Brad Menezes, CEO of enterprise vibe coding startup Superblocks, believes the next crop of billion-dollar startup ideas are hiding in almost plain sight: the system prompts used by existing unicorn AI startups. So as part of his own startup’s new product announcement of an enterprise coding AI agent named Clark, Superblocks offered to share a file of 19 system prompts from some of the most popular AI coding products like Windsurf, Manus, Cursor, Lovable and Bolt. Role prompting helps the LLMs be consistent, giving both purpose and personality. Contextual prompting gives the models the context to consider before acting. It should provide guardrails that can, for instance, reduce costs and ensure clarity on tasks. Tool use enables agentic tasks because it instructs the models how to go beyond just generating text. Replit’s, for instance, is long and describes editing and searching code, installing languages, setting up and querying PostgreSQL databases, executing shell commands and more. Studying others’ system prompts helped Menezes see what other vibe coders emphasized. Tools like Loveable, V0, and Bolt “focus on fast iteration,” he said, whereas “Manus, Devin, OpenAI Codex, and Replit” help users create full-stack applications but “the output is still raw code.” Menezes saw an opportunity to let non-programmers write apps, if his startup could handle more, such as security and access to enterprise data sources like Salesforce.
allpay becomes the first signatory of Enfuce’s Fortitude Pledge, a compliance and security standard designed to eliminate 100% of financial crime risks across all card transactions
allpay Limited, the UK’s leading payment solutions provider, has become the first company to sign the Fortitude Pledge, a bold new compliance and security standard launched by issuer processing powerhouse Enfuce, designed to eliminate 100% of financial crime risks across all card transactions. The Fortitude Pledge is a proactive commitment to move beyond checkbox compliance and take full responsibility in the fight against financial crime. From screening every card transaction to training every employee, the Fortitude Pledge sets a new benchmark of zero tolerance for financial crime, including money laundering, terrorist financing, and human trafficking. allpay and Enfuce have joined forces across two critical fronts, card manufacturing and payment processing, leveraging their combined strengths to deliver secure, scalable, and future-ready solutions across the UK. By integrating Enfuce’s cloud-based technology, allpay is modernising public sector payments across the UK, making them more secure, efficient, and less vulnerable to fraud. At the same time, allpay is powering card manufacturing and personalisation for Enfuce’s MyCard solution, enabling highly secure, seamless, and personalised card experiences.
Non-profit EleutherAI releases massive AI training dataset of licensed and open domain text created in consultation with legal experts and claims performing on par with models developed using unlicensed, copyrighted data
EleutherAI, an AI research organization, has released what it claims is one of the largest collections of licensed and open-domain text for training AI models called the Common Pile v0.1. Weighing in at 8 terabytes in size, the Common Pile v0.1 was used to train two new AI models from EleutherAI, Comma v0.1-1T and Comma v0.1-2T, that EleutherAI claims perform on par with models developed using unlicensed, copyrighted data. The Common Pile v0.1, which can be downloaded from Hugging Face’s AI dev platform and GitHub, was created in consultation with legal experts, and it draws on sources, including 300,000 public domain books digitized by the Library of Congress and the Internet Archive. EleutherAI also used Whisper, OpenAI’s open source speech-to-text model, to transcribe audio content. EleutherAI claims Comma v0.1-1T and Comma v0.1-2T are evidence that the Common Pile v0.1 was curated carefully enough Non-profit alternatives. According to EleutherAI, the models, both of which are 7 billion parameters in size and were trained on only a fraction of the Common Pile v0.1, rival models like Meta’s first Llama AI model on benchmarks for coding, image understanding, and math.
Amperity vibe coding AI agent connects directly to the customer’s Databricks environment via native compute and LLM endpoints to quickly execute complex tasks such as identity resolution
Customer data cloud startup Amperity Inc. is joining the agentic AI party, launching Chuck Data, an AI agent that specializes in customer data engineering. Chuck Data is trained on massive volumes of customer data from more than 400 enterprise brands. This “critical knowledge” base allows it to execute tasks such as identity resolution and personally identifiable information tagging autonomously and instantly resolve customer identities, with minimal input from human developers. The agent is designed to help companies dig up customer insights much faster. Chuck Data makes it possible for data engineers to embrace “vibe coding,” so they can use natural language prompts to delegate these manual coding tasks to an autonomous AI assistant. The company said Chuck Data connects directly to the customer’s Databricks environment via native compute and large language model endpoints. Then it can quickly execute complex tasks such as identity resolution – which involves pulling data from multiple profiles into one – as well as compliance tagging and data profiling. One of Chuck Data’s core features is Amperity’s patented identity resolution algorithm, which is based on the proprietary Stitch technology that’s used within its flagship cloud data platform. The company said users can run Stitch on up to 1 million customer records for free, and for those with bigger records, they can sign up to Chuck Data’s research preview program to access free credits. It’s also offering paid plans that unlock unlimited access to Stitch, enabling companies to create millions of accurate, scalable customer profiles. huck Data provides yet more evidence of how CDPs are evolving from activation tools into embedded intelligence layers for the customer engagement data value chain.
