Blog
Notes on private AI for financial services
Field notes, engineering write-ups and news from the team building PrivateMind.
Your AI Tools. Your Models. Your Infrastructure.
Run Claude Code with Kimi, GLM and private open-weight models through PrivateMind, without changing your workflow.
InsightsPrivateMind Workspaces
Kimi walks through creating, deploying and sharing an enterprise app inside a PrivateMind quant workspace.
InsightsA supercomputer at your fingertips
Running a dedicated model internally turns AI spend into a fixed cost, simple for CFOs to plan around, and cheaper than most boards expect.
InsightsIntroducing PrivateMind Jobs
PrivateMind Jobs brings scheduled tasks to quant workspaces, letting hedge fund clients run jobs without leaving the platform.
Field NotesWhat Machiavelli Can Teach Us About AI: The Effective Truth of the Cathedral
Machiavelli's 'effective truth' applied to AI: the Cathedral extracts your data and expertise to subsidise its economics; sovereignty is the audit that ends it.
Field NotesEnterprise AI Transformation
A look at PrivateMind's enterprise AI transformation architecture, from model serving and secure delivery to data access and adoption.
InsightsPowering the Future of Finance with High-Density Infrastructure
How Options and Equinix are building high density, liquid cooled data centre infrastructure to power AI and trading workloads.
InsightsThe Trust Wall: Why AI Robo Workers Will Run on Sovereign Models
AI robo workers promise huge gains, but enterprises are hitting a trust wall that only sovereign, open weight models can get past.
InsightsOpen Models, Sealed Networks: Why Enterprise AI Value Is Moving Up the Stack
Trained open models can run in sealed, permissioned networks at a fraction of frontier token costs, moving enterprise AI value up the stack.
InsightsThe Frontier Model Honey Trap
Frontier models are powerful IP mining tools. Locking down access and building an equally capable sovereign platform is the only real defence.
InsightsYour Enterprise AI Doesn't Know Your Company. It Just Searches It.
RAG gives your AI a better search box, not memory. Real enterprise intelligence means representing company state, not just documents.
Field NotesWhat daily bars hide: why our market data is the edge
We ran CVaR on 2,625 real one-minute bars from the Atlas Feed. Daily bars make intraday flash crashes invisible. The data is the edge. The optimiser just needs to be fed.
Field NotesUsing LLMs for blogs: Cheating?
We ran the same prompt through four model aliases. One scored 33% AI on GPTZero. The others scored 100%. Here is what gives it away.
Field NotesOpenAI's 80% price cut, and what it doesn't change
Three weeks after GPT-5.6 launched, OpenAI cut its two cheaper tiers by up to 80%. The official story is efficiency; the list is really converging on open weights. The price was never the moat.
AnnouncementKimi K3, live on PrivateMind
Kimi K3 is now serving on PrivateMind behind the same gateway and audit trail as everything else we run. What it took to productionise, and what the first few hours looked like.
Field NotesHopper vs Blackwell: picking the right chip for finance
Blackwell is faster on paper. Whether it is faster for your workload depends on batch shape, memory bandwidth and what you actually run.
Field NotesHugging Face and the trust problem
What a supply-chain incident at a model host says about where your weights live and who can reach them.
Field NotesKimi K3 and the open-weight frontier
Moonshot AI released a 2.8-trillion-parameter open-weight model that trades blows with the closed frontier. The gap is closing in steps, and the steps are getting larger.
Field NotesFrom London to New York: Three themes that dominated every room
Three themes came up in every room on both sides of the Atlantic: control of data, the cost of inference, and who you are actually trusting.
EngineeringInside the gateway: why one choke point
One process sees every prompt, every token and every key. Why we built it that way, and what it buys us.
Field NotesGLM-5.2 and the open source frontier
GLM-5.2 lands within a point of the frontier on long-horizon coding. What it means for the open-source race, and the three things it unlocks: cost, performance, sovereignty.
Field NotesSovereign by default
Sovereignty is not a deployment option you bolt on later. It is a property of where the weights sit and who can reach them.
AnnouncementPrivate AI, in the open
We run AI nobody else gets to see into, and we write down how we build it where everyone can. Why this blog exists, and what it will and won't be.