• Demonstrators gathered outside Poetica Coffee’s Williamsburg location on Wednesday.
• The protest came after Poetica said on social media that Rep. Dan Goldman was not welcome back after a visit.
• Protesters and counter-protesters traded insults. One passerby suggested the barbs hyperbolic.

Roughly 65 protesters and counter-protesters gathered outside a Brooklyn coffee shop on Wednesday after the chain banned Rep. Dan Goldman over his stance on Israel.

The demonstrations at Williamsburg Poetica Coffee shop came days after the chain’s now-deleted Instagram post showing Goldman inside its Park Slope café — and said he was not welcome back.

In the caption, Poetica said it did not serve “genocide enablers” and told the New York Democrat not to return. It also said it had refunded his order.

“Hey Congressman Dan Goldman, we see that you stopped by our shop today for a coffee,” the deleted post read. “Do you see how it doesn’t taste like genocide juice?”

On Wednesday morning, Business Insider observed a few dozen protesters outside Poetica’s Williamsburg location, along with two smaller, less organized counter-protest groups. NYPD officers separated the groups with barricades as demonstrators shouted competing chants into microphones and megaphones.

The demonstration — organized by #EndJewHatred, a Jewish advocacy group — included several speakers who called on Poetica to apologize, praised the Justice Department’s decision to open a civil-rights investigation into the chain, denounced New York City Mayor Zohran Mamdani, and led a chant supporting the New York City Police Department.

In two other barricaded areas, counter-protesters held signs condemning Israel’s military actions in Gaza, the West Bank, and Lebanon. Several of them took turns using a megaphone to chant or address the pro-Israel demonstrators.

NYPD officers directed protesters, counter-protesters, and onlookers to stay behind temporary fencing. Officers also directed traffic, which was mildly slowed by camera crews outside the coffee shop.

A small stream of customers continued to walk into the shop past the protesting groups. One man ate a muffin inside the shop while watching the protests.

Business Insider spoke to several people who walked by, as well as people who came to support both sides. While some demonstrators said they were motivated by strongly held views about antisemitism, Israel, and Palestinian rights, several passersby said they had not heard about the protest beforehand and had stopped because of the spectacle.

One passerby suggested the groups were being hyperbolic. “It’s a lot of mental gymnastics both groups are doing,” the man said.

The controversy surrounding Poetica has already drawn national attention and federal scrutiny. The Justice Department’s Civil Rights Division said it had opened an investigation into the incident, citing federal laws that bar public companies, including coffee shops, from discriminating on the basis of race, religion, or national origin.

Poetica did not immediately respond to a request for comment. The chain’s Instagram account was unavailable on Wednesday afternoon.

Poetica Coffee’s Yelp page also displayed a “public attention alert” after the business received an influx of one-star reviews.

Goldman, who is Jewish and represents parts of Manhattan and Brooklyn, said earlier this week that he had stopped by the café with his young daughter and had a pleasant interaction with the staff. In an Instagram comment, he called the shop’s post “sad.”

In a later interview with CNN, the US representative for New York’s 10th congressional district said that he supported human rights and “what is going on in the Middle East is horrific”

“Now, I may disagree as to whether or not there’s a genocide, but come on, we’re better than this, and we need to be better than this,” Goldman said about the incident. He added that Justice Department resources might be better used on antisemitism cases involving people without his public platform.

Goldman lost Tuesday’s Democratic primary for New York’s 10th congressional district to Brad Lander, the former New York City comptroller.

• The release date for Google’s new Gemini 3.5 Pro model has been pushed back to July.
• The company has been improving the model based on feedback from early testers and its Flash model.
• The pressure is on Google to deliver amid intense competition between rival AI labs.

The release date for Google’s next frontier AI model has been pushed to July, Business Insider has learned.

The company previously said it planned to roll out the new Gemini 3.5 Pro model in June. However, it is now targeting a July launch as it spends extra time gathering feedback from early testers and tweaking the model, according to a person familiar with the matter.

Google teased the new model at its I/O developer conference in May but said it wasn’t quite ready. At the time, CEO Sundar Pichai said the model would launch “next month.”

A Google spokesperson declined to comment.

With this upcoming model, the pressure is on for Google at a moment of intense competition among the AI labs. While Gemini 3 outperformed expectations last year, Anthropic and OpenAI are continuing to pull ahead of Google in coding, which has emerged as the first major enterprise use case for modern AI.

