Or Itzahary

—

Unclaimed · AI-structured profileOpenAlex

AI researcher building synthetic enterprise estates with exact ground truth to benchmark LLM agents (Era by Eon)

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—

Output cannot be scored yet: the snapshot records only 3 works in the recent three-year window, far too thin to gauge productivity or venue mix; this axis would measure publication volume, rate, and venue tiers.

Impact—

Impact data is insufficient for scoring: the profile shows an h-index of 2 and an i10-index of 0, so citation influence cannot yet be characterized; this axis would measure citation accumulation and uptake in follow-up work.

Focus0/100

Focus scores 0 because the field distribution splits exactly three ways — LLM agent benchmarking (0.333), LLM agent evaluation (0.333), and synthetic data generation for agentic AI (0.333) — with no dominant area. In substance the three overlap heavily, since all cataloged papers serve one Era by Eon benchmark pipeline, so the low score reflects even tagging across near-identical topics rather than genuinely scattered research.

Collaboration39/100

Collaboration scores 39 from 6 distinct coauthors in the snapshot. All three cataloged papers are multi-author, with Itzahary in middle or last position and never corresponding author, suggesting a fixed role inside a small, stable team. Recurring collaborators are Benjamin Gruenbaum, Doron Porat, Assaf Natanzon, Roy Zavida, Chen Dinachi, and Omer Niv, and the same group spans the entire Era by Eon series.

Activity40/100

Activity scores 40. The OpenAlex counts-by-year profile shows 2 works in 2019 and 3 in 2026, with no recorded output in between, and 3 works fall inside the recent three-year window. Read alongside the catalog, current activity is a compact 2026 burst — three arXiv preprints in one series — rather than sustained annual output across the intervening years.

Reproducibility43/100

Reproducibility scores 43, the mean across 3 assessed papers. Evolve's reviews pin down concrete limits: one benchmark extends evaluation with a single seed and contains internal count errors; the synthetic-data paper's reference-free evaluation is circular and externally unvalidated; and the core benchmark paper has unreleased code. Readiness is rated medium, medium, and low — determinism is the strength, artifact release the gap.

Research summary

Or Itzahary is an early-career AI researcher whose visible work forms one coherent program: building synthetic enterprise data and deterministic benchmarks for evaluating LLM agents. Three 2026 arXiv papers reviewed in the Evolve catalog — "The Era by Eon Benchmark", "Generating a Consistent Enterprise", and "Era by Eon: Benchmarking Enterprise Agents on Hidden Knowledge" — read as a pipeline: seed-generated fictional companies with exact answer keys, deterministic multi-system business data with reference-free evaluation, and hidden-knowledge question suites layered on top. Evolve's reviews credit the work with elegant benchmark constructs and sound engineering while flagging single-seed evaluation, circular reference-free evaluation, internal count errors, and unreleased code, yielding repro-readiness ratings of medium, medium, and low. Bibliometric signals are correspondingly thin: an OpenAlex snapshot records an h-index of 2, an i10-index of 0, 6 distinct coauthors, and 3 works in the recent three-year window, with an even 0.333 split across agent benchmarking, agent evaluation, and synthetic data generation for agentic AI. Itzahary appears in middle or last author positions and never as corresponding author, consistent with a specialist role on an engineering-driven team alongside Benjamin Gruenbaum, Doron Porat, Assaf Natanzon, Roy Zavida, Chen Dinachi, and Omer Niv. Output and impact dimensions remain unscored for lack of data.

Collaboration pattern

Itzahary publishes with a compact, stable team: 6 distinct coauthors, all recurring across the three-paper Era by Eon series — Benjamin Gruenbaum, Doron Porat, Assaf Natanzon, Roy Zavida, Chen Dinachi, and Omer Niv. Author positions are middle or last, never corresponding, consistent with a specialist contributor role in an engineering-driven group rather than an independent lab lead. No cross-institutional or external collaborations appear in the snapshot.

Research keywords

LLM agentsbenchmark generationhidden knowledgeenterprise datasynthetic enterprise datadeterministic generationentity graphreference-free evaluationadversarial detectorquestion-conditioned generationexact ground truthtool use evaluation

Highlights

Generated enterprise estate with exact ground truth
The Era by Eon Benchmark seed-generates fictional companies with exact answer keys for LLM-agent evaluation; Evolve finds the validation sound but notes the headline evaluation covers one small estate and the code is unreleased (repro readiness: low).
Deterministic synthetic multi-system business data
"Generating a Consistent Enterprise" builds deterministic synthetic enterprise data across systems of record with reference-free evaluation; Evolve credits sound engineering but flags a circular evaluation with no external validation (repro readiness: medium).
Hidden-knowledge questions probe enterprise agents
"Era by Eon: Benchmarking Enterprise Agents on Hidden Knowledge" adds 8 deterministic hidden-knowledge questions to the benchmark; Evolve calls the construct elegant and the limits honestly stated, but notes single-seed evaluation and internal count errors (repro readiness: medium).

Timeline

2019OpenAlex records 2 works under this name in this year; titles are not resolved in this profile's data.
2026The Era by Eon Benchmark introduces a generated enterprise estate with exact ground truth for benchmarking LLM agents (arXiv preprint).
2026"Generating a Consistent Enterprise" presents deterministic synthesis and reference-free evaluation of multi-system business data (arXiv preprint).
2026"Era by Eon: Benchmarking Enterprise Agents on Hidden Knowledge" extends the benchmark with 8 deterministic hidden-knowledge questions (arXiv preprint).

Papers in the Evolve catalog

Frequent co-authors

Benjamin Gruenbaum×3 Doron Porat×3 Assaf Natanzon×3 Roy Zavida×3 Chen Dinachi×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.

Bibliometrics come from an OpenAlex snapshot only (identity: openalex); institution, country, and ORCID are unconfirmed, so this profile is name-resolved from public bibliographic data.
OpenAlex topic associations include respiratory themes (asthma, COPD, inhalation drug delivery, topic modeling) that do not match the cataloged arXiv agent-benchmark papers, a likely author-disambiguation artifact.
Output and impact are unscored: citation data (h-index 2, i10-index 0) is too thin, and the evidence base is only 3 catalog-reviewed 2026 arXiv preprints.
Reproducibility rests on Evolve's single-pass reviews of these preprints; no independent replication attempts are recorded.