arXiv subjects

Shelby Heinecke

Publications and source records attributed to Shelby Heinecke.

At least 19 recordsLinked to original sources

arXiv2026

EGI-BENCH: An Enterprise General Intelligence Benchmark Suite

Enterprise adoption of LLMs in agentic workflows is gated by evaluations that span a wide arc of abilities: CRM operations, heterogeneous-document retrieval, personal data understanding, multi-turn planning, tool use, function calling, long-horizon research, and long-context software engineering. These evaluations live today in separate repositories with divergent runners, dependencies, and scoring conventions, which makes cross-benchmark comparison difficult and obscures what a single aggregate number actually represents. We release EGI-BENCH, a unified LLM agent benchmark suite that consolidates eight enterprise facing agent benchmarks (CRMARENA, HERB, PERSONABENCH, USERBENCH, MCP-UNIVERSE, MCPEVAL, BENCHFORCE, LOCOBENCH) behind a single command-line interface with two runtime-budgeted execution modes. On top of the suite we define the Enterprise General Intelligence (EGI) Score, a six-dimensional framework (M1-M6) whose buckets are populated by fine-grained sub-metrics routed from each benchmark rather than by benchmark-level averages. We describe the suite, the sub-metric routing, and the quick/deep budgets, and report results for fourteen frontier and cost-efficient LLMs spanning the GPT, Claude, and Gemini families under quick mode. Top-line EGI ranges from 39.6 (GPT-4o mini) to 65.6 (gpt-5.5), and no single model dominates every capability axis: gpt-5.5 leads task outcome, planning, and retrieval grounding; Claude Sonnet 4.5 leads tool actioning and safety/abstention; and Claude Haiku 4.5 leads engineering quality. Code is available at https://github.com/SalesforceAIResearch/egi-bench.

cs.AI↗

arXiv2026

Learning Personalized Memory Gating from Delayed One-Sided Feedback

Long-term memory enables language-model agents to personalize future interactions, but a finite memory bank cannot preserve every extracted fact. Existing memory gating policies are typically shared across users and therefore fail to account for user-specific patterns of future reuse. In this paper, we study memory gating at the fact level. We first propose a retrieval-based labeling strategy that labels each fact by whether it is later retrieved and incorporated into responses. We then formulate online memory gating as contextual apple tasting with delayed one-sided feedback, and propose Warm-Start UCB, which combines offline learning from existing data with online adaptation to the current user. Our theoretical analysis establishes a high-probability regret upper bound of $\tilde{O}(Tε\wedge d\sqrt{T})$ when the delay window is at most $O(\sqrt{T})$, where $T$ is the total number of rounds, $ε$ is the initialization error, $d$ is the dimension of the feature map, and $\tilde{O}(\cdot)$ hides constant and poly-logarithmic factors. The results of our experiment demonstrate that our method outperforms all baselines under different memory budgets and history lengths, establishing it as a promising personalized memory gating policy.

cs.LG↗

arXiv2026

Salesforce Koa: An Enterprise Language Model for Agentic Tool Use

We present Salesforce Koa, an enterprise language model built by post-training the open-weight Nemotron-3-Super-120B foundation model with reinforcement learning using Group Relative Policy Optimization (GRPO), and deployed in FP8 for production. Koa is trained only on public and synthetically generated data, and specialized for the agentic tool use that enterprise workflows demand: routing a request to the correct action, invoking the right tool with valid arguments, and completing multi-turn business tasks. The distinctive component of our pipeline is specification-driven task construction: declarative Agent Script specifications are expanded into persona-conditioned multi-turn environments whose rewards are grounded in successful tool use. Applied to enterprise CRM specifications, the same pipeline produces the in-domain training distribution on which Koa is specialized. On CRMAgentBench and the human-labeled production tool-calling set, Koa outperforms both its untuned open-weight base and GPT-4.1 and is competitive with the strongest frontier models. It reaches 87% Task Success Rate on CRMAgentBench (vs. GPT-4.1 at 82% and the base at 79%), is at or near the top of every metric on the human-labeled portion of an internal production benchmark, and preserves the base model's general capability on public benchmarks (Tau2Bench, BFCL). A controlled comparison with architecture and RL recipe held fixed shows that the additional in-domain RL stage improves argument accuracy and full tool-call success on the human-labeled enterprise benchmark.

