• Project Flux
  • Posts
  • OpenAI's 2028 AI Researcher Bombshell: Are WeReady for Autonomous Discovery?

OpenAI's 2028 AI Researcher Bombshell: Are WeReady for Autonomous Discovery?

Anthropic conquers finance whilst Nvidia hits $5 trillion and UK planning gets an AI overhaul—plus construction's biggest merger yet.

AuthorAuthor

Yoshi Soornack & James Garner
November 04, 2025

Proudly sponsored by ConstructAI, brought to you by Weston Analytics.

Hello Project AI enthusiasts,

OpenAI just announced they're targeting fully autonomous AI researchers by 2028—systems capable of conducting original scientific breakthroughs without human intervention. Meanwhile, Anthropic is quietly dominating enterprise AI with their finance-focused Claude, Nvidia became the first company to hit $5 trillion, and the UK government is finally tackling the planning bottleneck with AI trials.

Oh, and construction tech just witnessed its most significant consolidation yet with OpenSpace acquiring Disperse. If you're in project delivery, this week's developments aren't just interesting—they're career-defining.

In This Edition

Flux check-in

The AI race just shifted into overdrive. OpenAI CEO Sam Altman announced an ambitious roadmap: an intern-level AI research assistant by September 2026, followed by a fully autonomous AI researcher by 2028 capable of conducting original scientific research end-to-end.

This isn't about automating spreadsheets—it's about AI systems that can formulate hypotheses, design experiments, and publish breakthrough discoveries in medicine, physics, and technology.

Meanwhile, Anthropic launched Claude for financial services, embedding deeply into enterprise stacks with specialised connectors to platforms like LSEG and Moody's, plus finance-specific skills like discounted cash flow modelling. Read the full breakdown →

What Does This Mean for Me?

This represents a fundamental shift in how we approach complex problem-solving. Imagine AI systems that don't just assist with risk analysis or schedule optimisation, they actively conduct research into new methodologies, test innovative approaches to project challenges, and generate evidence-based recommendations. The distinction between "AI tool" and "AI colleague" is dissolving rapidly. Those who learn to collaborate with these systems now—defining problems, evaluating AI-generated research, and integrating insights into decision-making, will have an enormous advantage. The question isn't whether AI will transform your role; it's whether you'll be leading that transformation or scrambling to catch up.

Key Themes

  • Autonomous research capabilities: AI systems conducting end-to-end scientific investigations without human intervention

  • Enterprise AI dominance: Anthropic's API revenue reportedly surpassing OpenAI through deep enterprise integration

  • Specialised financial tools: Claude for Excel with finance-specific connectors and analytical capabilities

  • Timeline acceleration: Intern-level AI assistants expected by September 2026, full researchers by 2028

Down the Rabbit Hole

AECOM has urged the UK government to appoint a dedicated infrastructure minister and embrace AI to accelerate project delivery. This comes as the government launched its AI Skills Tools Package, a comprehensive initiative designed to upskill the UK workforce for the AI era. The package includes practical resources, training frameworks, and guidance specifically tailored for sectors like construction and infrastructure. Read the full breakdown →

What Does This Mean for Me?

The UK government is finally acknowledging what we've known for years: AI adoption in infrastructure isn't optional, it's essential for national competitiveness. The AI Skills Tools Package represents a genuine opportunity to upskill your team without massive investment. For project managers and delivery leaders, this is your moment to advocate for AI training budgets and pilot programmes. AECOM's call for an infrastructure minister signals growing recognition that project delivery needs dedicated political attention. If you're working on major infrastructure projects, expect increased scrutiny on AI adoption, digital delivery methods, and skills development. Those who can demonstrate AI competency and drive digital transformation will be positioned for leadership roles as the sector modernises.

Key Themes

  • Political advocacy: AECOM pushing for dedicated infrastructure minister to prioritise project delivery •

  • National AI upskilling: Government's comprehensive AI Skills Tools Package targeting workforce transformation •

  • Sector-specific training: Resources tailored for construction, infrastructure, and project delivery professionals •

  • Competitive imperative: AI adoption framed as essential for UK's global infrastructure competitiveness

Down the Rabbit Hole

OpenSpace has acquired Disperse, the AI-powered progress tracking platform, in what may be the most significant consolidation in construction technology to date. OpenSpace's 360° reality capture technology combined with Disperse's computer vision and progress tracking creates an end-to-end solution for automated site monitoring, progress verification, and quality control. This isn't just a merger—it's a signal that construction AI is maturing from point solutions to integrated platforms. Read the full breakdown →

What Does This Mean for Me?

