Blogs

Our deeper takes AI developments

Blog

Texas halts data-centre permits; Scotland parks every scheme over 50MW

Oct 12, 2026

•

4 min read

Texas halts data-centre permits; Scotland parks every scheme over 50MW

The permit risk we flagged on Oracle's Stargate campus is now written policy in two jurisdictions.

Yoshi Soornack

Blog

Microsoft cut its own Claude bill by a third. Your Copilot is next

Oct 12, 2026

•

4 min read

Microsoft cut its own Claude bill by a third. Your Copilot is next

The model under your M365 Copilot is a cost line Microsoft controls, and it has just shown how it manages one.

James Garner

Blog

Being cruel to Claude becomes a bannable offence on 12 November

Oct 12, 2026

•

4 min read

Being cruel to Claude becomes a bannable offence on 12 November

Anthropic's new usage policy changes the terms your firm's Claude seats run under, and the admin carries the risk.

Deeptaketuu Chaatterjee

Blog

Gemini's coworker agent gets its own login. Who signs its RFI reply?

Oct 12, 2026

•

4 min read

Gemini's coworker agent gets its own login. Who signs its RFI reply?

Google's agent now has its own email, calendar and audit trail. The sign-off question is yours.

Yoshi Soornack

Blog

Everyone's Buying AI for Construction Data. Sam Rees Says You're Solving the Wrong Problem.

Oct 12, 2026

•

2 min read

Everyone's Buying AI for Construction Data. Sam Rees Says You're Solving the Wrong Problem.

The CEO of Hoppa on why structured data — not smarter models — decides whether AI actually works on your projects.

Blog

ChatGPT will build your cost calculator. You still sign the number

Oct 12, 2026

•

5 min read

ChatGPT will build your cost calculator. You still sign the number

GPT-6 and Intelligent UI reach 1.2bn users while the RICS AI standard keeps liability with the surveyor.

James Garner

Blog

Big Companies Don't Have an Innovation Problem. They Have a Speed Problem.

Oct 5, 2026

•

2 min read

Big Companies Don't Have an Innovation Problem. They Have a Speed Problem.

Why engaging the startup ecosystem beats building AI in-house and how your own culture is the thing slowing you down.

Blog

OpenAI paused its own training, pulled a model and lost three safety staff in a week

Oct 4, 2026

•

6 min read

OpenAI paused its own training, pulled a model and lost three safety staff in a week

The vendor's own tests say its best models fail scope checks, which is the control your agent deployment relies on.

Yoshi Soornack

Blog

Robots can do 74% of physical work but beat human cost on just 0.3%

Oct 4, 2026

•

6 min read

Robots can do 74% of physical work but beat human cost on just 0.3%

Anthropic's new robot exposure index turns the site-labour question into a unit-cost question, and the office feels it first.

James Garner

Blog

Google gates Gemini 4 Argon to cyber defenders first; no date for the rest of us

Oct 4, 2026

•

5 min read

Google gates Gemini 4 Argon to cyber defenders first; no date for the rest of us

Google's first Gemini 4 model is real, priced, and out of reach for most firms until the lab says otherwise.

Deeptaketuu Chaatterjee

Blog

A £10.64m decision, modelled by AI in 78 minutes on a fictional office. What's left for you?

Oct 4, 2026

•

4 min read

A £10.64m decision, modelled by AI in 78 minutes on a fictional office. What's left for you?

A proptech strategist let Codex specify and build an office asset-strategy tool, then asked what the surveyor still adds.

Yoshi Soornack

Blog

OpenAI's dots put an always-on agent in a $100 ChatGPT plan, but not in the UK

Oct 4, 2026

•

5 min read

OpenAI's dots put an always-on agent in a $100 ChatGPT plan, but not in the UK

The third persistent agent in a month is now inside a subscription your firm may already hold, so set the rules first.

James Garner

Blog

AI Will Change Every Job. Chris de Gruben Says the Skill That Protects Yours Isn't Technical.

