BNP Paribas Deploys Internal AI Tool to Streamline Investment Banking
BNP Paribas has introduced a proprietary AI-powered tool, IB Portal, to streamline the creation of client pitches for its investment banking division. The tool operates as a sophisticated search engine, scanning the bank’s historical pitch materials to identify relevant slides and analyses, aiming to reduce repetitive tasks. According to the bank, this innovation can shorten research time by days, enabling teams to focus more on strategic development. The system is built on the bank’s internal “LLM as a Service” platform, which leverages multiple large language models within a secure, private infrastructure. All information generated by the tool requires human review to ensure accuracy and the proper handling of sensitive data.
Sources:
MIT’s DisCIPL System Enables Small AI Models to Tackle Complex Reasoning
MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) has developed DisCIPL, a novel system that enables small language models to collaborate to solve complex reasoning tasks under strict constraints. The framework utilizes a large language model as a “planner” to strategize and delegate sub-tasks to smaller, more efficient “follower” models. This approach has demonstrated higher accuracy than leading LLMs like GPT-4o on specific tasks and approaches the precision of top reasoning systems while being more cost-effective. In experiments, DisCIPL achieved an 80.2% cost reduction over comparable reasoning models. The system functions by translating complex requests into a formal programming language, which guides the smaller models to produce accurate and compliant results.
Sources:
G&J Pepsi Expands Autonomous Vehicle Fleet with Cyngn’s DriveMod Technology
Cyngn, a developer of AI-powered autonomous vehicle (AV) technology, announced an expanded agreement with G&J Pepsi, the largest independent Pepsi bottler in the United States. G&J Pepsi has placed a new purchase order for its DriveMod Tugger program, significantly expanding its fleet of autonomous industrial vehicles. This decision to acquire more vehicles before finalizing their deployment locations signals strong confidence in the AV technology’s performance and its applicability across diverse workflows. The deal expands upon a previous deployment at a single bottling plant announced in October. Cyngn’s technology aims to address critical industrial challenges like labor shortages and safety incidents.
Johns Hopkins Launches HopGPT, a Secure Generative AI Platform for Research and Education
Johns Hopkins University has launched HopGPT, a secure, in-house generative AI platform for its students, faculty, and staff. The platform provides secure, authenticated access to leading AI models like ChatGPT, Claude, and Llama through a university-managed interface. Formerly known as the Hopkins AI Lab, HopGPT is designed to support research, learning, teaching, clinical, and administrative work across the institution. A key feature is its approval for use with sensitive data, including Protected Health Information (PHI) and Personally Identifiable Information (PII), under the university’s strict data protection policies. The university has also launched a website with training, tutorials, and responsible-use guidelines to support the community in using these new AI tools effectively and ethically. HopGPT is available free of charge to all members of the Johns Hopkins community, aiming to keep the university at the forefront of innovation.
Sources:
AWS re:Invent 2025: Key Announcements Include Graviton5, Lambda Managed Instances, and GuardDuty Updates
At re:Invent 2025, AWS announced several major updates, including the new Graviton5 processor, which delivers up to 25% better performance than its predecessor. A significant announcement for developers was the introduction of Lambda Managed Instances, a new deployment model allowing Lambda functions to run on dedicated EC2 instances, effectively eliminating cold starts and optimizing costs for predictable, high-volume workloads. In security, Amazon GuardDuty Extended Threat Detection now covers Amazon EC2 instances and Amazon ECS tasks, providing a more unified view of potential multi-stage attacks. Additionally, Amazon S3 Tables now support the Intelligent-Tiering storage class and automatic replication across AWS Regions to optimize storage costs and simplify data management.
Sources:
Qilimanjaro and Oxigen Data Center to Integrate Quantum Computers into HPC Environments
Barcelona-based Qilimanjaro Quantum Tech is collaborating with Oxigen Data Center to integrate multimodal quantum computers directly into commercial high-performance computing (HPC) data center environments. This strategic partnership aims to establish the technical and operational standards required to house both analog and digital quantum systems alongside classical HPC infrastructure. The collaboration will focus on creating a hybrid quantum-classical framework to prepare for the next generation of quantum cloud services in Europe. Qilimanjaro’s analog quantum computers, which are well-suited for complex simulations and AI, will be a key component of the integrated system. This initiative aims to position Barcelona as a key European hub for the emerging quantum cloud ecosystem.
Sources:
Western Digital Strategically Invests in Qolab to Scale Quantum Hardware Manufacturing
Quantum hardware company Qolab has secured a strategic investment from data storage leader Western Digital to accelerate the manufacturing of scalable quantum computing systems. The partnership will leverage Qolab’s innovative superconducting qubit designs with Western Digital’s deep expertise in materials science, precision manufacturing, and large-scale nanofabrication. The collaboration aims to develop next-generation fabrication processes to enhance qubit performance, reliability, and scalability, transitioning quantum hardware from the laboratory to real-world applications. This investment highlights the growing interest from established technology companies in the commercialization of quantum computing. The joint effort is also expected to bolster U.S. leadership in semiconductor research and create technology jobs in California.
Sources:
Tantalum-Based Qubits Achieve Breakthrough in Quantum Coherence Time
Scientists have developed a new fabrication method for superconducting qubits that significantly extends their coherence time—a critical metric for building powerful quantum computers. The new technique utilizes the chemical element tantalum on a specialized silicon substrate, creating some of the most coherent superconducting qubits ever demonstrated. Researchers demonstrated coherence times reaching up to 1.68 milliseconds, which is reportedly up to 15 times longer than the superconducting qubits used in some of today’s leading quantum processors. The improved performance is attributed to tantalum’s resistance to contaminants and corrosion during manufacturing. This breakthrough, detailed in the journal Nature, could pave the way for more reliable and powerful quantum computers.
Sources:
The Four Biggest Open Source Trends of 2025: AI, Security, and Licensing
The year 2025 in open source has been defined by four pivotal trends: the influence of artificial intelligence, the evolution of licensing and governance, a heightened focus on software supply chain security, and significant shifts in commercial open-source business models. These areas have shaped the most important developments and discussions within the open-source community. The pervasive influence of AI, in particular, has been a recurring theme across the open-source landscape.
Sources: