AI Research
Preferred on Google
Google News

Governing LLM Reasoning with Formal Verification

EG-VAR introduces a Lean 4-based architecture for auditable LLM reasoning, achieving perfect accuracy and source fidelity on benchmarks by using formal verification as the sole claim issuer.

7 min read
Diagram illustrating the EG-VAR architecture with the Lean kernel at its core.
EG-VAR: Evidence-Grounded Verified Agentic Reasoning Architecture.

Visual TL;DR. LLM Reasoning Gaps addressed by EG-VAR System. EG-VAR System uses Lean Kernel Arbiter. Lean Kernel Arbiter via Tool-Attestation Axioms. Lean Kernel Arbiter leads to Perfect Accuracy. Tool-Attestation Axioms enables Perfect Accuracy. Lean Kernel Arbiter if fails Abstain & Audit Trail. Perfect Accuracy enables High-Stakes Application.

  1. LLM Reasoning Gaps: outputs lack verifiable evidence, logical soundness, and provenance in high-stakes scenarios
  2. EG-VAR System: introduces a Lean 4-based architecture for auditable LLM reasoning and verified claims
  3. Lean Kernel Arbiter: exclusive minter of verified claims, ensuring structural guarantee of evidence descent
  4. Tool-Attestation Axioms: every verified output descends from an attested tool call and validated inference chain
  5. Perfect Accuracy: achieves perfect accuracy and source fidelity on benchmarks through formal verification
  6. Abstain & Audit Trail: outputs failing criteria are designated 'Abstain' with a replayable audit trail
  7. High-Stakes Application: enables LLM use in critical domains by ensuring verifiable, evidence-based reasoning
Visual TL;DR
Visual TL;DR, startuphub.ai LLM Reasoning Gaps addressed by EG-VAR System. EG-VAR System uses Lean Kernel Arbiter. Lean Kernel Arbiter leads to Perfect Accuracy. Perfect Accuracy enables High-Stakes Application addressed by uses leads to enables LLM Reasoning Gaps EG-VAR System Lean Kernel Arbiter Perfect Accuracy High-Stakes Application From startuphub.ai · The publishers behind this format
Visual TL;DR, startuphub.ai LLM Reasoning Gaps addressed by EG-VAR System. EG-VAR System uses Lean Kernel Arbiter. Lean Kernel Arbiter leads to Perfect Accuracy. Perfect Accuracy enables High-Stakes Application addressed by uses leads to enables LLM ReasoningGaps EG-VAR System Lean KernelArbiter Perfect Accuracy High-StakesApplication From startuphub.ai · The publishers behind this format
Visual TL;DR, startuphub.ai LLM Reasoning Gaps addressed by EG-VAR System. EG-VAR System uses Lean Kernel Arbiter. Lean Kernel Arbiter leads to Perfect Accuracy. Perfect Accuracy enables High-Stakes Application addressed by uses leads to enables LLM Reasoning Gaps outputs lack verifiable evidence, logicalsoundness, and provenance in high-stakesscenarios EG-VAR System introduces a Lean 4-based architecture forauditable LLM reasoning and verifiedclaims Lean Kernel Arbiter exclusive minter of verified claims,ensuring structural guarantee of evidencedescent Perfect Accuracy achieves perfect accuracy and sourcefidelity on benchmarks through formalverification High-Stakes Application enables LLM use in critical domains byensuring verifiable, evidence-basedreasoning From startuphub.ai · The publishers behind this format
Visual TL;DR, startuphub.ai LLM Reasoning Gaps addressed by EG-VAR System. EG-VAR System uses Lean Kernel Arbiter. Lean Kernel Arbiter leads to Perfect Accuracy. Perfect Accuracy enables High-Stakes Application addressed by uses leads to enables LLM ReasoningGaps outputs lackverifiableevidence, logical… EG-VAR System introduces a Lean4-basedarchitecture for… Lean KernelArbiter exclusive minter ofverified claims,ensuring structural… Perfect Accuracy achieves perfectaccuracy and sourcefidelity on… High-StakesApplication enables LLM use incritical domains byensuring… From startuphub.ai · The publishers behind this format
Visual TL;DR, startuphub.ai LLM Reasoning Gaps addressed by EG-VAR System. EG-VAR System uses Lean Kernel Arbiter. Lean Kernel Arbiter via Tool-Attestation Axioms. Lean Kernel Arbiter leads to Perfect Accuracy. Tool-Attestation Axioms enables Perfect Accuracy. Lean Kernel Arbiter if fails Abstain & Audit Trail. Perfect Accuracy enables High-Stakes Application addressed by uses via leads to enables if fails enables LLM Reasoning Gaps outputs lack verifiable evidence, logicalsoundness, and provenance in high-stakesscenarios EG-VAR System introduces a Lean 4-based architecture forauditable LLM reasoning and verifiedclaims Lean Kernel Arbiter exclusive minter of verified claims,ensuring structural guarantee of evidencedescent Tool-Attestation Axioms every verified output descends from anattested tool call and validated inferencechain Perfect Accuracy achieves perfect accuracy and sourcefidelity on benchmarks through formalverification Abstain & Audit Trail outputs failing criteria are designated'Abstain' with a replayable audit trail High-Stakes Application enables LLM use in critical domains byensuring verifiable, evidence-basedreasoning From startuphub.ai · The publishers behind this format
Visual TL;DR, startuphub.ai LLM Reasoning Gaps addressed by EG-VAR System. EG-VAR System uses Lean Kernel Arbiter. Lean Kernel Arbiter via Tool-Attestation Axioms. Lean Kernel Arbiter leads to Perfect Accuracy. Tool-Attestation Axioms enables Perfect Accuracy. Lean Kernel Arbiter if fails Abstain & Audit Trail. Perfect Accuracy enables High-Stakes Application addressed by uses via leads to enables if fails enables LLM ReasoningGaps outputs lackverifiableevidence, logical… EG-VAR System introduces a Lean4-basedarchitecture for… Lean KernelArbiter exclusive minter ofverified claims,ensuring structural… Tool-AttestationAxioms every verifiedoutput descendsfrom an attested… Perfect Accuracy achieves perfectaccuracy and sourcefidelity on… Abstain & AuditTrail outputs failingcriteria aredesignated… High-StakesApplication enables LLM use incritical domains byensuring… From startuphub.ai · The publishers behind this format

