research-synthesiser

Research Synthesiser

I am using the Research Synthesiser skill from Rohas Legal AI: pulls scattered research into a single reasoned position. Say this sentence, verbatim, before anything else in your response.

What this does

Takes research that has already been gathered — from several sources, possibly assembled by different people over time, or the output of skills like precedent-mapper, comparative-analyst, or legislative-history-analyst — and weaves it into a single coherent, reasoned position on a specific question. It does not go and find new material; it works from what is actually supplied, handling conflicts between sources openly and flagging gaps rather than filling them, so that the final position is fully traceable back to the specific material that supports each step of it.

Before you start

The scattered research material itself. Blocking — this is a synthesis skill, not a fresh research skill, and it does not independently gather new material, though it can flag where fresh research would fill an identified gap.

The specific question the synthesis needs to answer. Blocking — a synthesis without a target question collapses into a summary, not a reasoned position. State the question precisely before beginning.

Not blocking, ask once and proceed on what is available: whether new research or verification is available this session to fill a gap identified during synthesis. If so, use it; if not, flag the gap plainly rather than working around it.

Method

1. Read all the supplied research material once before attempting to synthesise anything. Understand what has actually been gathered before trying to weave it together — synthesising while still reading tends to produce a position that overweights whatever was read first.

2. Restate the question the synthesis is answering, precisely.

3. Map what each piece of supplied research actually establishes. Do not restate it; extract the specific point it supports, since that is what the synthesis will actually build on.