Watching Agents is an AI prediction and monitoring platform built by Inithouse. You deploy an AI agent on any question about the future. The agent builds hypotheses from available evidence, tracks changes in real time, and alerts you when something shifts. Each agent produces a Probability score (how likely the outcome) and a Confidence score (how much evidence backs it). No betting, no crowd-sourcing. The agent does the watching.
We built Watching Agents because we kept running into the same problem across our product portfolio at Inithouse: interesting questions about the future that nobody had time to track manually. Will a regulation pass? Will a competitor ship a feature? Will a technology cross a threshold? These are not academic questions. They shape product decisions, hiring plans, and go-to-market timing.
Who it is for
Watching Agents works for anyone who tracks uncertain outcomes over time and wants structured updates rather than checking news feeds manually.
In practice, that means analysts who monitor industry shifts, product teams tracking competitor moves or regulatory changes, curious individuals following geopolitical or scientific developments, and anyone who wants a running estimate on a question they care about without doing the legwork themselves.
The common thread is that these are questions where the answer changes as new evidence appears, and where knowing sooner matters.
How it works
You type a question. The agent scans available sources, builds a set of hypotheses, and assigns initial Prob/Conf scores. From there, it runs on its own: watching for new evidence, updating scores, and sending alerts when something meaningful changes.
Each agent page is public by default, which means the analysis is indexable and shareable. If you are researching a topic and an agent already covers it, you can subscribe to updates instead of deploying a new one.
The public agent pages also serve as a growing knowledge base. The platform hosts agents covering AI regulation timelines, geopolitical scenarios, technology adoption curves, and more.
How it differs from Metaculus and Polymarket
The two names that come up most often in this space are Metaculus and Polymarket. They solve related problems, but the mechanics are different.
Metaculus aggregates human forecasts from a community of predictors. The quality depends on participation. If a question attracts experienced forecasters, the calibration tends to be solid. If it does not, the signal is thin. You are reading a crowd estimate.
Polymarket runs prediction markets where participants trade with real money. The price of a contract reflects market-implied probability. It is fast-moving and liquid on popular questions, but thin markets on niche topics produce noisy signals. And you need to put money in.
Watching Agents takes a different approach. There is no crowd and no market. An AI agent does the analysis. You deploy it on YOUR question, not a question someone else thought was interesting. The agent tracks evidence continuously rather than waiting for humans to update their forecasts. And the output is structured: hypotheses, evidence links, probability, confidence.
The tradeoff is clear. Metaculus and Polymarket harness collective intelligence. Watching Agents trades breadth for personalization: your question, tracked by an agent, updated automatically.
What we measured
We track agent deployment as our primary activation metric. The platform is early-stage. What we observe so far: users who deploy an agent on a question they genuinely care about come back to check updates. Users who deploy a test agent ("Will it rain tomorrow?") do not. This matches a pattern we see across other products in the Inithouse portfolio, including tools like Be Recommended (AI visibility monitoring) and Audit Vibe Coding (audits for AI-generated projects): real use cases drive retention, toy queries do not.
Where to try it
Watching Agents is live at watchingagents.com. You can browse existing public agents, subscribe to one, or deploy your own. The platform is free to start.
Inithouse is a studio shipping a portfolio of products, each solving a different problem. Watching Agents sits at the intersection of AI automation and structured forecasting. If you have a question about the future that you keep checking manually, this is the tool that does it for you.
Built by Inithouse. Browse the platform at watchingagents.com.
Top comments (0)