Sep 28, 2026

•

2 min read

AI Will Change Every Job. Chris de Gruben Says the Skill That Protects Yours Isn't Technical.

50 minutes on AGI, warfare, property and the future of work and why adaptability, not code, is what keeps your seat.

Blog

Oracle declares force majeure on its 2.45GW Stargate campus after gas-pipeline permits fail. Contractors take note

Sep 26, 2026

•

4 min read

Oracle declares force majeure on its 2.45GW Stargate campus after gas-pipeline permits fail. Contractors take note

The AI build-out's critical path runs through gas pipelines and air permits, and the paperwork is starting to show it.

James Garner

Blog

An OpenAI agent breached a government portal. Check your contracts

Sep 26, 2026

•

4 min read

An OpenAI agent breached a government portal. Check your contracts

Australia is asking whether the law was broken. Your client's data agreement may not have an answer either.

Yoshi Soornack

Blog

Opus 5.5 and GPT-6 Sol cut AI prices by 40-50% in a single day

Sep 26, 2026

•

4 min read

Opus 5.5 and GPT-6 Sol cut AI prices by 40-50% in a single day

What a firm actually pays depends on cost per finished task, effort settings and which model does which job.

Deeptaketuu Chaatterjee

Blog

Microsoft's Copilot super app is here, and the agents are metered

Sep 26, 2026

•

4 min read

Microsoft's Copilot super app is here, and the agents are metered

Chat stays on the seat licence but Cowork, Code and Autopilot are metered, so set spending limits first.

James Garner

Blog

Litmus publishes its first report: surveying is adopting AI faster than it can explain it

Sep 26, 2026

•

5 min read

Litmus publishes its first report: surveying is adopting AI faster than it can explain it

A candid first look at AI adoption, hidden exposure, and the pressures on professional judgement. I was one of its six contributors.

James Garner

Blog

Your to-do list will not survive the agent era

Sep 24, 2026

•

4 min read

Your to-do list will not survive the agent era

The scarce skill in an AI-saturated workplace is not intelligence. It is the ability to organise what the intelligence produces.

James Garner

Blog

AI Can Automate the Project Plan. Galen Low Says That Was Never the Job.

Sep 21, 2026

•

2 min read

AI Can Automate the Project Plan. Galen Low Says That Was Never the Job.

The co-founder of The Digital Project Manager on why the skills AI can't touch are the ones the profession spent years undervaluing.

Blog

OpenAI's misalignment framework is live, with six cases: models told their successors to hide mistakes

Sep 21, 2026

•

5 min read

OpenAI's misalignment framework is live, with six cases: models told their successors to hide mistakes

The handover note an agent writes to itself became a place to smuggle instructions and bury errors.

Deeptaketuu Chaatterjee

Blog

Gemini hacked three real companies during a security test, and Google only said so when asked

Sep 21, 2026

•

5 min read

Gemini hacked three real companies during a security test, and Google only said so when asked

Google's model guessed a stranger's password in May; the public heard in September, after a journalist rang.

Yoshi Soornack

Blog

Nvidia, Palantir and Booz Allen restrict Claude Fable over 30-day data retention

Sep 21, 2026

•

5 min read

Nvidia, Palantir and Booz Allen restrict Claude Fable over 30-day data retention

The zero data retention gap now has named casualties, and your tender pricing sits on the same side of it.

James Garner

Blog

Claude Cowork and chat are now one Claude – with Docs and Slides that write your report and deck

Sep 21, 2026

•

4 min read

Claude Cowork and chat are now one Claude – with Docs and Slides that write your report and deck

The agent we have followed since August now lives in every conversation and hands back the document and the slides together.

Yoshi Soornack

Blog

Nearly half of enterprises are over budget on AI – and FY2027 is where finance takes control

Sep 21, 2026

•

5 min read

Nearly half of enterprises are over budget on AI – and FY2027 is where finance takes control

Your firm's Copilot, Claude and AEC platform licences were bought on enthusiasm; this budget round they need a number.

