Connect with us

Business

Want AI agents to work better? Improve the way they retrieve information, Databricks says

Published

on



Hello and welcome to Eye on AI. In this edition…Nvidia snags the team and tech from AI chip startup Groq…Meta buys Manus AI…AI gets better at improving AI…but we might not know enough about the brain to reach AGI.

Happy New Year! A lot has happened in AI since we signed off for the year just before Christmas Eve. We’ll aim to catch you up in the Eye on AI News section below.

Meanwhile, as I’ve noted here before, 2025 was supposed to be the year of AI agents, but most companies struggled to implement them. As the year drew to a close, most companies were stuck in the pilot phase of experimenting with AI agents. I think that’s going to change this year, and one reason is that tech vendors are figuring out that simply offering AI models with agentic capabilities is not enough. They have to help their customers engineer the entire work flow around the AI agent—either directly, through forward deployed engineers who act as consultants and “customer success” sherpas; or through software solutions that make it super easy for customers to do this work on their own.

A key step in getting these workflows right is making sure AI agents have access to the right information. Since 2023, the standard way to do this has been with some kind of RAG, or retrieval augmented generation, process. Essentially, the idea is that the AI system has access to some kind of search engine that allows it to retrieve the most relevant documents or data from either internal corporate sources or the public internet and then the AI model bases its response or takes action based on that data, rather than relying on anything it learned during its training process. There are many different search tools that can be used for a RAG system—and many companies use a hybrid approach that combines vector databases, particularly for unstructured documents, as well as more traditional keyword search or even old-fashioned Boolean search.

But RAG is not a panacea and simple RAG AI processes can still suffer from relatively high error rates. One problem is that AI models often struggle to translate a user’s prompt into good search criteria. Another is that even if the search is conducted well, often the model fails to properly filter and sift the data from an initial search. This is sometimes because there are too many different data formats being retrieved, and sometimes because the human who is prompting the AI model has not written good instructions. In some cases, the AI models themselves are not reliable enough and they ignore some of the instructions.

But, most of the time, AI agents fail not because the agent “is not able to reason about data but the agent is not getting the right data in the first place,” Michael Bendersky, the research director at Databricks tells me. Bendersky was a long-time veteran of Google, where he worked on both Google Search and for Google DeepMind.

Databricks introduces a new retrieval ‘architecture’ that beats RAG

Today, Databricks (known for its data analytics software) is debuting a new architecture for retrieval-augmented AI agents called Instructed Retriever that it says solves most of RAG’s shortcomings.

The system translates a user’s prompt and any custom specifications that the model should always consider (such as the recency of a document or whether a product has good customer reviews) into a multi-step search plan for both structured and unstructured data—and, crucially, metadata—to get the right information to the AI model.

Much of this has to do with translating the natural language of the user’s prompt and the search specifications into specialized search query language. “The magic is in how you translate the natural language, and sometimes it is very difficult, and create a really good model to do the query translation,” Hanlin Tang, Databricks’ CTO for neural networks, says. (Tang was one of the cofounders of MosaicML, which Databricks acquired in 2023.)

On a suite of benchmark tests that Databricks designed that it says reflects real world enterprise use cases involving instruction-following, domain-specific search, report generation, list generation, and searching PDFs with complex layouts, the company’s Instructed Retriever architecture resulted in 70% better accuracy than a simple RAG method and, when used in a multi-step agentic process, delivered a 30% improvement over the same process built on RAG, while requiring 8% fewer steps on average to get to a result.

Improving results even with under-specified instructions

The company also created a new test to see how well the model can deal with queries that may not be well-specified. It is based partly on an existing benchmark dataset from Stanford University called StaRK (Semi-structured Retrieval Benchmark). In this case, Databricks looked at a subset of these queries related to Amazon product searches, called StaRK-Amazon, and then further augmented this dataset with additional examples. They wanted to look at search queries that have implied conditions. For instance, the query, “find a jacket from FooBrand that is best rated for cold weather,” has multiple implied constraints. It has to be a jacket. It has to be from FoodBrand. It has to be the FooBrand jacket that has the highest rating for cold weather. They also looked at queries where users want to exclude certain products or want the AI agent only to find products with recent reviews.