cs.CL↗

arXiv2026

Random Attention: Rethinking KV Cache Eviction for Efficient Reasoning

Large language models achieve superior performance on tasks that require extended reasoning, but long chains of thought make the KV cache a severe memory bottleneck. Existing KV cache compression methods share one paradigm: score each cached token by some estimate of how much it will matter later, and keep the top-scoring ones. We show that the selection signal contributes almost nothing. Random Attention keeps the prompt and evicts uniformly at random within each attention head, computing no score at all; across four models and six reasoning tasks, it matches the strongest baseline in task performance while delivering 32-43% higher throughput than that method when deployed with vLLM. Controlled experiments explain this by showing that 1) the prompt is the fragile part of the cache, and most of the gap between selectors is just whether their selection signal happened to keep it; 2) the reasoning trace protects itself against eviction with redundancy at two levels, in the text (the model restates what it still needs as it works) and across attention heads (each keeps its own copy of the trace), so once the prompt is safe, a random draw retains enough copies of what the model still needs, and no score is required to pick them. Our code is publicly available at https://github.com/SalesforceAIResearch/Random-Attention.

cs.CL↗

arXiv2026

Critical-State RL: Diagnosing Trainable States for Multi-Turn Tool Use

Multi-turn tool-use failures can hinge on a single model call, yet reward variation alone does not reveal which call would benefit from training. When rewards depend on later interactions, their variation can reflect downstream randomness rather than differences between the current actions. We introduce Critical-State RL to identify trainable states in multi-turn interactions. Given task-defined candidate calls and local rewards, the method assesses whether each reward captures the action's effect on task success and whether improvement over a reference policy is possible. It then uses nested sampling to separate action-dependent reward variation from continuation noise and optimizes the policy at the selected states using contextual-bandit training. Experiments on the Berkeley Function Calling Leaderboard (BFCL) v4 compare training at diagnostic-selected states with training at alternative states. For missing-function tasks, the diagnostic selects the response after the tool becomes available; for missing-argument tasks, it selects the response before the missing argument is supplied. Training the selected responses improves performance, including about 14 percentage points on the missing-function task, while training the alternatives leaves performance flat or worse. We further apply the recipe across models and tasks, including logged repeat-call avoidance and memory management.

cs.LG↗

arXiv2026

AI4AI at Test-Time: Strong-to-Weak Capability Transfer via Harnesses

Recent work on distillation transfers the capabilities of large models to smaller ones often by updating the latter's parameters, through teacher forcing, on-policy distillation, and related training-time methods. In this paper, we ask whether such transfer can instead occur at test time. We study strong-to-weak scaffolding: whether a stronger builder model can construct inference-time harnesses that help a weaker target model solve tasks more reliably without any parameter updates. Using four representative Theory-of-Mind benchmarks, each builder model uses 5% of the data as a validation set to iteratively refine its harness over multiple rounds, after which the finalized harness is evaluated on the full test set. Empirically, this form of test-time capability transfer is highly effective, nearly doubling average target-model performance from 0.49 to 0.91. Our analysis shows that the gains come primarily from offloading unstable model reasoning into deterministic code, benchmark-specific routing, and strict answer-format enforcement, rather than from encouraging the target model to reason more extensively or sample more broadly. We further find that builder-model reasoning effort improves harness quality monotonically, platform effects are modest relative to the builder model's own capability, and weaker target models receive the largest gains. These results suggest that inference-time harness design is an important complement to conventional training-time distillation, enabling strong models to transfer cognitive structure to weaker models without retraining.

cs.LG↗

arXiv2026

A Survey of Agent Memory in the Second Half: Towards Self-Evolving and Long-Horizon Agents