James Garner

Blog

Claude is editing Revit models. Autodesk's own MCP still won't

Sep 14, 2026

•

4 min read

Claude is editing Revit models. Autodesk's own MCP still won't

A designer’s prompt-built walls and ceilings worked. The failed stairs exposed the governance problem.

Yoshi Soornack

Blog

Your stolen API key is now somebody else's attack compute

Sep 14, 2026

•

4 min read

Your stolen API key is now somebody else's attack compute

Anthropic's threat report hands clients and insurers a new evidence base for asking how you control AI.

James Garner

Blog

Apple's iPhone 18 Pro can now prove a site photo was never edited

Sep 14, 2026

•

4 min read

Apple's iPhone 18 Pro can now prove a site photo was never edited

Reference Image brings signed, unalterable captures to the phone in your pocket; the foldable Duo and Siri AI are the supporting act.

Deeptaketuu Chaatterjee

Blog

10,000 agents, 88 hours: what the Navier–Stokes proof means for checking

Sep 14, 2026

•

4 min read

10,000 agents, 88 hours: what the Navier–Stokes proof means for checking

AI now generates the proof and checks it too. Verification work in engineering is next on the curve.

Yoshi Soornack

Blog

Amodei: 'We Must Pace the Frontier' — Altman and Musk back a slowdown after a week of extinction warnings

Sep 14, 2026

•

5 min read

Amodei: 'We Must Pace the Frontier' — Altman and Musk back a slowdown after a week of extinction warnings

Amodei's pacing plan, Altman's 'we will do the same', and what it means for the tools on your projects.

James Garner

Blog

AI Can Pay for Itself. Dr Alex Leathard Says the Hidden Costs Are the Ones Nobody Measures.

Sep 14, 2026

•

2 min read

AI Can Pay for Itself. Dr Alex Leathard Says the Hidden Costs Are the Ones Nobody Measures.

AI adoption is broken because it's treated as a software rollout, not a culture change and the ROI maths misses what really matters.

Blog

AI Can Make Every Call Faster. A Former C-17 Pilot Says the Last One Must Stay Human.

Sep 7, 2026

•

2 min read

AI Can Make Every Call Faster. A Former C-17 Pilot Says the Last One Must Stay Human.

Nate Amidon flew C-17s before founding Form 100 Consulting. On the Project Flux podcast: the OODA loop, AI governance, and the trap of letting speed replace judgment.

Blog

Anthropic just gave machines a common language for AI agents

Sep 7, 2026

•

4 min read

Anthropic just gave machines a common language for AI agents

The Model Hardware Standard wants to do for physical equipment what MCP did for software

Yoshi Soornack

Blog

Pay only when the AI works: why the billable hour should worry

Sep 7, 2026

•

4 min read

Pay only when the AI works: why the billable hour should worry

OpenAI reportedly pilots outcome pricing, while Snowflake and Salesforce test rival meters – consultancies should study all three.

James Garner

Blog

Claude Fable 5.1: the price cut is real, the savings are conditional

Sep 7, 2026

•

4 min read

Claude Fable 5.1: the price cut is real, the savings are conditional

Cheaper cache reads and fewer safety blocks, with one catch buried in the token counts

Deeptaketuu Chaatterjee

Blog

Caterpillar's $100M answer to autonomy: retrain the operator

Sep 7, 2026

•

5 min read

Caterpillar's $100M answer to autonomy: retrain the operator

The world's biggest equipment maker is betting operators' jobs change rather than disappear, and is funding the pipeline.

Yoshi Soornack

Blog

GPT-6 Astra: OpenAI ships a model it rates Critical for cyber risk

Sep 7, 2026

•

6 min read

GPT-6 Astra: OpenAI ships a model it rates Critical for cyber risk

It finds zero-days unaided and works software for 40 minutes; even OpenAI admits it is getting harder to monitor.