The idea of the Instructed Retriever architecture is that it turns these implied conditions into explicit search parameters. Bendersky says the breakthrough here is that Instructed Retriever knows how to turn a natural language query into one that will leverage meta data.

Databricks tested the Instructed Retriever architecture using OpenAI’s GPT-5 Nano and GPT-5.2, as well as Anthropic’s Claude-4.5 Sonnet AI models, and then also a fine-tuned small 4 billion parameter model they created specifically to handle these kind of queries, which they call InstructedRetriever-4B. They evaluated all of these against a traditional RAG architecture. Here they scored between 35% to 50% better in terms of the accuracy of the results. And the Instructed Retriever-4B scored about on par with the larger frontier models from OpenAI and Anthropic, while being cheaper to deploy.

As always with AI, having your data in the right place and formatted in the right way is the crucial first step to success. Bendersky says that Instructed Retriever should work well as long as  an enterprise’s dataset has a search index that includes metadata. (Databricks also offers products to help take completely unstructured datasets and produce this meta data.)

The company says that Instructed Retriever is available today to its beta test customers using its Knowledge Assistant product in its Agent Bricks AI agent building platform and should be in wide release soon.

This is just one example of the kinds of innovations we are almost certainly going to see more of this year from all the AI agent vendors. They might just make 2026 be the real year of AI agents.

With that, here’s more AI news.

Jeremy Kahn
jeremy.kahn@fortune.com
@jeremyakahn

This story was originally featured on Fortune.com



Source link

Continue Reading

Business

Allegiant to acquire Sun Country in deal valued at $1.5 billion

Published

on



Allegiant Travel Co. will acquire Sun Country Airlines Holdings Inc. in a cash-and-stock deal valued at $1.5 billion including Sun Country’s debt, the two carriers said in a joint statement on Sunday. 

Sun Country’s shareholders will receive 0.1557 shares of Allegiant common stock and $4.10 in cash per Sun Country share, the companies said. The offer represents a premium of 19.8% over Sun Country’s closing share price on Friday, according to the statement.

The combined entity will provide more than 650 routes, including 18 international destinations in Mexico, Canada, the Caribbean and Central America, the companies said.  

“Together, our complementary networks will expand our reach to more vacation destinations including international locations,” said Allegiant Chief Executive Officer Gregory C. Anderson in a statement. 

Join us at the Fortune Workplace Innovation Summit May 19–20, 2026, in Atlanta. The next era of workplace innovation is here—and the old playbook is being rewritten. At this exclusive, high-energy event, the world’s most innovative leaders will convene to explore how AI, humanity, and strategy converge to redefine, again, the future of work. Register now.



Source link

Continue Reading

Business

Iran edges closer to a revolution that would reshape the world

Published

on



As protesters pour into the streets of Iran night after night, leaders across the region and around the world are grappling with the possibility that the Islamic Republic could be overthrown — a seminal event that would transform global geopolitics and energy markets.

The regime of Supreme Leader Ayatollah Ali Khamenei has weathered bouts of protests many times, but demonstrations that began two weeks ago are spreading — by some accounts, hundreds of thousands of people defied authorities’ threats and a brutal crackdown to take to the streets over the weekend, from the capital Tehran to dozens of other cities across the nation of 90 million. They are being cheered on by President Donald Trump, fresh off the capture of Venezuela’s Nicolas Maduro, and the US leader has in recent days repeatedly threatened to strike Iran, suggesting that America is back in the regime change business.

World leaders and investors are watching closely. US commanders have briefed Trump on options for military strikes, according to a White House official. Brent crude surged more than 5% on Thursday and Friday to over $63 a barrel as investors priced in the possibility of supply disruptions in OPEC’s fourth-biggest producer.

“This is the biggest moment in Iran since 1979,” said William Usher, a former senior Middle East analyst at the Central Intelligence Agency, referring to the revolution that birthed the Islamic Republic, upended the balance of power in the region and led to decades of rancor between Tehran and the US and its allies. “The regime is in a very tough spot right now and the primary driver is the economy. I think they have a narrowing window to reassert control and a diminished toolset to do it.”