The traditional site visit, clipboard in hand, manually checking progress against drawings, is becoming obsolete. This acquisition creates a platform that automatically captures site conditions, tracks progress against schedules, identifies quality issues, and flags deviations from design.

For project managers in construction, this means dramatically reduced time spent on site inspections and progress reporting. For site engineers, it means shifting from manual documentation to exception management, focusing on problems the AI identifies rather than hunting for them. The consolidation also signals that construction tech is moving towards fewer, more comprehensive platforms rather than dozens of specialised tools. If you're evaluating construction tech, prioritise platforms with broad capabilities and integration potential over narrow point solutions.

Key Themes

  • Industry consolidation: Major acquisition signals construction AI market maturation and platform convergence •

  • Automated progress tracking: Integration of reality capture with AI-powered progress verification •

  • End-to-end visibility: Combined solution covers capture, analysis, tracking, and quality control •

  • Reduced site visits: Technology enabling remote monitoring and exception-based site management

Down the Rabbit Hole

Nvidia became the first public company to reach a $5 trillion valuation, driven by insatiable demand for its AI chips from OpenAI, Palantir, and major enterprises worldwide. CEO Jensen Huang announced plans for next-generation chips and a $500 billion revenue target through 2026. Not to be outdone, AMD secured a $1 billion contract with the US Department of Energy to build two AI supercomputers at Oak Ridge National Laboratory—the first arriving in early 2026 as America's first dedicated "AI Factory" for training foundation models aimed at scientific discovery. Read the full breakdown →

What Does This Mean for Me?

The AI infrastructure race directly impacts the tools you'll use tomorrow. Nvidia's $5 trillion valuation and AMD's billion-dollar supercomputer contracts signal that AI computing power will continue to expand dramatically, making today's cutting-edge models look primitive within months. For project delivery professionals, this means the AI tools you adopt today will become significantly more capable—and affordable—as competition intensifies.

The Department of Energy's investment in AI supercomputers for scientific discovery suggests that AI will soon tackle complex engineering and project delivery challenges currently beyond its reach. Budget for continuous AI tool evaluation and be prepared to upgrade or switch platforms as capabilities evolve. The question isn't whether to invest in AI—it's how to stay agile as the technology landscape shifts beneath your feet.

Key Themes

  • Historic valuation milestone: Nvidia becomes first company to reach $5 trillion market capitalisation

  • Government AI investment: US Department of Energy committing billions to AI supercomputing infrastructure

  • Scientific AI applications: Dedicated "AI Factory" for training foundation models for research breakthroughs

  • Competitive dynamics: AMD challenging Nvidia's dominance with energy-efficient supercomputer designs

Down the Rabbit Hole

The UK is launching AI trials to accelerate planning application processing, targeting the notorious 18-month bottleneck that has strangled infrastructure and construction projects for years. The trials will use AI to analyse planning documents, identify issues, flag inconsistencies, and expedite decision-making. If successful, this could fundamentally reshape how projects move from concept to construction. Read the full breakdown →

What Does This Mean for Me?

Planning delays have been the silent killer of project delivery for decades—derailing schedules, inflating costs, and creating uncertainty that ripples through entire programmes. AI-powered planning review could compress 18-month approval processes into weeks, fundamentally changing project timelines and risk profiles. For project planners and programme managers, this means rethinking how you structure early-stage activities. If planning approvals accelerate dramatically, the bottleneck shifts to design readiness and stakeholder engagement. Start preparing higher-quality planning submissions now—AI systems will likely be less forgiving of incomplete or poorly structured applications than human reviewers. The trials also signal government willingness to embrace AI in regulatory processes, potentially opening doors for AI-assisted building control, environmental impact assessment, and other approval workflows.