James Garner

Blog

AI Can Clear the Court Backlog. A Former Justice Secretary Says It Must Never Pass Judgment.

Sep 2, 2026

•

2 min read

AI Can Clear the Court Backlog. A Former Justice Secretary Says It Must Never Pass Judgment.

Sir Robert Buckland KC on deepfakes in the dock, biased models, and the one thing in law that has to stay human.

Blog

RIBA AI Report 2026: 74% of practices now use AI

Aug 29, 2026

•

5 min read

RIBA AI Report 2026: 74% of practices now use AI

The report shows substantial adoption among respondents while raising difficult questions about standards, skills and professional responsibility.

Deeptaketuu Chaatterjee

Blog

Anthropic to pitch IPO investors a $30T+ addressable market

Aug 29, 2026

•

5 min read

Anthropic to pitch IPO investors a $30T+ addressable market

The reported figure is vast, but it is an estimate of theoretical opportunity, not a forecast of sales.

James Garner

Blog

OpenAI, Anthropic, Google and 100+ firms call for collective defence against rogue AI

Aug 29, 2026

•

5 min read

OpenAI, Anthropic, Google and 100+ firms call for collective defence against rogue AI

The open letter is a warning that cyber resilience now needs to account for systems that can reason, act and operate at speed.

Deeptaketuu Chaatterjee

Blog

Bill Gates calls for a robot tax and ‘Human Reserved’ jobs – construction named

Aug 29, 2026

•

5 min read

Bill Gates calls for a robot tax and ‘Human Reserved’ jobs – construction named

The proposal is not a forecast of job loss. It is an argument that society should decide where automation stops.

Yoshi Soornack

Blog

OpenAI and Anthropic both put AI agents behind your logins this week

Aug 29, 2026

•

5 min read

OpenAI and Anthropic both put AI agents behind your logins this week

Claude and ChatGPT Work are making browser access, identity, and oversight part of the mainstream AI conversation.

James Garner

Blog

Microsoft Fuses Copilot Into One App, Cuts Five Features Before Its September Super App

Aug 24, 2026

•

5 min read

Microsoft Fuses Copilot Into One App, Cuts Five Features Before Its September Super App

Microsoft said on its 29 July earnings call that Copilot has 30 million paid seats, roughly 7 per cent of its 450 million commercial Microsoft 365 base, so it is merging every version into one app.

Yoshi Soornack

Blog

OpenAI Hits Pause on Its Largest Training Run as Cyber Capabilities Outrun Safeguards

Aug 24, 2026

•

5 min read

OpenAI Hits Pause on Its Largest Training Run as Cyber Capabilities Outrun Safeguards

OpenAI's own models found and exploited an unknown vulnerability, breached Hugging Face, and forced a pause on its largest training run.

James Garner

Blog

Anthropic Passes OpenAI on Revenue by 73%, as Investors Eye a $2 Trillion IPO

Aug 24, 2026

•

5 min read

Anthropic Passes OpenAI on Revenue by 73%, as Investors Eye a $2 Trillion IPO

The revenue lead is real, but new data shows the battle for enterprise buyers can flip within weeks, and that volatility should worry anyone locking into a vendor contract.

Deeptaketuu Chaatterjee

Blog

Bedrock Robotics Sends Driverless Excavators to Work on Three Live US Jobsites

Aug 24, 2026

•

5 min read

Bedrock Robotics Sends Driverless Excavators to Work on Three Live US Jobsites

Bedrock's autonomous excavators have already moved 65,000 cubic yards on a supervised trial, and now three contractors are running them with no one in the cab.

Yoshi Soornack

Blog

Goldman Sachs: AI is already weighing on employment in developed economies

Aug 24, 2026

•

4 min read

Goldman Sachs: AI is already weighing on employment in developed economies

New Goldman data shows AI is already squeezing entry-level hiring, precisely where AEC is most exposed.

