknowledge that Overview—and the deeper “AI Mode” that it encourages you to use—is simply better for many things, whether finding out if Saturday Night Live has a new episode, getting an explanation of an agentic harness, or even finding a link. When I searched for my WIRED article where I described the meeting in the Ouagadougou, the blue links were less than useful. But when I explained in plain language what I was looking for, I found it immediately.

So it’s working. Google claims that more than a billion people a month are searching with AI Mode, a separate tab on Google’s website where links are even more peripheral. AI Mode queries are doubling every quarter.

Investing your money at the right time and right place is key to fighting rapidly growing inflation. But when we invest our hard-earned money, it is essential to know where the money is going, who is managing it, and how. There are numerous schemes and ways to invest your money. While doing so, we often come across many heavy financial jargons, and one of them is AUM. You must have often heard this in mutual funds. So what is AUM in finance, and why is it so important? 

Keep reading to discover interesting financial facts about AUM. 

What does AUM mean?

AUM refers to Assets Under Management, which is the total market value of investments managed by an investment manager/organization on behalf of their investor, with their consent. These assets can be anything, including stocks, bonds, mutual funds, ETFs, and other investment options. 

The global asset under management (AUM) is expected to reach $200 trillion by 2030, growing at a CAGR of 6.2%. See the detailed market reports here. 

How to calculate AUM? 

It is calculated by adding the total current market value of all investments being managed.

AUM = ⅀ (Current Market Value of all Assets)

The formula, which is used to calculate the daily AUM value:

AUM_today = AUM_yesterday + Net Inflows + Market Profit/Losses

Here, net inflows are your new investments minus the redemptions, and market gains & losses are value changes of assets due to price movements. 

How is your money turned into assets? 

When you invest your money, it is converted into assets that are far more valuable than your paper money. These are the things bought with that money. And the value of these things keeps increasing faster than actual money. It can be anything, such as digital gold, ETFs, mutual funds, shares, stocks, and property. 

Types of AUM Assets

As discussed above, your money can be converted into different types of assets by using it to buy various things. Let me tell you about some of them.

• Equities or Stocks: These are shares, or say you buy a certain percentage of a publicly traded company, including large-cap, mid-cap, and small-cap stocks.
• Debt Funds: These assets are essential components of mutual funds and emphasize fixed-income securities such as government bonds, corporate bonds, treasury bills, and marketable securities. They often have lower risks and a predictive outcome. 
• Hybrid Funds: It is where investment managers diversify your money across assets like equities, debt, and sometimes even instruments like gold. It gives investors a wider portfolio and a good option for people choosing growth with less volatility.
• Thematic & Sectoral Investment: These are specified mutual funds for targeted industries or well-rounded investment themes. These have high growth potential but also come with significant risks due to condensed exposure. The risks are generally due to economic lows or sector-specific downfalls.
• IFs and ETFs: The index funds or exchange-traded funds are passive investments that track the performance of a specific market index. People like their simplicity and cost efficiency. Mostly, retailers and institutional investors prefer these investments. 
• Alternative Investments: These investment approaches are different than traditional ones. Here you can invest in land, real estate, commodities, and gold. This can be used to diversify a portfolio, offer great returns over time, but comes with high risk. 

Each of these investment types has more in-depth subcategories.

Factors that Affect AUM

The AUM value often keeps fluctuating; sometimes you gain, and sometimes you lose. There are different factors behind this, and some of them are listed below.

1. Market Graphs

The market often experiences upswings and downturns, and your underlying assets increase and decrease in value, respectively. Highly volatile assets such as stocks, commodities, crypto currencies etc can be frequently impacted. 

2. Investor Activities

Here, AUM is affected by the investor’s action. Where there are inflows (new investments by investment) like buying new units, increasing capital, etc., the value of AUM increases. On the other hand, if there are outflows or redemptions by investors pulling out their money directly decreases the AUM. 

3. Distribution

When a fund pays out dividends or interest, the AUM reduces, and if these payouts are compensated or reinvested, the value increases. When the funds with better performance outperform, the benchmarks tend to attract people to invest more, and this leads to increased AUM. 