Key Themes

  • Planning bottleneck solution: AI trials targeting 18-month approval delays that plague UK projects

  • Automated document analysis: AI reviewing planning applications for completeness and compliance

  • Regulatory AI adoption: Government embracing AI for public sector decision-making processes

  • Project timeline impact: Potential for dramatically compressed approval processes reshaping delivery schedules

Down the Rabbit Hole

If you work in project delivery you need to understand how AI is reshaping it. Renowned industry thinker — and friend of Project Flux — Antony Slumbers has just launched Cohort 14 of his acclaimed Generative AI in Real Estate course (starting 7 November). Over three weeks, you’ll master frontier tools, reconfigure workflows, and reimagine the future of property. Expect real-world case studies, hands-on sessions, and a network of innovators shaping what’s next.

👉 Join the course here and stay ahead of the curve.

The pulse check

Tips of the week

Getting AI to Challenge Your Thinking

A common frustration with AI tools is that they often agree with whatever you say. That’s not very helpful when what you really want is sharp, honest feedback.

The trick is to adjust your AI’s custom instructions. Try adding this:

“Provide incisive feedback that pushes the boundaries of my thinking. Challenge my assumptions while showing genuine curiosity and collaboration.”

This simple tweak can transform AI from a polite assistant into a genuine thought partner — one that tests your ideas and helps you think deeper.

Governance & Security

OpenAI completed its transition from nonprofit to a public benefit corporation, establishing the OpenAI Foundation which holds $130 billion in equity earmarked for health research and AI resilience. Microsoft's ownership stake adjusted to approximately 27%, with the new terms allowing both companies to pursue AGI with other partners whilst Microsoft retains IP rights through 2032. An independent expert panel will verify any AGI claims—a crucial accountability mechanism as we approach potential artificial general intelligence.

On the security front, OpenAI launched Aardvark, an autonomous AI agent based on GPT-5 that continuously scans codebases to detect, validate, and patch security vulnerabilities. In testing, it detected 92% of known and synthetic vulnerabilities, demonstrating high recall and effectiveness in complex environments. The company also open-sourced gpt-oss-safeguard, a pair of 120B and 20B parameter models designed for safety classification tasks, allowing easier updating and refinement of safety rules without retraining. These moves towards transparency and automated security represent important steps in responsible AI development.

However, significant risks remain. Security researchers, including Brave, revealed that AI browser agents like OpenAI's Atlas are vulnerable to prompt injection and data exfiltration attacks, with malicious websites potentially hijacking AI assistants to perform unauthorised actions. Character AI announced it will prohibit users under 18 from engaging in open-ended conversations with its AI chatbots starting November 25, 2025, following legal pressures linked to teen deaths. Meanwhile, OpenAI updated GPT-5 to better handle mental health emergencies, achieving 91% compliance with mental health protocols, up from 77% in GPT-4o—though ethical challenges and regulatory scrutiny continue to intensify.

The competitive landscape is also shifting dramatically. OpenAI focuses on consumer subscription revenue through ChatGPT, whilst Anthropic has reportedly surpassed OpenAI in API revenue by embedding AI deeply into enterprise stacks. Analysts suggest both approaches may succeed, mirroring Meta's consumer platform and Amazon's cloud infrastructure business models. As AI capabilities expand, the tension between innovation, safety, and regulation will only intensify—making governance and security expertise increasingly valuable for organisations deploying AI at scale.

Robotics

Amazon Introduces Blue Jay Multi-Arm Robot for Warehouse Automation
Amazon unveiled Blue Jay, a multi-arm robotic system that combines picking, stowing, and consolidating tasks into a single compact workspace. Developed with AI-driven digital twins, Blue Jay reduces physical strain on workers and accelerates order processing, currently in trials at a South Carolina facility. Read more

Unitree Robotics Releases Advanced Humanoid Robot H2 with Enhanced Mobility
Chinese company Unitree Robotics unveiled the H2 humanoid robot, featuring 31 degrees of freedom, enhanced flexibility, and fine motor control. Standing 1.8 metres tall, it can perform ballet spins, martial arts moves, and lifelike gestures, marking a significant advancement from its previous H1 model. Discover H2

Noetix Robotics Launches Affordable Consumer-Grade Humanoid Robot Bumi
Chinese startup Noetix Robotics launched Bumi, a consumer-grade humanoid robot priced below $1,400, aimed at education and home use. Bumi features voice command response, drag-and-drop programming, and up to two hours of battery life. Learn about Bumi