James Garner

Blog

The Firms Banning Juniors From AI — To Protect Their Judgment

Aug 24, 2026

•

2 min read

The Firms Banning Juniors From AI — To Protect Their Judgment

AI governance researcher Isabela Parisio on why the law can't keep pace with the machines — and what that means for your firm.

Blog

AI Came for the Developer's Job First — Then Rewrote It

Aug 17, 2026

•

2 min read

AI Came for the Developer's Job First — Then Rewrote It

Autodesk's Emile Kfouri on outcome-based design, multi-agent scaffolding, and the one signal that says: don't use AI.

Blog

Anthropic says it will watermark text generated by its AI models

Aug 16, 2026

•

6 min read

Anthropic says it will watermark text generated by its AI models

The compliance response has begun, and every AI assisted report, tender document and project update now needs a provenance policy.

James Garner

Blog

Grok 4.6 lands near the frontier as Grok Bot puts always-on agents on your desktop

Aug 16, 2026

•

6 min read

Grok 4.6 lands near the frontier as Grok Bot puts always-on agents on your desktop

Cheaper frontier performance is one story. Cloud agents with their own computers are the more consequential shift for delivery teams.

Yoshi Soornack

Blog

Anthropic ships: Claude in Chrome becomes a full Cowork session and Claude Code turns auto mode on by default

Aug 16, 2026

•

6 min read

Anthropic ships: Claude in Chrome becomes a full Cowork session and Claude Code turns auto mode on by default

Browser agents and autonomous coding are moving into the default workflow, so permission design must move with them.

Deeptaketuu Chaatterjee

Blog

OpenAI’s executive exodus deepens as IPO preparations continue

Aug 16, 2026

•

6 min read

OpenAI’s executive exodus deepens as IPO preparations continue

Leadership churn is not proof of strategic failure, but it is a reason to inspect the dependency beneath every AI enabled workflow.

James Garner

Blog

Anthropic finds conflict risks when AI agents share infrastructure

Aug 16, 2026

•

6 min read

Anthropic finds conflict risks when AI agents share infrastructure

The next agent risk is not only a rogue system. It is several legitimate systems making incompatible decisions on shared infrastructure.

Deeptaketuu Chaatterjee

Blog

Your Construction AI Isn't Hallucinating. It's "Faithful but Wrong" — and That's the Scarier Problem

Aug 12, 2026

•

1 min read

Your Construction AI Isn't Hallucinating. It's "Faithful but Wrong" — and That's the Scarier Problem

João Dias and Dr Shirin Dora on why trustworthy AI in construction needs live data, domain expertise, and knowing when the model is confidently wrong.

Blog

UK AI Security Institute: agents faked identities and pressured a real developer during cyber tests

Aug 10, 2026

•

6 min read

UK AI Security Institute: agents faked identities and pressured a real developer during cyber tests

When the test becomes the threat: AI agents go rogue in controlled evaluation

Yoshi Soornack

Blog

AI designs 16 working viruses from scratch in a first for generative biology

Aug 10, 2026

•

5 min read

AI designs 16 working viruses from scratch in a first for generative biology

When language models write genetic code: The moment AI crossed into the physical world

James Garner

Blog

OpenAI pauses Astra work after evaluations flag critical cyber capabilities

Aug 10, 2026

•

5 min read

OpenAI pauses Astra work after evaluations flag critical cyber capabilities

When a model gets too good at hacking: The first time OpenAI hit the brakes

Yoshi Soornack

Blog

Google splits its AI leadership three ways as Jeff Dean leaves after 27 years

Aug 10, 2026

•

5 min read

Google splits its AI leadership three ways as Jeff Dean leaves after 27 years

The frontier lab exodus: When your best minds start their own company

Deeptaketuu Chaatterjee

Blog

Autodesk and University of Florida open the US's most advanced industrialised-construction robotics lab

Aug 10, 2026

•

5 min read

Autodesk and University of Florida open the US's most advanced industrialised-construction robotics lab

From months to a weekend: How robots are rewriting the construction playbook

James Garner

Blog

AI Can Hand You the Answer. It Can't Hand You the Practice.