More than 500 protesters have been killed in the past two weeks, according to the AP, citing the US-based Human Rights Activists News Agency, and more than 10,000 have been arrested in demonstrations triggered by a currency crisis and economic collapse, but now also focused on the regime.

Authorities have tried to block the internet and telephone networks since Thursday, as they seek to quell Iranians’ growing outrage over government corruption, economic mismanagement and repression. Foreign airlines have canceled flights to the country.

Trump’s repeated warnings to Iran that the US will strike if it kills peaceful protesters come as the president escalates his assault on the post-World War II global order in a stunning assertion of American power that’s included claiming Venezuela’s oil after seizing Maduro, and threatening to take over Greenland from NATO ally Denmark.

Israel, which battered Iran during a US-assisted 12-day air war in June, is liaising closely with European governments about the situation on the ground, according to a senior European official, who asked not to be named discussing private talks. 

If the regime does fall, it would be a blow to Russian President Vladimir Putin, who would lose another foreign ally after Maduro this month and the overthrow of Syria’s Bashar al-Assad just over a year ago, the official added.

The stakes for oil traders are significant. But it’s unclear if Khuzestan, the main oil-pumping province, has seen unrest and so far there are no signs of reduced crude exports. On Saturday, Reza Pahlavi, the son of the former shah who’s exiled in the US and positioning himself as an opposition leader, urged petroleum workers to strike. Oil strikes in 1978 were one of the death knells of his father’s monarchy because of how they immediately hit the economy.

The market’s “focus has now shifted to Iran,” said Arne Lohmann Rasmussen, chief analyst at A/S Global Risk Management, which helps clients manage volatility in energy markets. “There is also growing concern in the market that the US, with Trump at the helm, could exploit the chaos to attempt to overthrow the regime, as we have seen in Venezuela.”

The White House is on a high after the tactical success of the operation against Maduro, as well as Trump’s decision to bomb Iranian nuclear facilities at the end of the 12-day war. American officials are also increasing pressure on Denmark to cede control of Greenland, signaling the administration has the appetite for more forays abroad.

Read More: Trump’s Ousting of Maduro Shows His New World Order Is Here

Trump may well be tempted, for all the risks, to try to topple a government that’s been an archenemy to the US and Israel for over 45 years. 

“The balance of power would change dramatically,” Mark Mobius, the veteran emerging markets investor, said of the downfall of the Islamic Republic. “The best outcome would be a complete change in the government. The worst outcome would be continued internal conflict and a continuing rule by the current regime.”

Trump at times ran against American adventurism in the region, where the ousting of longtime US enemy Saddam Hussein in Iraq unleashed a generation of chaos and terrorism, costing hundreds of thousands of lives and trillions of dollars.

It’s just that kind of potential power vacuum that’s worrying Arab leaders in the Gulf Cooperation Council, according to regional officials. While the group — which includes Saudi Arabia, the United Arab Emirates and Qatar — has often viewed Iran as an adversary, its members have sought to improve ties in recent years to ensure Tehran doesn’t lash out against any Israeli or US military action by attacking them. The specter of the Arab Spring, where dictators fell across the region only for chaos to follow, looms large.

Iran has warned that if it’s attacked, American assets in the region — where it has deep commercial ties and tens of thousands of troops stationed — and Israel will be “legitimate targets for us.”

Read More: How Sanctions and a Currency Crash Fueled Iran Unrest

The Islamic Republic has been severely weakened in the past two years, thanks to its stagnating economy, rampant inflation and Israel striking both it and its proxies. But it retains a large and sophisticated arsenal of ballistic missiles able to hit targets across the Middle East, from military bases to oil installations, and the regime still has the backing of the country’s myriad security forces, including the all-important Islamic Revolutionary Guard Corps.

