On-Demand Research Brief Agent

This agent turns a single research question into a polished, sourced brief. You send it a prompt like "Competitive landscape for vertical-farming startups in 2026" and it fans out parallel searches across Exa's neural search and Perplexity's answer engine, dedupes and ranks the results, then uses an OpenAI model to synthesize a structured brief — executive summary, key findings, and a numbered source list with live URLs. Every claim in the brief is tied back to a retrieved source, so the output is auditable rather than a hallucinated wall of text.

Reach for this when you need a fast, defensible answer and don't have time to open twenty tabs: pre-meeting prep, sizing a new market, vetting a vendor, building a literature scan, or arming a sales rep before a call. It's ideal for teams that want repeatable research output in a consistent format instead of ad-hoc ChatGPT prompts that lose their sources. Use it on demand from chat, or wire the same flow behind a webhook so other tools can request a brief programmatically.

The flow triggers from a chat message containing the research question. It runs two retrieval steps in parallel — Exa for high-recall semantic web results and Perplexity for an already-summarized answer with citations — giving the synthesizer both raw sources and a grounded baseline. A logic step merges the two result sets, drops duplicate domains, and trims to the strongest sources. OpenAI then receives the question plus the consolidated source set with an instruction to write the brief strictly from the provided material and to footnote each claim. The finished brief, with its citation list, is returned to the chat thread as the final output.