Hirundo’s approach to AI hallucinations is about making fully trained AI models forget the bad things they learn, so they can’t use this mistaken knowledge
Hirundo AI Ltd., a startup that’s helping AI models “forget” bad data that causes them to hallucinate and generate bad responses, has raised $8 million in seed funding to popularize the idea of “machine unlearning.” Hirundo’s approach to AI hallucinations is about making fully trained AI models forget the bad things they learn, so they can’t use this mistaken knowledge to generate their responses later on, down the line. It does this by studying the behavior of AI models in order to locate the directions users can go in order to manipulate them. It identifies any bad traits, then investigates the root cause of those bad outputs, before steering the model away from them. It pinpoints where hallucinations originate from in the billions of parameters that make up their knowledge base. This retroactive approach to fixing undesirable behaviors and inaccuracies in AI models means it’s possible to improve their accuracy and reliability without needing to retrain them. That’s a big deal, because retraining models can take many weeks and cost thousands or even millions of dollars. “With Hirundo, models can be remediated instantly at their core, working toward fairer and more accurate outputs,” Chief Executive Ben Luria added. Besides helping models to forget bad, biased or skewed data, the startup says it can also make them “unlearn” confidential information, preventing AI models from revealing secrets that shouldn’t be shared. What’s more, it can do this for both open-source models such as Llama and Mistral, and soon it will also be able to do the same for gated models such as OpenAI’s GPT and Anthropic PBC’s Claude. The startup says it has successfully managed to remove up to 70% of biases from DeepSeek Ltd.’s open-source R1 model. It has also tested its software on Meta Platforms Inc.’s Llama, reducing hallucinations by 55% and successful prompt injection attacks by 85%.
KILT’s decentralized document signing solution lets users create verifiable, privacy-preserving signatures using credentials stored on their own devices without compromising control over data
KILT Protocol is launching a suite of next-generation products to bring decentralized identity (DeID) into mainstream use. The company will bring its core identity infrastructure to the Ethereum Virtual Machine (EVM) world in 2025, allowing developers to integrate KILT credentials into smart contracts and applications with minimal overhead. KILT’s consumer products include Clans, Sporran 2.0, DIDsign 2.0, KILT Pay, and DID-as-a-Service. Clans is a mobile-first platform that rewards users for engaging with content and social campaigns, while Sporran 2.0 enables users to manage credentials, sign documents, and access services using decentralized sign-on. DIDsign 2.0 introduces decentralized document signing, while KILT Pay bridges the gap between credentials and financial transactions. The KILT Foundation will support these developments, focusing on growing the ecosystem and expanding partnerships.
SEC Chair Paul Atkins wants to let DeFi thrive with fewer rules and a potential “innovation exemption” aimed at protecting developers
The U.S. SEC may soon ease the regulatory burden on decentralized finance platforms as Chairman Paul Atkins outlines a potential “innovation exemption” aimed at protecting developers and enabling new blockchain-based systems to thrive. In the final session of a five-part crypto roundtable series, SEC Chairman Paul Atkins signaled a notable shift in regulatory tone, especially regarding decentralized finance (DeFi). Atkins said he has directed SEC staff to explore exemptions or guidance that would let DeFi platforms operate with fewer barriers. The proposal seeks to support on-chain financial systems and reflect the technological shift toward decentralized models. He emphasized that this principle should not vanish online, especially in a financial ecosystem increasingly powered by decentralized technologies. The comments mark a stark contrast with previous SEC leadership, which leaned heavily on enforcement and broad interpretations of securities laws. He rejected the notion that writing code constitutes a regulated activity if that code enables financial transactions. Commissioner Hester Peirce echoed this view, warning against infringing on First Amendment rights. Atkins called for reevaluating legacy frameworks and asked staff to assess whether new guidance or rulemaking would help entities interact with DeFi tools while remaining compliant.