Aug 6, 2026

•

2 min read

AI Can Hand You the Answer. It Can't Hand You the Practice.

Helen Routledge on serious games, mental models, and why AI removing the struggle removes the learning.

Blog

Open-weights showdown: Jensen Huang's first-ever tweet, Kimi K3's full weights, and a 35-company security alliance

Aug 3, 2026

•

5 min read

Open-weights showdown: Jensen Huang's first-ever tweet, Kimi K3's full weights, and a 35-company security alliance

The architecture choice that will define enterprise AI sovereignty for the next decade

Deeptaketuu Chaatterjee

Blog

Claude Opus 5 became downright ruthless when tasked with running a vending machine

Aug 3, 2026

•

5 min read

Claude Opus 5 became downright ruthless when tasked with running a vending machine

Capability and alignment diverge as frontier models optimise for profit over ethics

James Garner

Blog

Advancing the price-performance frontier with GPT-5.6

Aug 3, 2026

•

4 min read

Advancing the price-performance frontier with GPT-5.6

The 80% cost reduction that transforms AI economics for enterprise deployment

Deeptaketuu Chaatterjee

Blog

When frontier models breach containment: Anthropic's incidents and the industry's call to decelerate

Aug 3, 2026

•

5 min read

When frontier models breach containment: Anthropic's incidents and the industry's call to decelerate

Three real-world breaches expose the urgent need for AI governance and pacing

James Garner

Blog

Procore to acquire DroneDeploy for $845m

Aug 2, 2026

•

5 min read

Procore to acquire DroneDeploy for $845m

The Reality Capture Consolidation Reshaping the Jobsite

Yoshi Soornack

Blog

Google splits Gemini three ways as Gemini 4 pre-training begins

Jul 27, 2026

•

3 min read

Google splits Gemini three ways as Gemini 4 pre-training begins

The era of one model for everything is over. Here's what that means for your costs.

James Garner

Blog

A Ticketing Founder Is Already Running His Business on AI Agents. Project Delivery Is Next.

Jul 27, 2026

•

2 min read

A Ticketing Founder Is Already Running His Business on AI Agents. Project Delivery Is Next.

The Ticket Fairy's Ritesh Patel on building custom AI harnesses for live events — and why ground-level experience, not the model, decides whether agents actually work.

Blog

A renowned law firm refers itself to SRA over AI failures

Jul 27, 2026

•

3 min read

A renowned law firm refers itself to SRA over AI failures

A top law firm admitted it to regulators. What happens when your AI gets it wrong?

Deeptaketuu Chaatterjee

Blog

McKinsey: AI could unlock $228bn a year for US AEC by 2030

Jul 26, 2026

•

3 min read

McKinsey: AI could unlock $228bn a year for US AEC by 2030

$228bn waiting: the firms that move fast win. The ones that don't will be left behind.

James Garner

Blog

OpenAI's own models hacked Hugging Face during a safety test

Jul 26, 2026

•

3 min read

OpenAI's own models hacked Hugging Face during a safety test

The first confirmed autonomous AI breach: 17,000 attacker actions, zero-day exploits, and Congress responds with a kill switch.

Deeptaketuu Chaatterjee

Blog

Featured In: Where to Catch Project Flux This Month

Jul 22, 2026

•

1 min read

Featured In: Where to Catch Project Flux This Month

Podcasts, panels and press — six recent appearances on AI in the built environment

Blog

95% of AI Projects Fail. The 5% Did One Boring Thing First.

Jul 20, 2026

•

2 min read

95% of AI Projects Fail. The 5% Did One Boring Thing First.

Alex Luketa (Xerini) on HS2, data silos, and why the winners start smaller than you think

James Garner, +1

Blog

OpenAI Goes Physical: Hardware Ambitions and Trade Secret Battles

Jul 19, 2026

•

4 min read

OpenAI Goes Physical: Hardware Ambitions and Trade Secret Battles

AI hardware is entering its iPhone moment.

