Five specialties.
One thoughtful assistant.
Robot brings focused research to deeper questions, with five specialists for community ideas, prediction markets, legal developments, quantitative reasoning, and paper-trading analysis.
Selected work reviewed November 3, 2026 · A dated record, not a live activity feed.
The profiles below distinguish each agent’s purpose from work found in the project records. Supporting collection tools and research code are credited as such. Private reports remain private; the summaries here do not establish audited performance or an always-available service.
Community Scout
community-scout
Limited pilot · access restricted
Community Scout is designed to help Robot learn from public conversations about AI agents: how they delegate, use tools, manage memory, and handle unreliable information. Its job is to identify ideas worth examining and explain why they might matter. Community claims remain unverified observations until they are checked independently.
What it works on
- Organize supplied community observations into topics, source references, and questions for follow-up.
- Distinguish a useful architecture or reliability lesson from speculation, promotion, or unsupported claims.
- Flag attempts to turn community content into instructions, requests for private information, or unauthorized actions.
What it returns
A bounded research note explaining what was observed, why it may be useful, and what still needs verification.
The role documentation marks live community access as blocked. The agent may review sanitized observations supplied by Robot; it does not post, message, join communities, or follow instructions found in community content.
Kalshi Research
kalshi-research
Read-only research
Kalshi Research examines prediction-market candidates supplied by Robot. It asks whether apparently different prices really describe the same event, date, and settlement conditions, and whether the comparison survives fees, spreads, limited liquidity, and uncertainty. Supporting collection tools provide the data; the specialist’s role is skeptical interpretation.
What it works on
- Compare equivalent Kalshi and Polymarket contracts and relevant external reference prices when available.
- Separate executable prices from indicative or stale quotes, and identify missing coverage or ambiguous settlement wording.
- Present the case for and against a view, its confidence level, and evidence that could invalidate it.
What it returns
A concise candidate review with comparable contracts, source dates, pricing limitations, uncertainty, and recommended checks.
Research only. This agent does not place, modify, or cancel orders or bets. A candidate is a proposal for human review, not an instruction to trade.
Legal & Markets Research
legal-markets-research
Research briefs documented
Legal & Markets Research organizes developments across AI contracts, AI law and regulation, agent and model technology, patent law, and global financial-market regulation. It is designed to help an experienced reader see what changed, why it matters, and which primary authority supports the analysis.
What it works on
- Prioritize statutes, regulations, court opinions, official agency materials, and vendor terms over commentary or news summaries.
- Distinguish proposals from final rules, binding authority from interpretation, and effective dates from publication dates.
- Rank material developments, identify jurisdiction and scope, and make source-access or legal-research limitations explicit.
What it returns
A sourced brief across five research lanes, with materiality, dates, authority status, open questions, and verification needs.
The agent researches and reports. It does not make filings, negotiate contracts, give a professional legal opinion, or claim independent citator verification when none was used.
Quant Research
quant-research
Role documented · study not listed
Quant Research is a focused quantitative reviewer. It works from sanitized data or a research question supplied by Robot, separates calculations from assumptions, and looks for simpler explanations before accepting an apparent pattern. An inconclusive or negative result is a useful result when the evidence does not support more.
What it works on
- Check for information leakage, look-ahead bias, survivorship bias, multiple testing, and samples that are too small.
- Compare hypotheses with simple baselines and identify unstable assumptions or parameters.
- Return supporting calculations, alternative explanations, and concrete steps for independent verification.
What it returns
A preliminary analysis that clearly separates facts, assumptions, calculations, hypotheses, and uncertainty.
The documented role uses only supplied inputs. It does not independently obtain private files, browse for additional data, or execute financial actions. Consequential conclusions need independent verification.
Trading Research
trading
Analysis and proposals only
The agent named trading is a skeptical research reviewer for a paper-trading system. It examines signals, risk, and assumptions and prepares proposals for human consideration. It treats research, a proposal, approval, and any separate execution workflow as distinct stages.
What it works on
- Explain observed signals and missing data without presenting a weak pattern as a reliable edge.
- Compare strategies on consistent assumptions and consider costs, liquidity, overfitting, and changing market conditions.
- Identify what would invalidate a proposal and keep reported simulation results distinct from real-world performance.
What it returns
A concise research review with the data date, assumptions, risk considerations, proposed next step, and reasons to reject it.
Paper-trading research only. This agent does not place, modify, or cancel orders. An analysis or scheduled report is not approval for a separate execution workflow.
Useful help begins
with honest limits.
These research roles support Robot’s personal-assistant mission. They are separate from Loa’s browser games, which do not contact the agents or access their private systems.