There are also some other factors, like sales and marketing. Different fund types, such as open-ended funds and different fund structures, also impact the AUM. 

How do AUM Managers Earn?

You must be wondering if someone is using the brains and resources to invest your money, then what do they benefit? So here is how asset management companies make money. 

These companies generally sell the investment solutions as products to their clients. They sell mutual funds, ETFs, and manage private accounts of other companies. In return, they either charge a fee or a percentage of assets under management. 

The charges consider some factors such as investment type, asset class, investment sector, and transaction complexity. For instance, when an investment strategy involves a cultured process and tools like trading or taking short positions, then the clients can be charged a high fee. 

Ongoing charge fee (OCF), performance fees, initial and exit charges, etc., are some charges incurred by companies that these agents charge to clients. 

Types of Asset Management Companies

Different types of investment are managed by different specialized companies for the same purpose. 

1. Mutual Fund Companies

These companies use the investor’s money to buy stocks, bonds, and other securities that align with the fund’s objective. These companies are best chosen by retail investors. The clients get fund units, and returns as per market performance. 

2. Hedge Fund Companies

This is most opted for by high-net-worth people and institutional investors, where they use plans like leverage, short selling, and derivatives. The aim is to gain high returns in all sorts of market fluctuations. This involves high risk but has fewer regulatory restrictions. 

3. Private Equity Firms

These are companies that invest directly in unlisted/private companies, or they pool capital from institutional investors and high-net-worth clients to take over, restructure, and improve private companies. Their goal is to increase their company’s worth over a period of time before selling it for a profit. 

4. Real Estate Investment Trusts (REITs)

The firms invest in income-generating assets of real estate, like commercial spaces. The investors earn returns from real estate without actually owning the property. These corporations manage high-value real estate portfolios. Leasing, selling, and collecting rents, and later distributing among the shareholders as their incomes and dividends. 

AUM vs. NAV

Benefits of Asset Under Management

• Shows you trust and scalability, a higher AUM indicates that the fund or company is a trusted one. It reflects credibility, market position, and investor confidence.
• Larger AUM shows that fund managers can help you diversify your investments. 
• When a company has bigger AUM, it can spread fixed costs for clients, which leads to lower expense ratios for investors. 
• Fund houses with larger AUMs have better negotiating power and broader investment opportunities.
• Better AUM indicates the stability of a fund management company. 

Conclusion

I hope this blog helped you understand what is AUM in finance. When you invest, you must know how and where the investment is happening. Learning and understanding about assets under management is the first step to doing so. We have discussed various aspects of it, such as types, risks involved, benefits, and how you can calculate AUM. I have also stated the clear difference between NAV and AUM, which often confuses investors. Keep reading, keep learning. And let me know in the comments, how you choose to invest your money? 

Related: How Blockchain Can Transform the Financial Services Industry

Dun & Bradstreet has spent over 180 years building a comprehensive commercial database. Its Commercial Graph, covering 642 million businesses and their relationships, corporate hierarchies and risk profiles, was designed for people. Credit analysts, risk managers and sales professionals who could wait for query results and work through ambiguous entity matches. AI agents cannot do any of those things.

When D&B’s customers started pushing agents into credit, procurement and supply chain workflows, the Commercial Graph that had reliably served nearly 200,000 customers globally became a problem. The systems built to serve human analysts were the wrong architecture for machines. So D&B rebuilt.

“We need to think about agents as our new consumer category, evolving from our standard credit analysts or sales and marketing professionals, et cetera, to also now catering to these customers’ agents,” Gary Kotovets, Chief Data and Analytics Officer at Dun & Bradstreet, told VentureBeat.

What broke when agents started querying

The Commercial Graph was not a single database. It was a collection of separate systems built for different use cases and different markets, held together by custom integrations. Human analysts navigated that fragmentation through SQL queries or pre-built interfaces. Agents could not.

The scale of the underlying data compounded the problem. The database had nearly doubled in five years, expanding from more than 300 million to more than 642 million business records, with 11,000 fields per record, according to D&B. The firm now runs approximately 100 billion data quality checks per month as records move through its systems. Querying that at the sub-second latency agents require, against a fragmented architecture, was not workable.