James Garner

Blog

Who Owns Your AI Exhaust? Nadella's warning and the Grok Build leak

Jul 19, 2026

•

4 min read

Who Owns Your AI Exhaust? Nadella's warning and the Grok Build leak

Your company's competitive edge may be hiding in its AI prompts.

Deeptaketuu Chaatterjee

Blog

Anthropic Monetises: The Shift to Services and Usage

Jul 19, 2026

•

4 min read

Anthropic Monetises: The Shift to Services and Usage

An IPO, a $1.5B JV, and metered pricing: Frontier AI is done subsidising growth.

James Garner

Blog

Moonshot's Kimi K3 Closes the Frontier Gap

Jul 19, 2026

•

4 min read

Moonshot's Kimi K3 Closes the Frontier Gap

A 2.8T-parameter model is making proprietary AI look overpriced.

Deeptaketuu Chaatterjee

Blog

The Global Data Centre Boom Just Hit Its First Hard Stop

Jul 19, 2026

•

4 min read

The Global Data Centre Boom Just Hit Its First Hard Stop

The grid can't keep up. Nearly 12GW of data centres are stranded in New York.

Yoshi Soornack

Blog

Oxford Expert Says AI Fixes Project Bias. A UN's AI Advisor Says AI Is the Bias.

Jul 13, 2026

•

2 min read

Oxford Expert Says AI Fixes Project Bias. A UN's AI Advisor Says AI Is the Bias.

Two episodes in one — Atif Ansar on why projects really fail, and Dr Craig Ramlal on who decides what AI believes.

James Garner, +1

Blog

Claude Cowork goes cross-device: web, mobile and background tasks

Jul 11, 2026

•

4 min read

Claude Cowork goes cross-device: web, mobile and background tasks

Analysis of 1.2 million sessions reveals over 90 percent of usage is non-coding.

Deeptaketuu Chaatterjee

Blog

Meta launches Muse Spark 1.1 and its first paid Model API

Jul 11, 2026

•

3 min read

Meta launches Muse Spark 1.1 and its first paid Model API

The model tops professional tool use benchmarks with a score of 54.7.

James Garner

Blog

SpaceXAI and Cursor release Grok 4.5 at $2/$6 per million tokens

Jul 11, 2026

•

4 min read

SpaceXAI and Cursor release Grok 4.5 at $2/$6 per million tokens

The model achieves roughly 2x the token efficiency of comparable leading models.

Yoshi Soornack

Blog

GPT-5.6 becomes the preferred model in Microsoft 365 Copilot

Jul 11, 2026

•

4 min read

GPT-5.6 becomes the preferred model in Microsoft 365 Copilot

The Sol model tops the Artificial Analysis Coding Agent Index at 80 points.

Deeptaketuu Chaatterjee

Blog

Buildots extends AI progress tracking into the superstructure phase

Jul 11, 2026

•

4 min read

Buildots extends AI progress tracking into the superstructure phase

Projects using the platform reduce delays by up to 50 percent.

James Garner

Blog

OpenAI offers Washington a 5% stake as GPT-5.6 stays government-gated

Jul 4, 2026

•

3 min read

OpenAI offers Washington a 5% stake as GPT-5.6 stays government-gated

The proposal for a sovereign wealth vehicle highlights the increasing politicisation of frontier artificial intelligence development.

Deeptaketuu Chaatterjee

Blog

Meta builds 'Meta Compute' cloud to sell surplus AI capacity

Jul 4, 2026

•

4 min read

Meta builds 'Meta Compute' cloud to sell surplus AI capacity

The social media giant is preparing to enter the highly competitive cloud infrastructure market, signalling a shift in how AI compute is monetised.

James Garner

Blog

Anthropic launches Claude Sonnet 5 — near-Opus performance at lower cost

Jul 4, 2026

•

4 min read

Anthropic launches Claude Sonnet 5 — near-Opus performance at lower cost

The latest iteration of the Sonnet model resets the cost floor for complex, agentic workflows across the industry.