For the GCC and the likes of Turkey and Pakistan, the worst outcome would be chaos in Iran, said Ellie Geranmayeh, deputy program director for the Middle East and North Africa at the European Council on Foreign Relations. It’s an eventuality made more possible by the sheer diversity of Iranian protesters, who include everyone from urban, secular elites to religious conservatives and lack a unifying leader.

“With the GCC reconciliation of the past few years with Tehran, there’s a sense of better the devil you know rather than complete chaos or an unknown power structure that is alien to them,” said Geranmayeh.

US and Israeli strikes might even strengthen the government and reduce the appeal of the protest movement. In June, there was a surge in nationalism as the Jewish state and Washington rained down bombs.

The Islamic Republic probably won’t survive in its current form by the end of 2026, according to Dina Esfandiary, a Middle East analyst at Bloomberg Economics. The most likely scenario, she said, is a leadership reshuffle that largely preserves the system or a coup by the IRGC, which could mean greater social freedom — the organization is run by generals rather than clerics — but less political liberty and a more militaristic foreign policy.

The chances of a revolution are still fairly low, she said.

“A collapse appears unlikely for now,” she said. “Iranians are frightened of chaos, having seen it wreak havoc in neighboring Iraq and Syria. More importantly, the government is cracking down hard.”

On Sunday, President Masoud Pezeshkian, a former heart surgeon and a moderate relative to others at the top of the Iranian government, struck a conciliatory note, offering condolences to families affected by the “tragic consequences.”

“Let’s sit down together, hand in hand, and solve the problems,” he said on state TV.

It’s unlikely many protesters will believe him. The supreme leader, a much more powerful figure, as well as members of the security forces, are increasingly bellicose, floating the death penalty and making clear they’re prepared to respond as they always have — with brutal force.

“I don’t think a collapse of the regime would be pretty,” said Usher, the former CIA analyst. “Short-term, I could imagine some fracturing of the country as ethnic minority groups and some provinces pursue autonomy from Tehran. The IRGC will fight vigorously to save the regime so I think there’d be strong possibility for large-scale violence.”



Source link

Continue Reading

Business

Britain is in talks with NATO to boost Arctic security, agreeing with Trump on Russia and China

Published

on



Britain is discussing with NATO allies how it can help beef up security in the Arctic to counter threats from Russia and China, a government minister said Sunday.

Transport Secretary Heidi Alexander said the talks are “business as usual” rather than a response to recent threats by U.S. President Donald Trump to take over Greenland.

Trump said Friday that he would like to make a deal to acquire Greenland, a semiautonomous region of NATO ally Denmark, to prevent Russia or China from taking it over.

“We are going to do something on Greenland whether they like it or not,” Trump said Friday.

Greenland, with a population of around 57,000, is defended by Denmark, whose military is dwarfed by that of the U.S., which has a military base on the island. Denmark’s prime minister has warned that a takeover would threaten NATO.

The U.K. agrees with Trump that Russia and China are increasingly becoming more competitive in the Arctic Circle, Alexander said.

“Whilst we haven’t seen the appalling consequences in that part of the world that we’ve seen in Ukraine, it is really important that we do everything that we can with all of our NATO allies to ensure that we have an effective deterrent in that part of the globe against (Russian President Vladimir) Putin,” Alexander told the BBC.

Britain’s former ambassador to the U.S., Peter Mandelson, who was sacked last year because of his friendship with the disgraced financier Jeffrey Epstein, said he did not think Trump would take Greenland by force.

“He’s not a fool,” Mandelson said. “We are all going to have to wake up to the reality that the Arctic needs securing against China and Russia. And if you ask me who is going to lead in that effort to secure, we all know, don’t we, that it’s going to be the United States.”

Ed Davey, leader of the Liberal Democrat Party, suggested Britain offer to deploy troops to Greenland in a joint command with Denmark.

“If Trump is serious about security, he’d agree to participate and drop his outrageous threats,” Davey said. “Tearing the NATO alliance apart would only play into the hands of Putin.”

It’s unclear how remaining NATO members would respond if the U.S. decided to forcibly take control of the island or if they would come to Denmark’s aid.



Source link

Continue Reading

Trending

Copyright © Miami Select.