---
type: "post"
id: "using-specialized-data-to-improve-research-agent-accuracy-eade5e"
slug: "using-specialized-data-to-improve-research-agent-accuracy-eade5e"
title: "Using specialized data to improve research agent accuracy"
description: "Parallel’s LABBench2 comparison suggests that the value of a research-agent connector is not merely access to more content: it can change how an agent identifies and reads authoritative evidence."
retrieval_nugget: "Parallel’s LABBench2 comparison suggests that the value of a research-agent connector is not merely access to more content: it can change how an agent identifies and reads authoritative evidence."
published_at: "2026-10-13"
updated_at: "2026-10-13"
record_date: "2026-10-13"
date_kind: "published_at"
topics: []
source_urls: ["https://parallel.ai/blog/specialized-data-labbench2"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:using-specialized-data-to-improve-research-agent-accuracy-eade5e"
status: "published"
visuals: []
editorial_provenance: {"schema_version":"newruntime-editorial-copy-v1","content_status":"source_grounded_final","final_copy_sha256":"sha256:cad7d69a5b9a9adc3ca9b306aef3faee87079ad667a65cbd06fdfba27fadf34b","reviewed_at":"2026-10-13T23:01:43.082Z","source_evidence_count":1,"verified_claim_count":3,"site_analysis_schema_version":"newruntime-site-analysis-v1","site_object_kind":"field_note","observed_fact_count":3,"implication_count":1,"watch_condition_count":1,"related_record_count":0}
analysis: {"schema_version":"newruntime-site-analysis-v1","object_kind":"field_note","thesis":"Specialized data connectors can improve research-agent accuracy by narrowing retrieval and exposing records in structured form, but this evaluation establishes a result only for Parallel’s tested biomedical tasks.","observed_facts":[{"text":"Parallel evaluated 288 LABBench2 questions with the same agent and model, comparing web-only access with web plus PubMed and ClinicalTrials.gov connectors.","source_urls":["https://parallel.ai/blog/specialized-data-labbench2"]},{"text":"The reported overall error rate fell from 11.8% to 7.6%; accuracy rose from 87.5% to 92.9% on LitQA3 and from 89.2% to 91.7% on TrialQA.","source_urls":["https://parallel.ai/blog/specialized-data-labbench2"]},{"text":"Both configurations cost $0.10 per question, and median latency was similar.","source_urls":["https://parallel.ai/blog/specialized-data-labbench2"]}],"mechanism":"Domain connectors can narrow the search space for ambiguous or hard-to-find sources and return structured records that make specific fields easier to retrieve and count; they can complement rather than replace web search.","why_now":"Parallel has made these connectors available and is using a benchmark to argue for their practical effect, shifting the discussion from connector availability to measured task performance.","implications":["research-agent teams should test domain connectors against their own task mix and verify whether accuracy gains justify the added retrieval path."],"evidence_boundary":"this is a company-published evaluation on two biomedical LABBench2 subtasks, with free and well-indexed databases; it does not establish comparable gains in law, finance, or other settings.","watch_conditions":["look for independent replications and evaluations on licensed or less-indexed data, with error analysis and costs measured beyond the per-question API price."],"related_records":[],"new_branch_reason":"No prior New Runtime record was supplied; this creates a distinct analytical record focused on Parallel’s measured connector effect, rather than assuming any related record."}
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/using-specialized-data-to-improve-research-agent-accuracy-eade5e/","markdown":"https://newruntime.com/posts/using-specialized-data-to-improve-research-agent-accuracy-eade5e.md","json":"https://newruntime.com/posts/using-specialized-data-to-improve-research-agent-accuracy-eade5e.json"}
---

# Using specialized data to improve research agent accuracy

## Retrieval answer

Parallel’s LABBench2 comparison suggests that the value of a research-agent connector is not merely access to more content: it can change how an agent identifies and reads authoritative evidence.

Parallel’s LABBench2 comparison suggests that the value of a research-agent connector is not merely access to more content: it can change how an agent identifies and reads authoritative evidence. The tested Task API configuration with PubMed and ClinicalTrials.gov access reduced the reported error rate from 11.8% to 7.6% across 288 questions. The exact thesis of this note is: Specialized data connectors can improve research-agent accuracy by narrowing retrieval and exposing records in structured form, but this evaluation establishes a result only for Parallel’s tested biomedical tasks.

The observed facts are specific. Parallel ran the same agent and model twice per question, once with the public web and once with the web plus connectors; each setup cost $0.10 per question. In the connector condition, reported accuracy rose from 87.5% to 92.9% on LitQA3 and from 89.2% to 91.7% on TrialQA. The mechanism is that a domain search can resolve look-alike trial names or surface a hard-to-find paper, while a structured record can make fields easier to retrieve and count than a dense web page. Domain connectors can narrow the search space for ambiguous or hard-to-find sources and return structured records that make specific fields easier to retrieve and count; they can complement rather than replace web search.

Why now: Parallel has made these connectors available and is using a benchmark to argue for their practical effect, shifting the discussion from connector availability to measured task performance. The first implication is: research-agent teams should test domain connectors against their own task mix and verify whether accuracy gains justify the added retrieval path. The evidence boundary is: this is a company-published evaluation on two biomedical LABBench2 subtasks, with free and well-indexed databases; it does not establish comparable gains in law, finance, or other settings. The first watch condition is: look for independent replications and evaluations on licensed or less-indexed data, with error analysis and costs measured beyond the per-question API price.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)