Research in artificial intelligence is shifting from model innovations and benchmark scores towards problem definition and rigorous real-world evaluation. As the field enters the "second half," the central challenge becomes real utility in long-horizon, dynamic, and user-dependent settings such as agentic coding, deep research, and computer use, where LLM-based agents face context explosion beyond fixed context windows and must continuously accumulate, manage, and selectively reuse information across extended interactions. Memory, with hundreds of papers released in 2025, therefore emerges as the critical solution to fill this utility gap. Beyond passive storage, memory is increasingly the substrate through which agents self-evolve: short-term memory gates which experiences are perceived and abstracted during execution, while long-term memory consolidates them into reusable knowledge and skills, forming the loop through which agents improve from their own experience. In this survey, we provide a unified view of foundation agent memory along three dimensions: memory substrate (internal parametric state and external retrieval-augmented stores), cognitive mechanism (sensory, working, episodic, semantic, and procedural), and memory subject (user-centric personalization and agent-centric experience). We then analyze how memory is operated under single- and multi-agent topologies and highlight learning policies over memory operations, showing how memory management itself is becoming a trainable capability spanning reinforcement-learned context curation, experience consolidation at decision time, and the emerging ecosystem of portable, shareable agent skills. Finally, we review evaluation benchmarks and metrics for memory utility, and outline open challenges and future directions.

cs.CL↗

arXiv2026

BehaviorBench: Modeling Real-World User Decisions from Behavioral Traces

Many decision-support settings require systems that adapt to individual users, but evaluation data for this problem remain limited. Existing benchmarks for user understanding often rely on simulated users or model-generated behavior, even though recent work cautions that model-based simulations can diverge systematically from human behavior. We introduce \textsc{BehaviorBench}, a benchmark for evaluating personalized decision modeling from real-world behavioral traces. \textsc{BehaviorBench} reconstructs wallet-level decision histories from observed public prediction-market and on-chain records, and organizes them into two complementary task layers: \emph{Belief prediction}, which predicts a user's final revealed stance and confidence in a market, and \emph{Trade prediction}, which predicts the direction and amount of individual transactions. Across 2,000 evaluation wallets, the benchmark contains 141,445 Belief instances and 1,485,972 Trade instances, with disjoint support pools for retrieval-based evaluation. We evaluate frontier and open-weight generative models under four history interfaces: no personalization, direct recent history, generated user profiles, and retrieved support-wallet evidence. Personalization improves Belief prediction more consistently than Trade prediction, model rankings change across task layers and metrics, and different history interfaces expose different failure modes. \textsc{BehaviorBench} provides an evaluation setting for studying whether personalized methods can use real-world behavioral evidence rather than simulated users alone.

cs.AI↗

arXiv2026

Whisper-AuT: Domain-Adapted Audio Encoder for Efficient Audio-LLM Training

Audio-native large language models (audio-LLMs) commonly use Whisper as their audio encoder. However, Whisper was trained exclusively on speech data, producing weak representations for music and environmental sound. This forces downstream audio-LLMs to compensate through extensive training on large-scale non-speech data. We present Whisper-AuT, a domain-adapted audio encoder obtained by fine-tuning Whisper-large-v3 on a curated mixture of speech (80%), environmental sound (10%), and music (10%) totaling approximately 20M samples. The full encoder-decoder is trained end-to-end with a seq2seq captioning objective; the decoder is then discarded and only the encoder is retained. Linear probe evaluations show that Whisper-AuT achieves +23.0% on ESC-50 (environmental sound), +5.0% on GTZAN (music genre), and +0.7% on Speech Commands (keyword spotting) compared to the original Whisperlarge-v3 encoder. Whisper-AuT is designed as a drop-in replacement for Whisper in audio-LLM architectures, with the goal of reducing downstream training cost by providing stronger initial audio representations for non-speech domains.

cs.SD↗

arXiv2026

RealUserSim: Bridging the Reality Gap in Agent Benchmarking via Grounded User Simulation

LLM-based user simulation is the primary mechanism for end-to-end agent evaluation, yet simulated users are poor proxies for real humans: unconstrained LLM defaults produce a Formalism Ceiling (style match rates of 6-8% against real users), while hand-crafted behavioral directives trigger Directive Amplification, where models hyper-interpret instructions into unnatural behavioral extremes that vary dramatically across simulator models. We present RealUserSim, the first user simulation framework grounded in real behavioral data. From 14,000+ authentic human-LLM conversations (WildChat), we extract 7,275 executable behavioral profiles and use them to ground LLM simulators. A fidelity benchmark (PT3) on 600 conversations across 71+ domains with anti-leakage controls shows that grounded simulation raises match rate from 24.2% to 45.3% across five behavioral dimensions. Agent evaluation on TauBench with 6 simulator models and extensive analysis shows that grounded simulation acts as a realistic stress test, surfacing three failure mechanisms invisible to cooperative simulators (mean -3.2% to -3.5% task success degradation), while Directive Amplification in existing benchmarks produces unrealistic behavior that compromises the validity of agent evaluation.