The relationships the graph tracked were also the wrong kind. Legacy systems recorded static connections between entities. A CEO was linked to a company. That was the line. Agents working on credit assessments or third-party risk need dynamic relationships: when that CEO leaves for a new company, which organization does their track record follow? When a subsidiary changes ownership, how does that propagate across a corporate hierarchy? Those questions required custom analyst work before. Agents cannot wait for custom analyst work.

The broader problem is not unique to D&B. Kotovets said he has spoken with hundreds of CDOs and CIOs over the past six months and consistently heard the same constraint: they could not build what they wanted in AI because their data foundations were not standardized, normalized or agent-queryable. D&B had that foundation, built over decades to serve human analysts. It still had to rebuild for agents.

What they actually built

The rebuild started with consolidation. D&B migrated its fragmented databases to cloud infrastructure, redesigned the underlying schema and built a data fabric layer that normalizes records across markets while preserving regional compliance requirements. The result is a unified knowledge graph that tracks billions of relationships across 642 million companies, continuously updated and enriched by AI-driven data processing.

On top of that graph, D&B built a structured access layer for agents. Raw SQL access at agent query volumes and latency requirements was not the answer. Instead, D&B created a set of tools and skills available through MCP that package data with context and route agents to the right records for specific queries. A match and entity resolution engine sits behind every query, confirming that when an agent asks about a company, the answer resolves to a verified, specific entity rather than a name match.

D&B solved agent identity from both directions

Rebuilding the graph and adding MCP access solved the data retrieval problem. It did not solve the identity problem. Agents are not humans, and the authentication model built for human users did not extend to machines.

D&B built a new registration model for agents. They must map to a verified IP address and register an individual access key, treated as an authenticated identity in the same pipeline as a human user.

“We actually have a concept of Know Your Agent, similar to know your customer, that does those additional verifications,” Kotovets said.

That handles the inbound problem: knowing which company an agent belongs to and what data it is entitled to query. But D&B also built for the outbound problem: what happens when a customer’s own multi-agent workflow loses track of which company it is analyzing.

In a workflow that chains a credit check agent, a KYC agent and a third-party risk agent, each queries D&B at a different step. Without a mechanism to confirm they are all referencing the same entity, a workflow can complete while operating on divergent records.

“They have to come back to our verification agent to ensure that they’re still talking to each other about the same entity,” Kotovets said. “It’s almost like a digital handshake, in a sense.”

D&B’s business verification agent can be embedded into any workflow as a persistent reference point and is available on Google’s A2A protocol regardless of which orchestration tool a customer uses.

Four things enterprises must get right before deploying AI agents

The rebuild exposed requirements that go beyond D&B’s own stack.

1. Data foundations come before agent infrastructure. The CDOs and CIOs Kotovets spoke with over the past six months consistently hit the same wall: they cannot build what they want in AI until their data is clean, normalized and consolidated. D&B had that foundation already. Most enterprises do not, and they will feel it.

2. Design for dynamic relationships, not static ones. Enterprise data systems typically record point-in-time connections: a person belongs to a company, an asset belongs to a subsidiary. Agents working on credit, risk or supply chain decisions need to reason across relationships that shift over time. If the underlying data only captures the static line, the agent will too.

3. Build entity consistency checks into multi-agent workflows. When multiple agents touch the same entity at different steps, there is no guarantee they are all referencing the same record by the time the workflow completes. That gap needs to be engineered for explicitly. Entity verification is a workflow design requirement, not an optional guardrail.

4. Embed lineage from the start, not as an afterthought. Every agent-produced answer should carry a traceable path back to its source. In credit, risk and supply chain decisions, the cost of an error is concrete. Lineage needs to be built in before scaling, not added after problems surface.

“You could always click and see where it came from, and validate it all the way back to the original source,” Kotovets said. “That’s been the key for us in unlocking a lot of other capabilities, because we have that level of certainty in the things that we’ve done.”