|
|
|
|
|
|
Run AI workloads on any of those four for a regulated UK client and your vendor now answers to the Bank of England.
|
|
JULY 14TH, 2026 · BY JONATHAN HILDEBRANDT
|
|
HM Treasury designated Amazon Web Services, Google Cloud, Microsoft, and Oracle as Critical Third Parties effective July 13, the first activation of powers under the Financial Services and Markets Act 2023. The Bank of England, PRA, and FCA can now demand information, run resilience tests, and enforce cloud-specific rules on those four vendors. A 2024 BoE/FCA survey put Microsoft, Google, and Amazon at 73% of UK financial-sector cloud (full breakdown). Why it matters. Run AI workloads on AWS, Azure, or GCP for a regulated UK client and your vendor now answers to the Bank of England, not just its own SLA. The regime is rolling; more providers can land on the list.
|
|
Read the full regulatory breakdown →
|
|
|
Apple rebuilt Siri on Google Gemini
JULY 14TH · PONDERO NEWSDESK
iOS 27 public beta shipped July 13 with a Siri that holds conversations, reads on-screen context, and takes app actions, with Gemini handling world knowledge after the OpenAI deal collapsed. Apple claims apps launch 30% faster and Photos loads 70% faster; you need an iPhone 15 Pro or newer. Our take →
|
Mistral shipped an 8B robot navigator with no LiDAR
JULY 14TH · PONDERO NEWSDESK
Robostral Navigate turns an RGB camera plus plain-language instructions into navigation commands, hitting 76.6% success on the R2R-CE unseen-environment benchmark, 9.7 points over the best prior single-camera method, per Bloomberg. It runs on wheeled, legged, and flying robots. Our take →
|
Five enterprise giants line up against MCP
JULY 14TH · PONDERO NEWSDESK
Google, Microsoft, Salesforce, Snowflake, and ServiceNow agreed to back a shared agent-backend protocol, framed as a counter to Anthropic's MCP, per The Information. MCP has the head start; this coalition owns the data-layer distribution that decides default enterprise deployments. Our take →
|
Make
JULY 14TH · PONDERO TOOLS
Visual automation for wiring AI agents into the apps you already run, without writing glue code between them. Try Make →
|
Cloudways
JULY 14TH · PONDERO TOOLS
Managed cloud hosting to self-host agents like OpenClaw or n8n in a few minutes instead of paying per seat. Try Cloudways →
|
|
•
|
US floats capping open-source models at China's level. The Trump administration and industry are discussing a framework to cap US open-weight model capability at leading Chinese open-source levels, currently about 7 months behind the US frontier, per the Washington Post. No rule or order exists yet. Source →
|
|
•
|
Nvidia cut more than half its Asia distributors. Nvidia built a distributor whitelist amid probes into Southeast Asian resellers suspected of routing restricted chips into China, per the FT. Firms that were cut can reapply after compliance changes. Source →
|
|
•
|
Nous Research is raising $75M at a $1.5B valuation. The Hermes Agent maker, at 214K GitHub stars, is in talks for at least $75M led by Robot Ventures with USV, up from $70M raised to date, per TechCrunch. The money funds cloud-hosted Hermes services. Source →
|
|
•
|
Gemini 3.5 Pro is targeting July 17, still unconfirmed. Third-party reports peg July 17 as Google's internal GA target after a full architecture rebuild, but there is no model card, blog post, or pricing from Google, per TechTimes. Treat it as a leak, not a date. Source →
|
Claude Fable 5 vs GPT-5.6 Sol: decide before July 19
JULY 14TH · PONDERO COMPARISON
Fable 5 is free for Claude subscribers only through July 19, so we compared it against GPT-5.6 Sol on pricing and coding tasks before the window shuts. For a solo dev already on a Claude plan, ride Fable 5 now; a team standardized on GPT tooling should read the task breakdown before switching. Read it →
|
Grok 4.5 at $6/M vs Opus 4.8 at $25/M in Cursor
JULY 12TH · PONDERO COMPARISON
A task-by-task Cursor model pick with the six conditions that flip each call. Default to Grok 4.5 for high-volume refactors on output cost; reach for Opus 4.8 on the hard reasoning passes where it earns the premium. Read it →
|
OpenClaw vs Lindy: self-host or pay per seat
JULY 11TH · PONDERO COMPARISON
OpenClaw's foundation is backed by OpenAI, NVIDIA, and Microsoft, and self-hosting runs roughly $15-40 a month versus Lindy at $49.99 and up. Builders who can run their own infra should self-host OpenClaw; teams that want managed setup and support should take Lindy. Read it → or try Lindy →
|
Lumo 2.0 vs ChatGPT for privacy-bound teams
JULY 13TH · PONDERO COMPARISON
Lumo 2.0 posts an Intelligence Index score for the first time, giving compliance teams a number to compare against ChatGPT and Claude alongside its privacy architecture. If your stack cannot send data to US servers, this is the pick to evaluate. Read it →
|
|
|
|
|
Jonathan Hildebrandt
Co-founder and primary operator of Pondero. Writes the Pondero Brief.
|
|
|
Affiliate disclosure
·
Pondero earns commissions on some links. This does not affect our editorial picks.
|
|
