4/22/2026
Show HN: AICommander – Reliable System-Level Automation
AICommander: Windows desktop app for AI agents with specialized agent routing, task automation, and browser control via local execution.
4/22/2026
Microsoft to Shift GitHub Copilot Users to Token-Based Billing
Microsoft shifting GitHub Copilot to token-based billing model, pausing new signups for student/individual tiers per leaked documents.
4/22/2026
Rcarmo/go-AI: A mildly sane inference API library for go
Go library providing sane inference API abstraction for LLM integration and model interaction.
4/22/2026
Meta employees are up in arms over a mandatory program to train AI on their
Meta employees resist mandatory keystroke/mouse tracking software installed to train AI models on human computer usage patterns.
4/22/2026
Cursor-autoresearch: pi-autoresearch port. Like Karpathy/autoresearch
MCP-driven optimization loop for code editors enabling automated research workflows with benchmarking and metric tracking.
4/22/2026
Show HN: Tesseron – an API for AI agents, defined by the app developer
Tesseron: API framework allowing developers to define custom interfaces for AI agent capabilities and interactions.
4/22/2026
I asked my AI agent its last wrong belief. It queried its own database
Technical exploration of AI agent cognition: memory structures, belief decay, error detection, and self-query capabilities in agent architecture.
4/22/2026
Mastra Code
Terminal AI coding agent connecting to 70+ models with file reading, searching, editing, and code execution tools; Node.js-based with npm distribution.
4/22/2026
Show HN: We open-sourced a 6-library governance stack for AI agents (Python)
Open-sourced 6-library governance stack (Python) for enterprise AI agent deployment covering reliability, policy, orchestration, monitoring, and identity.
4/22/2026
Show HN: Kdts, an optimization-first TypeScript compiler
Optimization-first TypeScript compiler using type information throughout compilation process. Bun-only with fast and optimized modes.
4/22/2026
Why Every AI-Coded App Is an Island
Analysis of challenges in deploying AI-generated apps: isolated databases, auth, and infrastructure; strategies for integration across generated applications.
4/22/2026
Show HN: LibreThinker, free AI assistant for LibreOffice Writer, 10k installs
LibreOffice Writer AI copilot extension with 10k installs; free online LLM integration, no signup required.
4/22/2026
Why do AI models hallucinate? [video]
Video explaining mechanisms behind AI model hallucinations.
4/22/2026
Show HN: We benchmarked 18 LLMs on OCR (7K+ calls) – cheaper models win
Benchmark comparing OCR performance of 18 LLMs on business documents, measuring quality, consistency, latency, and cost across 7K+ calls.
4/22/2026
Git for web services – everything as a file for coding agents
Overview of coding agents and Model Context Protocol (MCP) for connecting LLM chatbots to external services and tools.
4/22/2026
Deep Research Max
Google releases Deep Research Max agents with MCP support, native visualizations, and autonomous research capabilities via Interactions API.
4/22/2026
Show HN: Data-driven GEO and marketing agent platform
Marketing platform tracking brand presence in AI responses and analyzing user prompts for SEO opportunities via chatbots.
4/22/2026
We found most apps send PII to LLMs and built a 2 line fix
PII scrubber middleware that detects and removes personal information before sending to LLM APIs with 2-line integration.
4/22/2026
Mistral Vibe
Mistral agentic coding assistant handling boilerplate, tests, documentation, and codebase refactoring autonomously.
4/22/2026
Zappa: An AI powered mitmproxy
AI-powered proxy tool enabling autonomous internet browsing and app interaction with near-human behavior.
4/22/2026
Pioneer: Vibetune Your LLMs
Pioneer platform that automatically retrains open-source LLMs on live inference data to improve performance over time.
4/22/2026
Agents with Taste – How to transfer taste into an AI
Method for encoding design preferences into skill files for AI agents to follow visual/creative guidelines.
4/22/2026
Debug2Fix: Can Interactive Debugging Help Coding Agents Fix More Bugs?
Framework enabling coding agents to fix bugs using interactive debugging tools, improving upon static analysis and test-fix cycles.
4/22/2026
PhysMem: Scaling Test-time Physical Memory for Robot Manipulation
Memory framework enabling vision-language model robot planners to learn and apply physical property predictions for manipulation tasks.
4/22/2026
From Verbatim to Gist: Distilling Pyramidal Multimodal Memory via Semantic Information Bottleneck for Long-Horizon Video Agents
Long-horizon video understanding for multimodal agents using semantic information bottleneck to compress visual memory into gist representations.
4/22/2026
AD-Copilot: A Vision-Language Assistant for Industrial Anomaly Detection via Visual In-context Comparison
AD-Copilot: vision-language assistant for industrial anomaly detection using visual in-context comparison and multimodal learning.
4/22/2026
ExpertGen: Scalable Sim-to-Real Expert Policy Learning from Imperfect Behavior Priors
ExpertGen: framework automating expert policy learning in simulation for scalable sim-to-real transfer in robotic behavior cloning.