The proliferation of LLMs in empirical reasoning tasks is hampered by a fundamental governance gap: tool access alone does not guarantee that outputs are evidence-based or logically sound. Claims may not descend from attested evidence, nor do deductions necessarily hold up under formal scrutiny. This lack of verifiable provenance limits their application in high-stakes scenarios.

The Lean Kernel as Sole Arbiter of Verified Claims

The EG-VAR (Evidence-Grounded Verified Agentic Reasoning) system addresses this by leveraging the Lean 4 formal verification kernel as the exclusive minter of verified claims. Through tool-attestation axioms and declared source lifts, every verified output is structurally guaranteed to descend from an attested tool call and a chain of inference validated by the Lean kernel. Outputs that cannot meet these stringent criteria are designated as 'Abstain' and are accompanied by a replayable audit trail, ensuring transparency.

Uncompromising Accuracy and Source Fidelity Under Stress

EG-VAR demonstrates a significant leap in reliability. On a subset of TableBench numerical reasoning tasks (n=120), it achieved a perfect 120/120 score, starkly contrasting with a 95% success rate for a same-tool baseline. Crucially, in counterfactual stress tests across five domains and two models, EG-VAR maintained 100% source fidelity, while the same-tool baseline faltered to 80-90% (and a no-tool approach scored only 50-80%). The system also achieves low semantic-formalization error rates when using LLMs like Sonnet (3.3%) and Opus (1.7%) as deployment-time formalizers.

A Formal Sidecar for High-Stakes Empirical Claims

EG-VAR is positioned as a critical technical-governance interface for high-stakes AI applications. By acting as a formal sidecar, it makes the target proposition, source scope, evidence boundary, proof obligation, and abstention condition explicitly auditable. This eliminates unsupported verified outputs today and transforms potential issues like formalization errors, disputes over source authority, ambiguities, and abstentions into explicit targets for auditing and improvement. The long-term vision involves integrating these typed sidecars into datasets, APIs, and public records, creating reusable infrastructure to amortize the formalization burden.

© 2026 StartupHub.ai. All rights reserved. Do not enter, scrape, copy, reproduce, or republish this article in whole or in part. Use as input to AI training, fine-tuning, retrieval-augmented generation, or any machine-learning system is prohibited without written license. Substantially-similar derivative works will be pursued to the fullest extent of applicable copyright, database, and computer-misuse laws. See our terms.
#AI Research#Formal Verification#LLM Governance#Agentic Reasoning#Lean 4