Skip to content

Deep Research Team

The native NataPulse Deep Research workflow uses eleven logical roles across six phases.

Defines the research question, scope, horizon, evidence needs, and specialist work plan. It identifies missing context and decides which analyst roles are relevant.

Examines company, project, filing, business, financial, and structural evidence applicable to the scope.

Examines market structure, price history, volatility, anomalies, and technical context. It does not convert a chart pattern into a guaranteed prediction.

Examines news catalysts, regulatory developments, macro conditions, and cross-asset context.

Examines source identity, social attention, sentiment change, rumor risk, and corroboration.

Examines wallet, protocol, transfer, exchange, bridge, and blockchain evidence when relevant.

Builds the strongest evidence-based positive interpretation. It must cite supporting evidence and identify assumptions.

Builds the strongest evidence-based negative or skeptical interpretation. It tests source weaknesses, counterevidence, and failure conditions.

Converts the competing analyses into a structured research stance, horizon, scenarios, and monitoring plan. “Strategy” describes analytical framing, not trade execution.

Challenges the provisional view, identifies hidden dependencies, data-quality problems, tail risks, invalidation conditions, and confidence limits.

Produces the final source-cited research view, integrating analyst evidence, debate, risk review, confidence, limitations, and what to monitor.

Not every role must run in every case. An on-chain analyst may be irrelevant to a conventional equity filing. A fundamental analyst may have little evidence for a short-lived market microstructure event. The supervisor adapts the plan to the subject and available data.

The run detail exposes phase and step progress, specialist summaries, evidence, and limitations. It does not expose private prompts, hidden reasoning traces, model costs, or operational secrets.