University of Göttingen researcher on LLM agents for automated experiments and conversational meeting participation
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
Output metrics are not yet available: no publication counts, venue list, or yearly volume feed this axis. Once a fuller bibliographic snapshot exists, it would measure the scale and consistency of this researcher's publication record.
Impact—
Impact data is unavailable: citation counts, h-index, and per-paper influence are not in the current snapshot, so no score is offered; this axis would measure citation-based influence across the full publication record once ingested.
Focus0/100
Focus scores at the bottom of the scale (0/100) on current evidence: the reviewed work splits evenly across two distinct fields — natural language processing for conversational AI agents and recommender systems / LLM-based research automation (0.5 / 0.5 field distribution). The two 2026 papers share an LLM-agent core but target different application domains, so the profile reads as a deliberately broad, agent-centric agenda rather than a single narrow specialty.
Collaboration44/100
Collaboration scores moderately (44/100). The reviewed record lists 8 distinct co-authors, with a recurring circle that includes Muneeb Khan, Frederic Kirstein, Terry Ruas, Joeran Beel, and Justus Krell among others. Both 2026 papers are multi-author efforts — a middle authorship position on AutoRecLab and the last position on Speak for Me — suggesting steady team-based work rather than solo writing or large-consortium output.
Activity—
Activity metrics are not yet available: no per-year publication timeline beyond the two 2026 arXiv papers feeds this axis. Once a fuller record is ingested, it would measure publishing cadence and recent productivity year over year.
Reproducibility78/100
Reproducibility scores high (78/100, the mean across 2 assessed papers). Evolve rates AutoRecLab's reproducibility readiness as high — natural-language prompts compile into runnable recommender-system experiments, with 8 of 9 runs succeeding at roughly US$1 per run — while Speak for Me is rated medium, since LLM-judge credit assignment and offline replay weaken confidence in the online agent's behavior.
Research summary
Bela Gipp is a University of Göttingen researcher (ORCID 0000-0001-6522-3019) whose reviewed work converges on a single theme: turning large language models from text generators into agents that act inside real systems. The Evolve catalog contains two 2026 arXiv papers. AutoRecLab ('Describe the Experiment, Get the Code!', middle authorship) compiles natural-language descriptions into runnable recommender-system experiments, succeeding on 8 of 9 runs at roughly US$1 each; Evolve calls it a promising proof of concept while noting success is LLM-judged and evaluation thin, and rates reproducibility readiness high. Speak for Me (last authorship) gives LLMs situational awareness to participate in meetings: structured meeting state cuts agent silence on the AMI corpus from 51.4% to 2.5%, though LLM-judge credit assignment and offline replay temper confidence, and reproducibility readiness is rated medium. Radar metrics mirror this two-front profile: an even field split (0.5/0.5) between conversational-AI NLP and LLM-based research automation yields a low focus score (0/100); collaboration scores moderately (44/100) across 8 distinct co-authors including Muneeb Khan, Frederic Kirstein, Terry Ruas, and Joeran Beel; and reproducibility scores high (78/100, mean over 2 assessed papers). Output, impact, and activity axes currently lack underlying data and are left unscored rather than estimated.
Collaboration pattern
The reviewed record shows a compact, team-based pattern: 8 distinct co-authors, with a recurring circle that includes Muneeb Khan, Frederic Kirstein, Terry Ruas, Moritz Baumgart, Philipp Meister, Justus Krell, Michael Schmidt, and Joeran Beel. Authorship positions vary — middle on AutoRecLab and last on Speak for Me, with neither paper listing this researcher as corresponding author — consistent with contributing inside a stable group rather than solo-first or large-consortium work.
Research keywords
Highlights
AutoRecLab: describing an experiment in plain language yields runnable RecSys code
Structured meeting state cuts agent silence on AMI from 51.4% to 2.5%
A two-front LLM-agent agenda: automating experiments and joining meetings
Across both 2026 papers the same agent-building toolkit — LLM agents, retrieval-augmented generation, tree search, POMDP-inspired state modeling — is aimed at two targets: running recommender-system experiments automatically and giving models situational awareness in human meetings, an even 0.5/0.5 field split that underlies the low focus score.
Timeline
- 2026Publishes AutoRecLab on arXiv: natural-language experiment descriptions become runnable recommender-system code, with 8 of 9 runs succeeding at roughly US$1 each.
- 2026Publishes Speak for Me on arXiv: structured meeting state lowers agent silence on the AMI corpus from 51.4% to 2.5%.
- 2026Both Evolve-reviewed papers appear in the same year, marking a two-front LLM-agent agenda spanning research automation and conversational meeting participation.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Profile metrics derive from Evolve's bibliometric snapshot and its review of only two papers; coverage of the full publication record is incomplete.
- Output, impact, and activity axes lack underlying data (no citation counts, h-index, or yearly volumes) and are left unscored.
- Both reviewed papers are 2026 arXiv preprints; peer-reviewed venue acceptance is not reflected in this data.
- Country is absent from the source record; identity is anchored only by the provided ORCID (0000-0001-6522-3019) and the University of Göttingen affiliation.