Arena Blog

TradingView's chart copilot can connect to your own MCP servers. Tested on the Backtesting Arena connector: it answered market questions correctly and invented instructions about its own settings four times. Why that happens, and how to tell which answers to check.

Agent payments are standardised: authorisation, identity, mandate, settlement. Whether the work was any good, nobody checks. Why that is a boundary and not a gap.

Your most thorough reader gets the least to read. Ten places where building for agents differs from building for people, with numbers from production.

An AI swarm turned $1,000 into nearly a million? The number falls apart at its own source. Why consensus from correlated simulations is not an edge.

Everyone talks about the agent economy. From the provider side — I run x402-paid financial endpoints — what agents actually buy today is sober: ingredients. And where the line runs between real and vision.

Asking which company is well-positioned in the AI drug-discovery race is really three questions at once. We sort the field soberly into shovels, bets, and giants — and show which layer earns regardless of any single trial outcome.

A thesis says trading platforms are going "agentic" — AI monitors, signals, executes. The structural part is right. The blind spot sits where money is lost.

"Invested in AI healthcare" can mean two almost opposite things. Two real companies show it: an annuity machine with ~86% recurring revenue — and an option with not a single approved drug. Same label, opposite risk DNA.

We wrote 500 lines of rules. Claude read them — and still wrote text-[13px] instead of text-xs. The mistake wasn't Claude. The mistake was confusing documentation with enforcement.

"AI-crypto" trades as one sector but is at least four different things on different layers of the stack: payment rails (Solana, Base, Sui, Tempo), agent frameworks (Virtuals, ElizaOS, Fetch/ASI), decentralized intelligence (Bittensor) and decentralized compute (Render, Akash, io.net). A map of the most relevant projects per category — and one test cutting across all of them to show where real substance sits and where a narrative was simply built around a token: is the token structurally necessary, or just present?