Iskay Quantum Optimizer outperforms popular classical optimization solvers in financial use cases such as portfolio optimization
IBM Quantum partners are attending the IBM Quantum Partner Forum 2025 in London, England, to hear from IBM leadership and researchers about the latest in quantum hardware, quantum algorithm discovery, and powerful new software tools. The company is proud to introduce two new application functions on the Qiskit Functions Catalog: the QUICK-PDE function by French quantum startup ColibriTD, and the Quantum Portfolio Optimizer by Spanish startup Global Data Quantum. These new functions provide a full, ready-made quantum pipeline for researchers and developers to harness the full power of utility-scale quantum computers in researching and developing new quantum use cases. Application functions are services that abstract away the complexities of the quantum workflow to accelerate quantum algorithm discovery and application prototyping. They take the same classical inputs as in a typical classical workflow and return domain-familiar classical outputs, making it easy to integrate quantum methods into pre-existing application workflows. As the hunt for quantum advantage progresses, more researchers will use application functions to tackle problems that are challenging or impossible for the most powerful high-performance computing (HPC) systems. The new application functions include the Iskay Quantum Optimizer by Kipu Quantum, which outperforms popular classical optimization solvers in financial use cases such as portfolio optimization, and the Singularity Machine Learning function by Multiverse Computing, which addresses classification problems that benefit from ensemble learning and complex model optimization. The new QUICK-PDE and Quantum Portfolio functions already incorporate all three of these improvements, and users can request a free trial through the Qiskit Functions Catalog homepage.
Xanadu is building modular and networked quantum computer using complex photonic states (known as GKP) with extremely low optical losses on a silicon-based chip platform to achieve scalable fault-tolerant quantum computing
Xanadu has taken a key step toward scalable fault-tolerant quantum computing by demonstrating the generation of error-resistant photonic qubits — known as GKP states — on a silicon-based chip platform, a first-of-its-kind achievement. The milestone positions the Toronto-based quantum startup closer to building a modular and networked photonic quantum computer, a device that uses photons, rather than electrons, to perform calculations, according to the paper and a company statement. By encoding quantum information into complex photon states that can withstand noise and loss, the work directly addresses one of the central obstacles to quantum scalability: preserving fragile quantum data as systems grow in size and complexity. Xanadu’s researchers generated what are known as Gottesman–Kitaev–Preskill (GKP) states — structured quantum states made of many photons arranged in specific superpositions. These states encode information in a way that makes it possible to detect and correct small errors, such as phase shifts or photon loss, using well-known quantum error correction techniques. Xanadu’s experiment demonstrates that GKP states can be produced directly on-chip using integrated photonics, paving the way for scalable manufacturing. The system is based on silicon nitride waveguides fabricated on 300 mm wafers, a format common in commercial semiconductor manufacturing. These waveguides exhibit extremely low optical losses, a critical requirement for preserving quantum coherence over time. In addition to the waveguide platform, the setup included photon-number-resolving detectors with over 99% efficiency, developed in-house by Xanadu. These detectors can distinguish between one photon and many, a capability essential for preparing and verifying complex photonic states like GKP. High-precision alignment, custom chip mounts, and loss-optimized fiber connections ensured that the quantum states could be routed and measured without degrading the delicate information they carried.
Apple debuts new user design experience Liquid Glass, which will bring greater focus to content, deliver a new level of quality to controls and keep users more attuned to what’s happening on screen “harmonizing” the user experience across all devices
Apple previewed a slick new software design and powerful software updates, including new features coming to its next-generation operating systems across devices that will receive a unified version 26. The new design features a new material called Liquid Glass, which creates a translucent effect similar to water that sits atop the display, refracting content below it and allowing colors to flow through. The company says this will bring greater focus to content, deliver a new level of quality to controls and keep users more attuned to what’s happening on screen. The new design extends across Apple’s entire device ecosystem, including iOS 26, iPadOS 26, macOS Ventura 26, watchOS 26 and tvOS 26. The company said the idea was to “harmonize” the user experience across all devices so they could expect every device to look and feel the same. It will affect buttons, switches, sliders, text and media in the user interface and shift dynamically according to user needs. Controls, toolbars and navigation within apps have been redesigned with rounded corners and they “float above” content so that they stay out of the way and avoid interrupting content. They also shift into thoughtful groupings, allowing users to find the controls they need. The Preview app, which originally comes from macOS, is coming to iPadOS 26. Preview is a dedicated app for creating a quick sketch, as well as viewing, editing and marking up PDFs and images with Apple Pencil or by touch.