NEO Launch Signals Humanoids Becoming Commercial Products
With the launch of NEO, humanoid robots are transitioning from research prototypes to commercial products, signalling a major milestone in robotics development and potential widespread deployment. View announcement

  • Anthropic Launches Claude for Excel with Finance-Specific Skills and Connectors
    Anthropic released Claude for Excel in beta, integrating its AI assistant directly into Excel to read, analyse, modify, and create spreadsheets. The launch includes seven new connectors to financial data platforms such as Aiera, LSEG, and Moody's, plus finance-specific agent skills like discounted cash flow modelling and company analysis. Explore Claude for Excel

  • Cursor Launches Composer Coding Model and Multi-Agent Platform 2.0 with Voice Control
    Cursor introduced Composer, its first proprietary coding model that is 4x faster than competitors whilst maintaining state-of-the-art intelligence. The new Cursor 2.0 platform allows running up to eight AI coding agents simultaneously in isolated environments without interference, with integrated web browsing, native browser testing tools, and voice command input. Try Cursor 2.0

  • OpenAI Partners with PayPal to Enable Shopping Payments via ChatGPT in 2026
    OpenAI partnered with PayPal to enable shopping payments directly within ChatGPT starting in 2026. The integration uses OpenAI's Agentic Commerce Protocol, allowing merchants to list products without building custom integrations, with PayPal's 400 million users gaining seamless access to AI-driven shopping. Read the announcement

  • Google AI Studio Launches Vibe Coding for No-Code AI App Development
    Google updated AI Studio with "vibe coding," enabling users to create AI-powered apps from plain language prompts without manual API or SDK integration. Powered by Gemini models, this lowers technical barriers for developers and non-developers alike, accelerating AI adoption and experimentation. Try vibe coding

  • Dropbox Launches Dash AI as Context-Aware Intelligent Workspace
    Dropbox released Dash, an AI teammate that connects apps, files, and conversations into one intelligent workspace. Dash indexes cloud content to provide context-aware search and insights, automates workflows, and integrates multimodal capabilities to handle text, images, audio, and video. Explore Dash

  • YouTube Will Use AI to Upscale Low-Res Videos
    YouTube announced AI-powered upscaling technology to enhance low-resolution videos, improving viewing quality for older content and videos uploaded at lower resolutions. Learn more

Other things we’re loving

Community

The Spotlight Podcast

Noel Hurley Explains the Next Big Shift in Artificial Intelligence

In this episode of the Project Flux podcast, James Garner and Yoshi Soornack speak with Noel Hurley, CEO of Literal Labs, about his journey from ARM to founding Literal Labs. They discuss the evolution of AI technology, the innovative approach of settling machines, and the importance of decision intelligence in various industries. Noel emphasises the need for explainability in AI, the human element in decision-making, and the future of work in the context of AI advancements. The conversation also touches on the UK's position in the global AI landscape and the potential of humanoid robots in construction. This is essential listening for anyone seeking to understand where AI is heading and how it will reshape project delivery. Listen now

OpenAI CEO Sam Altman and Chief Scientist Jakub Pachocki Host Public Q&A
OpenAI leaders Sam Altman and Jakub Pachocki engaged with the public in a livestream Q&A discussing the company's future, AI social impact, and research directions. The session demonstrated openness and transparency despite challenging questions from the audience. Following OpenAI's recapitalization and transition to a public benefit corporation, the conversation covered the company's roadmap towards AGI, the timeline for autonomous AI researchers, and the implications of the restructured Microsoft partnership. This candid discussion offers rare insight into the thinking of leaders at the forefront of AI development. Watch the Q&A

Event of the Week

Megaprojects Conference – P3GQA
The Megaprojects Conference brings together industry leaders, project delivery professionals, and government representatives to discuss the challenges and opportunities in delivering large-scale infrastructure projects. Topics include AI adoption, digital delivery, risk management, and stakeholder engagement. This is an essential event for anyone involved in major project delivery, offering networking opportunities and insights from the world's most complex programmes. Register now

One more thing

That’s it for today!

Before you go we’d love to know what you thought of today's newsletter to help us improve The Project Flux experience for you.

Login or Subscribe to participate in polls.

See you soon,

James, Yoshi and Aaron—Project Flux