Doron Porat

—

Unclaimed · AI-structured profileOpenAlex

Builds deterministic synthetic-enterprise benchmarks with exact ground truth for evaluating LLM agents

papers here

reviewed by Evolve

first-author

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—

No total-publication count is available in this snapshot, so output is not scored. This axis would measure the volume of works over time; only the 3 reviewed 2026 arXiv papers are currently verified, which is not enough to estimate lifetime output.

Impact—

Citation-based impact is not yet scored: h-index and i10-index are recorded as 0 and no citation counts are available, which for a 2026-only record reflects recency rather than measured influence. This axis would track citations and uptake.

Focus0/100

Focus scores 0, derived from an even field distribution: the three assigned fields — LLM agent benchmarking, LLM agent evaluation, and synthetic data generation for agentic AI — each hold 0.333 of the record. In practice the labels describe one coherent enterprise-agent-benchmark agenda, but the classifier splits output evenly; topic counts (Topic Modeling 3, AI Applications 2, NLP 1) still indicate a narrow topical core.

Collaboration39/100

Collaboration scores 39 (0-100 scale). The record shows 6 distinct coauthors, and all three cataloged 2026 arXiv papers are written with the same close-knit team — Benjamin Gruenbaum, Assaf Natanzon, Roy Zavida, Chen Dinachi, Or Itzahary, and Omer Niv. Doron Porat holds middle author positions on every paper with no corresponding-author role, consistent with a small, stable, repeat-collaboration group rather than a wide external network.

Activity40/100

Activity scores 40. All measured output falls in 2026: countsByYear records 3 works in that year, and recent3yWorks is 3 — the entire visible record is recent. No earlier publication activity appears in this snapshot, so the score reflects a short observable window rather than a slowdown; a year-over-year trend cannot be computed from three same-year papers.

Reproducibility43/100

Reproducibility scores 43, the mean readiness across 3 assessed papers. Evolve rates two medium (the hidden-knowledge benchmark extension and the synthetic-enterprise-data paper) and one low (the Era by Eon benchmark paper, whose headline evaluation covers one small estate and whose code is unreleased). Recurring weaknesses noted in reviews include single-seed evaluation, circular evaluation, and missing external validation.

Research summary

Doron Porat is an early-stage researcher whose visible record centers on one coherent problem: making large language model (LLM) agents measurable in enterprise settings. Three arXiv papers, all from 2026 and all written with the same six-person team, assemble a benchmark stack called Era by Eon. The foundational paper seed-generates fictional companies — a synthetic enterprise estate — with exact ground-truth answer keys, so agent answers can be scored deterministically instead of being judged by another model. A companion paper handles the data-engineering side: deterministic synthesis of consistent multi-system business data, plus reference-free evaluation built around an adversarial detector. The third extends the benchmark with eight deterministic hidden-knowledge questions probing whether agents can surface information buried in enterprise records. Evolve's reviews credit the work with elegant constructs and honest limitation sections, while flagging single-seed evaluation, circular evaluation, internal count errors, and unreleased code; reproducibility readiness averages 43 across the three papers. Bibliometric coverage comes from OpenAlex only: 6 distinct coauthors, middle author positions on every paper with no corresponding-author role, an even three-way field split (0.333 each), and h-index and i10-index of 0 — expected for a 2026-only record. Technically distinctive, but too new for citation-based impact to be assessed.

Collaboration pattern

Every cataloged paper is a team effort with the same six coauthors — Benjamin Gruenbaum, Assaf Natanzon, Roy Zavida, Chen Dinachi, Or Itzahary, and Omer Niv — giving 6 distinct collaborators overall. Doron Porat sits in a middle author position on all three papers, with no corresponding-author role recorded, suggesting a stable in-house group producing the Era by Eon benchmark stack together rather than an externally networked collaboration pattern.

Research keywords

LLM agentsagent benchmarkingsynthetic enterprise datadeterministic data generationexact ground truthreference-free evaluationhidden knowledge evaluationentity graphstool use evaluationquestion-conditioned generation

Highlights

Generated enterprise estate with exact ground truth for agent benchmarks
The Era by Eon benchmark seed-generates fictional companies so LLM agents can be scored against exact answer keys rather than model judgment. Evolve judges the validation sound, but notes the headline evaluation covers one small estate and code is unreleased (reproducibility: low).
Deterministic synthesis of consistent multi-system enterprise data
A companion paper generates consistent business data across systems deterministically and proposes reference-free evaluation with an adversarial detector. Evolve calls the engineering sound but finds the evaluation circular and lacking external validation (reproducibility: medium).
Probing hidden knowledge in enterprise agents
Adds eight deterministic hidden-knowledge questions to Era by Eon, testing whether agents can surface information buried in enterprise records. Evolve praises the elegant construct and honest limits, while flagging single-seed evaluation and internal count errors (reproducibility: medium).

Timeline

2026Published 'The Era by Eon Benchmark', introducing a seed-generated enterprise estate with exact ground truth for benchmarking LLM agents.
2026Published 'Generating a Consistent Enterprise', presenting deterministic synthesis of multi-system business data with reference-free evaluation.
2026Extended the benchmark with 'Era by Eon: Benchmarking Enterprise Agents on Hidden Knowledge', adding eight deterministic hidden-knowledge questions.

Papers in the Evolve catalog

Frequent co-authors

Benjamin Gruenbaum×3 Assaf Natanzon×3 Roy Zavida×3 Chen Dinachi×3 Or Itzahary×3 Omer Niv×1

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.

Bibliometric scores are derived from OpenAlex records only (identity: openalex); no other citation or publication databases are included.
Institution, country, and ORCID are unavailable; this profile is name-resolved from public bibliographic data and may not uniquely disambiguate the individual.
The visible record spans only 2026 and covers the 3 arXiv papers reviewed by Evolve; no earlier career history, awards, or affiliations are known.
h-index and i10-index are recorded as 0, consistent with very recent output; citation-based impact cannot yet be assessed.