Jester is Becoming [REDACTED]
Jester is becoming [REDACTED] Systems. This has been coming for a while, and to be honest, it is less of a rebrand than it is finally admitting that what we have built is no longer the thing we originally named. Jester started as one TradingView strategy and a bot. That was basically it. At that scale, the name was perfect. But over the last three years the strategy became a platform, the platform needed execution, execution needed risk, risk needed portfolio context, then we added backtesting, optimization, market data, automation, research tooling, machine learning hooks, multiple venues, APIs, MCP, autonomous agents, agent coordination, multiple front ends, and all the infrastructure underneath those things. Eventually Jester stopped describing the whole system and started describing one part of it. [REDACTED] is the name for what the system has actually become.
The biggest change is not even the technology, it is how we think people are going to use it. Originally, we assumed people would be talking to the bots, opening Crucible, inspecting strategies, running backtests, tweaking parameters and generally controlling everything themselves. That made sense when agents were still mostly a novelty and the human was obviously going to remain at the centre of the loop. Over the last year that assumption has changed very quickly. Agents are becoming capable enough that the human should increasingly be defining the objective, the capital, the permissions and the risk constraints, then letting the system do the work. Research the market, form a hypothesis, build a strategy, test it, compare it against alternatives, reject the bad ideas, understand the portfolio it is entering, apply risk, deploy it, monitor it and adapt. Other agents can challenge the thesis, specialize in execution, watch portfolio exposure, or handle completely different parts of the process. The human does not need to micromanage every step. The system should be able to orchestrate it.
That is the real shift behind [REDACTED]. We are no longer trying to build a smarter trading bot. We are building an environment where financial agents can actually operate. The simplest way to describe the system is research, coordination, risk and execution, all connected through the same infrastructure. Research is where agents investigate markets, generate and test strategies, compare ideas, optimize parameters and determine whether something is actually worth deploying. Coordination is how those agents share context, delegate work and operate as part of the same market loop instead of existing as isolated chat windows. Risk is the layer that prevents autonomy from becoming stupidity. Sizing, leverage, exposure, stops, drawdown, portfolio state, account policy and venue constraints all need to be enforced before an agent is allowed to touch capital. Then execution is the final step, where a strategy actually leaves the research environment and becomes a live action in a market. The goal is to carry context all the way from "I wonder if this works" to "this is now deployed with real capital inside defined risk."
This also changes how we think about our products. Oroboros, Saylis, Crucible, Tesseract and JESTER are not really separate systems. They are different surfaces into the same infrastructure. Oroboros is the perpetual and prop trading interface, and right now it is probably the clearest expression of the full [REDACTED] loop, from manual trading through assisted operation and eventually fully autonomous execution. Saylis is the onchain side, where the same underlying system can be used to research and operate across decentralized markets. Crucible is the research environment, where strategies are backtested, compared, optimized and hopefully broken before real money ever gets near them. Tesseract handles prediction markets and the broader world model layer, because not every market question begins with a candlestick. JESTER is another interface into the system, but JESTER itself is not [REDACTED]. Neither is Telegram, neither is Oroboros, neither is the web terminal. Those are clients. [REDACTED] is underneath all of them.
That distinction matters because it means our own front ends do not have to be the only way this technology wins. We would obviously love Oroboros and Saylis to become large products, but they do not have to replace every broker or trading platform in the world. Questrade does not necessarily have to be our competitor. Wealthsimple does not necessarily have to be our competitor. Interactive Brokers does not necessarily have to be our competitor. They already have users, accounts, interfaces and distribution. What they may not have is the autonomous research, risk, strategy and agent infrastructure underneath all of it. So keep the frontend. Use [REDACTED] underneath it. Use our backtesting, our risk engine, our execution layer, our agent coordination, or just one small piece of the system. The goal is not to force the entire financial world into one [REDACTED] application. The goal is to build infrastructure useful enough that a lot of different applications, brokers, agents and financial systems can use it. That is why the new positioning is the infrastructure behind autonomous finance.
It is also important to say that this is not a whitepaper for something we intend to start building after the relaunch. The system already exists. During Jester's alpha we have routed more than $9 million in real trading volume through the infrastructure. We have had real operators running strategies, automation and real capital through the system. We have had real execution problems, real risk problems, real bugs and real failures, and we have had to actually solve them. Over 2 years we have rebuilt pieces of the architecture again and again because every time the scope expanded we discovered another layer that had to exist underneath everything else. There is a very large difference between building an agent that can say "I think BTC is bullish" and building the infrastructure that lets that agent safely do something about it. Most of what we have spent the last few years building is the second part.
The timing of this is also kind of hilarious. The broader AI industry has been moving through almost the exact same progression. The first generation of agents was basically giving an LLM some tools and letting it loop until something happened. Then everyone rediscovered workflows because deterministic systems were more reliable. Then the harnesses got much better, agents became capable of longer-running work, and now the industry is moving toward systems where agents have defined tools, permissions, workflows, memory, specialized environments and infrastructure around them. That is basically where we have been heading with finance. We did not begin three years ago with some grand thesis about agent harnesses. We were trying to solve the next problem in front of us, and those problems kept forcing us toward this architecture. Now the rest of the industry is arriving at many of the same conclusions, which makes the direction we have been moving in feel much less strange than it did even a year ago.