4/22/2026
CounterRefine: Answer-Conditioned Counterevidence Retrieval for Inference-Time Knowledge Repair in Factual Question Answering
CounterRefine: inference-time repair layer for RAG systems gathering counterevidence to fix wrong answers despite relevant evidence retrieval.
4/22/2026
Bridging the High-Frequency Data Gap: A Millisecond-Resolution Network Dataset for Advancing Time Series Foundation Models
Millisecond-resolution time series dataset for training foundation models capturing high-frequency temporal patterns across domains.
4/22/2026
Resolving the Robustness-Precision Trade-off in Financial RAG through Hybrid Document-Routed Retrieval
RAG system for financial document Q&A using hybrid document-routed retrieval to resolve chunk confusion in structurally homogeneous corpora.
4/22/2026
Council Mode: Mitigating Hallucination and Bias in LLMs via Multi-Agent Consensus
Council Mode: multi-agent consensus approach mitigating hallucinations and bias in MoE LLMs through coordinated expert activation.
4/22/2026
Fine-tuning DeepSeek-OCR-2 for Molecular Structure Recognition
DeepSeek-OCR-2 adapted for molecular structure recognition from 2D diagrams using vision-language model fine-tuning for chemistry applications.
4/22/2026
GAIN: Multiplicative Modulation for Domain Adaptation
GAIN: multiplicative modulation technique for LLM domain adaptation preserving pretrained weight matrix column span to prevent catastrophic forgetting.
4/22/2026
MoBiE: Efficient Inference of Mixture of Binary Experts under Post-Training Quantization
MoBiE: weight binarization framework for Mixture-of-Experts LLMs addressing cross-expert redundancy and routing shifts for efficient inference.
4/22/2026
Semantic Intent Fragmentation: A Single-Shot Compositional Attack on Multi-Agent AI Pipelines
Semantic Intent Fragmentation: attack class against multi-agent LLM orchestration systems where composed subtasks violate security despite individual safety.
4/22/2026
THEIA: Learning Complete Kleene Three-Valued Logic in a Pure-Neural Modular Architecture
THEIA: 2.75M parameter modular neural network learning complete Kleene three-valued logic without symbolic inference or hand-coded gates.
4/22/2026
A Triadic Suffix Tokenization Scheme for Numerical Reasoning
Triadic Suffix Tokenization: deterministic tokenization scheme preserving digit structure in LLMs to improve arithmetic and numerical reasoning accuracy.
4/22/2026
Evaluating Cooperation in LLM Social Groups through Elected Leadership
Multi-agent LLM framework exploring whether elected leadership and governance mechanisms improve cooperation in common-pool resource management.
4/22/2026
Coding-Free and Privacy-Preserving Agentic Framework for Data-Driven Clinical Research
CARIS: AI agent system automating clinical research workflows including literature search, cohort construction, and IRB documentation without coding.
4/22/2026
JumpLoRA: Sparse Adapters for Continual Learning in Large Language Models
JumpLoRA framework for continual learning in LLMs using sparse adapters with jump-based sparse masking to mitigate catastrophic forgetting.
4/22/2026
Beyond the 'Diff': Addressing Agentic Entropy in Agentic Software Development
Addresses agentic entropy in autonomous coding agents, proposing methods to track system-level drift beyond code diffs and maintain alignment with architectural intent.
4/22/2026
Why AI Readiness Is an Organizational Learning Problem, Not a Technology Purchase
Analysis of AI project failures in organizations, finding that only 6% of firms see earnings impact despite $252B investment, attributing issues to organizational learning problems.
4/22/2026
Geometry-Aware CLIP Retrieval via Local Cross-Modal Alignment and Steering
Improves CLIP retrieval by addressing local geometric inconsistencies in cross-modal alignment for better ranking of nearby items.
4/22/2026
SCATR: Simple Calibrated Test-Time Ranking
SCATR method for test-time scaling in LLMs using efficient ranking scorers as alternative to expensive process reward models.
4/22/2026
REZE: Representation Regularization for Domain-adaptive Text Embedding Pre-finetuning
REZE framework for domain-adaptive text embeddings using contrastive pre-finetuning with representation regularization to prevent task-induced bias.
4/22/2026
Cat-DPO: Category-Adaptive Safety Alignment
Category-adaptive safety alignment method for LLMs that addresses safety across different harm categories rather than uniform safety scoring.
4/22/2026
Beyond the Bellman Fixed Point: Geometry and Fast Policy Identification in Value Iteration
Research paper analyzing geometric structure of Q-value iteration for faster policy identification in dynamic programming and Markov decision processes.
4/22/2026
DuQuant++: Fine-grained Rotation Enhances Microscaling FP4 Quantization
Research paper on DuQuant++ fine-grained rotation method improving MXFP4 quantization for efficient LLM inference on NVIDIA hardware.
4/22/2026
LEPO: Latent Reasoning Policy Optimization for Large Language Models
Latent reasoning policy optimization for LLMs with stochastic sampling via Gumbel-Softmax for improved reasoning diversity.
4/22/2026
Diversity Collapse in Multi-Agent LLM Systems: Structural Coupling and Collective Failure in Open-Ended Idea Generation
Study of diversity collapse in multi-agent LLM systems during open-ended idea generation.