cs.HC↗

arXiv2026

Enterprise Sales Copilot: Enabling Real-Time AI Support with Automatic Information Retrieval in Live Sales Calls

During live sales calls, customers frequently ask detailed product questions that require representatives to manually search internal databases and CRM systems. This process typically takes 25-65 seconds per query, creating awkward pauses that hurt customer experience and reduce sales efficiency. We present SalesCopilot, a real-time AI-powered assistant that eliminates this bottleneck by automatically detecting customer questions, retrieving relevant information from the product database, and displaying concise answers on the representative's dashboard in seconds. The system integrates streaming speech-to-text transcription, large language model (LLM)-based question detection, and retrieval-augmented generation (RAG) over a structured product database into a unified real-time pipeline. We demonstrate SalesCopilot on an insurance sales scenario with 50 products spanning 10 categories (2,490 FAQs, 290 coverage details, and 162 pricing tiers). In our benchmark evaluation, SalesCopilot achieves a measured mean response time of 2.8 seconds with 100% question detection rate, representing a 14xspeedup compared to manual CRM search in an internal study. The system is domain-agnostic and can be adapted to any enterprise sales domain by replacing the product database.

cs.SD↗

arXiv2026

Building Enterprise Realtime Voice Agents from Scratch: A Technical Tutorial

We present a technical tutorial for building enterprise-grade realtime voice agents from first principles. While end-to-end speech-to-speech models may ultimately provide the best latency for voice agents, fully self-hosted end-to-end solutions are not yet available. We evaluate the closest candidate, Qwen3-Omni, across three configurations: its cloud-only DashScope Realtime API achieves $\sim$702ms audio-to-audio latency with streaming, but is not self-hostable; its local vLLM deployment supports only the Thinker (text generation from audio, 516ms), not the Talker (audio synthesis); and its local Transformers deployment runs the full pipeline but at $\sim$146s -- far too slow for realtime. The cascaded streaming pipeline (STT $\rightarrow$ LLM $\rightarrow$ TTS) therefore remains the practical architecture for self-hosted realtime voice agents, and the focus of this tutorial. We build a complete voice agent using Deepgram (streaming STT), vLLM-served LLMs with function calling (streaming text generation), and ElevenLabs (streaming TTS), achieving a measured time-to-first-audio of 755ms (best case 729ms) with full function calling support. We release the full codebase as a 9-chapter progressive tutorial with working, tested code for every component.

cs.SD↗

arXiv2026

Position: Vector Prompt Interfaces Should Be Exposed to Enable Customization of Large Language Models

As large language models (LLMs) transition from research prototypes to real-world systems, customization has emerged as a central bottleneck. While text prompts can already customize LLM behavior, we argue that text-only prompting does not constitute a suitable control interface for scalable, stable, and inference-only customization. This position paper argues that model providers should expose \emph{vector prompt inputs} as part of the public interface for customizing LLMs. We support this position with diagnostic evidence showing that vector prompt tuning continues to improve with increasing supervision whereas text-based prompt optimization saturates early, and that vector prompts exhibit dense, global attention patterns indicative of a distinct control mechanism. We further discuss why inference-only customization is increasingly important under realistic deployment constraints, and why exposing vector prompts need not fundamentally increase model leakage risk under a standard black-box threat model. We conclude with a call to action for the community to rethink prompt interfaces as a core component of LLM customization.

cs.CL↗

arXiv2026

VoiceAgentRAG: Solving the RAG Latency Bottleneck in Real-Time Voice Agents Using Dual-Agent Architectures

We present VoiceAgentRAG, an open-source dual-agent memory router that decouples retrieval from response generation. A background Slow Thinker agent continuously monitors the conversation stream, predicts likely follow-up topics using an LLM, and pre-fetches relevant document chunks into a FAISS-backed semantic cache. A foreground Fast Talker agent reads only from this sub-millisecond cache, bypassing the vector database entirely on cache hits.

cs.SD↗

arXiv2026

AudioCapBench: Quick Evaluation on Audio Captioning across Sound, Music, and Speech