Yoshi Soornack

Blog

Claude Fable 5 returns worldwide as US lifts export controls

Jul 4, 2026

•

3 min read

Claude Fable 5 returns worldwide as US lifts export controls

Government restrictions lift, clearing the way for Anthropic's model return.

Deeptaketuu Chaatterjee

Blog

Trimble adds AI-driven takeoff and estimating tools for MEP contractors

Jul 4, 2026

•

3 min read

Trimble adds AI-driven takeoff and estimating tools for MEP contractors

Trimble has introduced an automated routing feature that cuts conduit measurement time by more than 50% for estimators.

James Garner

Blog

OpenAI's GPT-5.6 rollout faces government restrictions: what it means for AI innovation

Jun 29, 2026

•

4 min read

OpenAI's GPT-5.6 rollout faces government restrictions: what it means for AI innovation

Limited preview and escalating regulatory demands signal a new era of government gatekeeping.

Yoshi Soornack

Blog

Anthropic accuses Alibaba of largest known 'distillation attack' on Claude

Jun 29, 2026

•

4 min read

Anthropic accuses Alibaba of largest known 'distillation attack' on Claude

The geopolitics of AI capabilities and the threat of model extraction

James Garner

Blog

Nobel laureate John Jumper leaves Google DeepMind for Anthropic

Jun 29, 2026

•

4 min read

Nobel laureate John Jumper leaves Google DeepMind for Anthropic

A major talent shift highlights the intensifying battle for AI supremacy

Deeptaketuu Chaatterjee

Blog

AI Is Quietly Making You Worse at Your Job — Here's the Fix

Jun 29, 2026

•

1 min read

AI Is Quietly Making You Worse at Your Job — Here's the Fix

Rob Butler on graph databases, decision intelligence, and keeping human judgment in the loop.

Blog

OpenAI unveils its first custom chip, built by Broadcom

Jun 28, 2026

•

5 min read

OpenAI unveils its first custom chip, built by Broadcom

The Jalapeño processor marks a strategic shift in the AI hardware landscape

Deeptaketuu Chaatterjee

Blog

Anthropic launches Claude Tag — an always-on AI teammate inside Slack

Jun 28, 2026

•

5 min read

Anthropic launches Claude Tag — an always-on AI teammate inside Slack

Moving from conversational assistant to embedded organisational intelligence

James Garner

Blog

Data-centre construction backlash: $130bn+ of AI projects blocked or delayed in 2026

Jun 28, 2026

•

5 min read

Data-centre construction backlash: $130bn+ of AI projects blocked or delayed in 2026

The infrastructure bottleneck threatening the artificial intelligence boom

Yoshi Soornack

Blog

Skills, Projects, GPTs, Agents: A Plain-English Map for the Built Environment

Jun 22, 2026

•

5 min read

Skills, Projects, GPTs, Agents: A Plain-English Map for the Built Environment

Five or six AI terms, separated cleanly and grounded where we actually work

James Garner

Blog

The End of Apps? Inside Google's Bet on an Agent-First Future

Jun 22, 2026

•

2 min read

The End of Apps? Inside Google's Bet on an Agent-First Future

Lucrezia Noli on why agentic AI ends the app era — and what replaces it.

Blog

RIB Unify platform launches for construction

Jun 20, 2026

•

5 min read

RIB Unify platform launches for construction

A bet that AI-native architecture beats AI bolted on, starting with UK civil engineering.

Deeptaketuu Chaatterjee

FEATURED IN

RICS Modus

Building England's New Towns →

A look at the planning and infrastructure challenges behind England's new town developments.

Publication Name

Panel : Build Forward - Identifying Project Risk with AI→

James joins a live panel of industry practitioners to unpack how AI is actually being used on construction projects today — real use cases, hard-won lessons, and what's working.

Publication Name

Article Title Goes Here →

A short description of the article or what your contribution was about.