The source said that Google pushed the launch date back so it could spend more time gathering real-world use cases from early testers. The new model has been available to some users on Google’s Antigravity platform and on the AI benchmarking site LMArena, they said.

The new Gemini 3.5 Pro model is expected to be better at long-horizon tasks and powering agents.

Google has also incorporated feedback from its recent Flash 3.5 model into 3.5 Pro, the source said, confirming a theory that Business Insider floated at I/O. That includes criticisms that Flash consumed tokens too quickly.

Have something to share? Contact this reporter via email at hlangley@businessinsider.com or Signal at 628-228-1836. Use a personal email address and a non-work device; here’s our guide to sharing information securely.

• A new legal filing is pushing to stop the transition of millions of student-loan borrowers off the SAVE plan.
• Beginning July 1, borrowers will start to receive their 90-day timeframe to leave the plan.
• The lawsuit said that borrowers who reached the relief threshold would be harmed by the transfer.

A new legal filing asks a federal judge to pause the forced transfer of millions of student-loan borrowers off of SAVE, the income-driven repayment plan that the Trump administration eliminated in March.

On Tuesday night, the law firm Public Goods Practice filed a motion in a US District Court seeking to block borrowers enrolled in SAVE from being automatically moved to a new, more expensive repayment plan while their broader lawsuit continues.

The case, filed in March, challenges the Department of Education’s decision to eliminate SAVE, a Biden-era plan that reduced monthly payments and shortened the timeline for debt relief.

The motion argues that legal procedure requires the department to make the benefits of REPAYE, a precursor to SAVE, available to borrowers and to pause forced transfers to a new plan while the case proceeds.

A Department of Education spokesperson disputed the merits of the argument and recommended that borrowers enroll in “a lawful repayment option,” such as the new Repayment Assistance Plan, which will become available on July 1.

Also beginning July 1, borrowers enrolled in SAVE will start receiving notices from their servicers on their 90-day timeframe to transfer to a new plan. If they do not voluntarily switch, those borrowers will automatically be placed in the standard repayment plan or the new tiered plan, both of which are the most expensive options.

Austin Hinkle, a managing partner at Public Goods Practice, said he’s pushing to pause that transfer before borrowers reach the deadline.

“Once borrowers start getting transferred over, injuries for lots of borrowers could start to happen right away,” Hinkle said, referring to the higher monthly payments borrowers would face under new repayment plans.

Democratic lawmakers have previously pushed the department to give borrowers more time to transition to a new plan. They wrote in an April letter that “borrowers deserve to have the time, critical information, and support necessary to successfully enroll in another affordable repayment plan and continue to pay down their loans.”

Have a story to share about student loans? Contact this reporter at asheffey@businessinsider.com.

• Garfield, an AI-powered law firm, has helped a freelancer win a court case in England.
• The freelancer used Garfield’s software to draft a legal letter to try to recover unpaid fees.
• Garfield is part of a new class of startups using technology to make legal services more accessible.

An English court has handed a victory to a law firm that uses artificial intelligence more than lawyers to bring legal claims.

Garfield, a tech company that is also a regulated law firm, helped a freelancer recover £7,000 in unpaid fees from a former client after a trial in Wandsworth County Court in May.

The case was small, but its implications are big. Garfield is part of a new class of startups trying to use artificial intelligence to make legal services cheap enough for the everyman to pursue. The company believes it is the first robo-firm to win a case in an English court.

Garfield’s software helps users chase unpaid invoices by uploading documents such as contracts and invoices, then producing legal letters and court documents. Its founder, Philip Young, a longtime litigator, said the company is starting with small debt claims because they are common, painful, and often costly for businesses to solve with traditional lawyers.

“You don’t want to spend a lot of money on lawyers to collect a £4,000 debt,” Young said. “It’s just not worth it.”

In this case, Tamires Camal Taquidir, a human-resources consultant, said she had not been paid for work she performed for a hospitality company. Young said she was originally owed about £6,000.

When she pressed the claim, the defendant denied owing her anything and brought a counterclaim of about £1,500, Young said. He said he believed the counterclaim was meant to pressure her into dropping the case or accepting a steep discount.

“To her credit, because she did have a meritorious claim, she wasn’t willing to accept that,” he said.