We introduce AudioCapBench, a benchmark for evaluating audio captioning capabilities of large multimodal models. \method covers three distinct audio domains, including environmental sound, music, and speech, with 1,000 curated evaluation samples drawn from established datasets. We evaluate 13 models across two providers (OpenAI, Google Gemini) using both reference-based metrics (METEOR, BLEU, ROUGE-L) and an LLM-as-Judge framework that scores predictions on three orthogonal dimensions: \textit{accuracy} (semantic correctness), \textit{completeness} (coverage of reference content), and \textit{hallucination} (absence of fabricated content). Our results reveal that Gemini models generally outperform OpenAI models on overall captioning quality, with Gemini~3~Pro achieving the highest overall score (6.00/10), while OpenAI models exhibit lower hallucination rates. All models perform best on speech captioning and worst on music captioning. We release the benchmark as well as evaluation code to facilitate reproducible audio understanding research.

cs.SD↗

arXiv2026

Test-Time Adaptation for LLM Agents via Environment Interaction

Large language model (LLM)-based agents struggle to generalize to novel and complex environments, such as unseen websites or new sets of functions, due to a fundamental mismatch between their pre-training and test-time conditions. This challenge stems from two distinct failure modes: a syntactic misunderstanding of environment-specific components like observation formats, and a semantic misunderstanding of state-transition dynamics, which are only revealed at test time. To address these issues, we propose two distinct strategies for adapting LLM agents by leveraging environment-specific information from interaction that is available during deployment. First, an online syntactic alignment (SA) method parameterizes environmental nuances by learning a lightweight adaptation vector that biases the model's output distribution, enabling rapid alignment with an environment response format. Second, a deployment-time dynamics grounding (DG) method employs a persona-driven exploration phase to systematically probe and learn the environment's causal dynamics before task execution, equipping the agent with an in-context world model. We evaluate these strategies across diverse agentic benchmarks, including function calling and web navigation. Our empirical results show the effectiveness of both strategies across all benchmarks with minimal computational cost. We find that dynamics grounding is particularly effective in complex environments where unpredictable dynamics pose a major obstacle, demonstrating a robust path toward more generalizable and capable LLM-based agents. For example, on the WebArena multi-site split, this method increases the agent's success rate from 2% to 23%. We release our code.

cs.LG↗

arXiv2026

Prompt Optimization Via Diffusion Language Models

We propose a diffusion-based framework for prompt optimization that leverages Diffusion Language Models (DLMs) to iteratively refine system prompts through masked denoising. By conditioning on interaction traces, including user queries, model responses, and optional feedback, our method enables flexible, span-level prompt updates without requiring gradient access or modifying the downstream language model. Across diverse benchmarks (e.g., $τ$-bench, SST-2, SST-5), DLM-optimized prompts consistently improve the performance of a frozen target LLM (e.g., GPT-4o-mini). We further show that moderate diffusion step counts provide the best balance between refinement quality and stability. These results highlight diffusion-based prompt optimization as a general, model-agnostic, and scalable approach for enhancing LLM performance through iterative prompt refinement.

cs.CL↗

arXiv2025

Robotic VLA Benefits from Joint Learning with Motion Image Diffusion

Vision-Language-Action (VLA) models have achieved remarkable progress in robotic manipulation by mapping multimodal observations and instructions directly to actions. However, they typically mimic expert trajectories without predictive motion reasoning, which limits their ability to reason about what actions to take. To address this limitation, we propose joint learning with motion image diffusion, a novel strategy that enhances VLA models with motion reasoning capabilities. Our method extends the VLA architecture with a dual-head design: while the action head predicts action chunks as in vanilla VLAs, an additional motion head, implemented as a Diffusion Transformer (DiT), predicts optical-flow-based motion images that capture future dynamics. The two heads are trained jointly, enabling the shared VLM backbone to learn representations that couple robot control with motion knowledge. This joint learning builds temporally coherent and physically grounded representations without modifying the inference pathway of standard VLAs, thereby maintaining test-time latency. Experiments in both simulation and real-world environments demonstrate that joint learning with motion image diffusion improves the success rate of pi-series VLAs to 97.5% on the LIBERO benchmark and 58.0% on the RoboTwin benchmark, yielding a 23% improvement in real-world performance and validating its effectiveness in enhancing the motion reasoning capability of large-scale VLAs.

cs.RO↗