Garfield drafted the pretrial materials. A human barrister ultimately represented Camal Taquidir in court. She paid Garfield about £400.

After the hearing, the judge issued a decision weeks later. Her claim succeeded, and the counterclaim failed.

Young said the outcome was “very satisfactory” for Camal Taquidir and for Garfield.

The ruling has already brought Garfield a surge of attention. Young said visits to the company’s website spiked 1,000% on Monday, after The Financial Times and The Guardian published articles about the case.

Garfield has processed more than 600 claims and recovered about £500,000 for clients, Young said. He said usage has been increasing over the past six months, with early adopters giving way to larger businesses and even a regulator in England using the platform.

Young began his career at the white-shoe law firm Baker McKenzie before starting his own boutique, Cooke, Young, & Keidan. After retiring from the London firm, he started playing with ChatGPT on a family road trip. He believed the technology would transform how legal services are delivered.

Last May, Garfield became the first regulated law firm of its kind when it won approval from the Solicitors Regulation Authority, the ruling body for lawyers in England and Wales. Domestic rules let non-lawyers own or invest in law firms — a structure that makes room for law firms to take on outside capital.

Even so, Young said Garfield has not raised institutional capital. So far, the company has been funded by him and his close friends.

Young said the idea for Garfield was inspired in part by his brother-in-law, a plumber in South Yorkshire who would call him when customers failed to pay.

“In England, we’ve got a choice,” Young said. “Either we can build things to solve access to justice gaps, or we can rearrange it so that every plumber has a brother who happens to be a litigation partner.”

Manufacturers have invested in connected sensors, machine monitoring, and dashboards to gain a clearer understanding of what is happening across their operations. This visibility has delivered real value, helping teams identify inefficiencies, reduce downtime, and respond to problems faster. Today, however,  industrial firms look beyond visibility to focus on how technology can respond as conditions change. Most industrial environments already generate more information than teams can reasonably analyze. More and more, what’s needed is faster, more effective responses when operational issues emerge. This is  Industrial IoT beyond monitoring toward autonomous decision systems that can recommend actions, trigger workflows, and, in some cases, automatically execute predefined responses.

Why Full Autonomy Is Not the Goal

Despite growing interest in autonomous decision systems, many companies are still working to extract value from basic monitoring and analytics initiatives. The transition toward automation is rarely linear, and sophistication levels vary significantly between industries and facilities.

Despite the attention surrounding autonomous factories, the reality is far more practical. Most industrial firms are not pursuing fully self-managing operations. Instead, they are focusing on targeted use cases where automation can improve speed, consistency, and performance without removing human oversight. In many facilities, the immediate priorities remain reducing downtime, improving quality inspection, and streamlining maintenance workflows rather than achieving full autonomy.

From Visibility to Action

Connected assets generate data that can be monitored remotely, giving operators a better understanding of equipment health and production performance. Analytics then helps identify anomalies, predict failures, and uncover opportunities for improvement. The next stage introduces decision support, where platforms don’t merely report issues, but also begin recommending responses.

Only after operators gain confidence in the underlying data and recommendations does automation typically enter the picture.

A common example is predictive maintenance. A company may initially monitor vibration and temperature data to detect abnormal equipment behavior. As confidence in the data grows, analytics can predict potential failures before they occur. Eventually, those same insights may automatically create maintenance work orders, schedule inspections, or alert the appropriate teams.

Most organizations expand automation gradually through well-defined workflows rather than attempting large-scale transformation from the outset.

How Industrial Autonomy Is Applied in Practice

Autonomous decision systems in industrial settings allow platforms to execute specific actions within tightly defined rules, safety limits, and escalation paths. The degree of automation depends heavily on industry risk, meaning adoption varies significantly between use cases. Warehouses may safely deploy autonomous robots to optimise routing and task coordination in real time, while sectors such as utilities, chemicals, and critical infrastructure require far more validation before automating any operational decisions.

In most cases, organizations are adopting bounded autonomy, where systems handle routine responses, and humans retain oversight of exceptions and high-impact decisions, reducing response times while maintaining control and accountability.

Why Edge AI and Interoperability Are Driving Change

Many industrial responses cannot wait for data to travel to the cloud and back. On a high-speed production line, waiting even a few seconds for a cloud response may mean hundreds of products have already passed through inspection. That is one reason companies are increasingly moving AI inference closer to machines and production assets.

Running AI models at the edge helps reduce latency and support faster responses when timing directly affects production performance.

Similar principles apply in other data-intensive industries that rely on real-time decision-making.  For example, a crypto trading platform may use AI models and streaming data to execute transactions based on predefined conditions, highlighting how low-latency decision systems are becoming increasingly important across a range of digital and industrial environments.

Faster responses also depend on how effectively data moves between technologies.

Many organizations still operate environments where decades-old equipment sits alongside modern cloud platforms. Sensors, PLCs, SCADA systems, historians, MES applications, and ERP platforms frequently exist in separate silos, making it difficult to move from insight to action.

Interoperability has become a critical part of Industrial IoT deployments. Standards such as OPC UA and MQTT are helping industrial firms connect data sources more effectively, making it easier to coordinate workflows across mixed-vendor environments.

Many Industrial IoT projects stall long before model performance becomes an issue because data quality, system integration, and coordination between teams remain unresolved. Data may already exist across the business, but it is often fragmented across departments, facilities, and platforms.

Where Adoption Is Advancing First

Adoption is happening fastest where the benefits are clear and automation is only allowed to act within set, controlled limits. Predictive maintenance is one of the most established use cases, with many companies moving from simple alerts to automated maintenance actions triggered by set thresholds.

Machine vision is also advancing quickly. AI inspection systems can detect defects in real time and trigger sorting, rework, or quality-control steps, particularly in automotive and electronics manufacturing, where results are easy to measure.

Energy optimization is another growing area, with systems adjusting schedules, equipment settings, and load levels to reduce waste while maintaining output. Most successful deployments focus on specific problems rather than full autonomy, delivering faster and more measurable returns.

The Challenges Ahead

Data quality, integration complexity, cybersecurity concerns, and skills shortages continue to slow progress across industrial environments. Creating reliable data foundations is often more difficult than deploying new analytics tools, and poor information can scale mistakes just as quickly as good information can scale efficiencies.

Before expanding automation, organizations need confidence that their data is accurate, recommendations are explainable, and appropriate safeguards are in place when outcomes differ from expectations.

The Shift Toward Trusted, Controlled Automation

The companies making the most progress are not necessarily those using the most advanced AI, but those with enough trust in their data and workflows to allow safe automation when speed matters. For most industrial firms, the future is not full autonomy, but steady adoption of automated workflows that handle routine decisions, while humans focus on complex, high-value judgement.

The post How Industrial IoT Is Evolving Into Autonomous Systems appeared first on IoT Business News.

California has a lot of billionaires, more than any other state and more than most countries. So a proposed wealth tax on its billionaires could be a windfall.

Under the Billionaire Tax Act, California residents with a net worth of over $1 billion would face a one-time tax equal to 5% of their net worth.

The tax plan got one step closer to becoming reality after garnering enough signatures to appear on the ballot. The deadline for ballot measures is June 25, so unless opponents can cut a deal with the groups backing the measure before then, California voters will decide on the tax in November.

If passed, the tax would apply retroactively to billionaires living in the state as of January 1, 2026. The tax would be due in 2027, with the option to spread the payment out over five years, with interest.

The tax plan has drawn sharp reactions from lawmakers and business leaders.

Google cofounders Larry Page and Sergey Brin moved entities tied to them out of the state in December ahead of the deadline, Business Insider first reported.

Nvidia CEO and billionaire Jensen Huang said he was “perfectly fine” with the tax. Palmer Luckey, the billionaire founder of defense tech startup Anduril, said it would force companies to “immediately pivot into profit obsession over mission or long-term sustainability.”

Critics of the tax, including Gov. Gavin Newsom, have warned it will encourage ultrawealthy residents to flee the state and hurt California’s economy. They’ve also said the one-time tax would not provide a sustainable source of funding.

Several left-leaning groups have come out against the tax in recent weeks, including the California Teachers Association and Planned Parenthood Affiliates of California.

Supporters, including SEIU-UHW, a healthcare workers union, say the tax would provide crucial funding for healthcare services to offset federal funding cuts, as well as for education and food assistance programs.

As of January 1, there were 214 billionaires in California, according to Forbes data compiled by Americans for Tax Fairness, a group that advocates for higher taxes.

Below is the full list of billionaires in California. Names with asterisks have recently moved at least some of their business entities out of the state.