Category intelligence

AI News Briefing — March 16, 2026

6 current items analyzed and ranked.

Executive synthesis

AI News Summary

AI agent infrastructure dominated this cycle, with LangChain releasing Deep Agents, a structured runtime for multi-step planning and memory management, and Volcengine open-sourcing OpenViking, a filesystem-based context database for agent memory.

Zhipu AI and Tsinghua University released GLM-OCR, a compact 0.9B-parameter multimodal model for document parsing and structured extraction
The Guardian published a major investigation into AI companies functioning as defense contractors, calling for regulation of AI warfare systems used in Gaza and Iran
Multiple tutorials covered enterprise AI governance with OpenClaw and structured LLM output with Outlines and Pydantic, reflecting growing focus on production reliability

Key Themes

AI Agent Infrastructure · 3AI Ethics & Military AI · 1Efficient Multimodal Models · 1AI Governance & Reliability · 2

Primary evidence

Top Ranked Signals

News MarkTechPost Mar 15

LangChain Releases Deep Agents: A Structured Runtime for Planning, Memory, and Context Isolation in Multi-Step AI Agents

By Michal Sutter

68 score

AI Analysis

LangChain released Deep Agents, a structured runtime library built on LangGraph for multi-step AI agents with planning, memory management, context isolation, and human-in-the-loop workflows. It packages defaults and built-in tools around the standard tool-calling loop rather than introducing a new reasoning model.

Most LLM agents work well for short tool-calling loops but start to break down when the task becomes multi-step, stateful, and artifact-heavy. LangChain’s Deep Agents is designed for that gap. The project is described by LangChain as an ‘agent harness‘: a standalone library built on top of LangChain’s agent building blocks and powered by the LangGraph runtime for durable execution, streaming, and human-in-the-loop workflows. The important point is that Deep Agents does not introd

AI AgentsOpen SourceDeveloper ToolsAgent Infrastructure

News AI (artificial intelligence) | The Guardian Mar 15

These aren’t AI firms, they’re defense contractors. We can’t let them hide behind their models

By Avner Gvaryahu

62 score

AI Analysis

Opinion/investigative piece arguing that AI companies acting as defense contractors should be regulated as such, citing AI-enabled warfare patterns in Gaza and Iran. Draws parallels between Israel's 'fog procedure' and systematized AI targeting systems.

From Gaza to Iran, the pattern is the same: precision weapons, chosen blindness, and dead children. The cost of failing to regulate AI warfare is already too highThere is an Israeli military strategy called the “fog procedure”. First used during the second intifada, it’s an unofficial rule that requires soldiers guarding military posts in conditions of low visibility to shoot bursts of gunfire into the darkness, on the theory that an invisible threat might be lurking.It’s violence licensed by bl

AI EthicsMilitary AIAI RegulationDefense

News MarkTechPost Mar 15

Zhipu AI Introduces GLM-OCR: A 0.9B Multimodal OCR Model for Document Parsing and Key Information Extraction (KIE)

By Asif Razzaq

60 score

AI Analysis

Zhipu AI and Tsinghua University released GLM-OCR, a compact 0.9B-parameter multimodal model for document parsing combining a 0.4B visual encoder and 0.5B language decoder. It targets practical document OCR including tables, formulas, and structured extraction at low computational cost.

Why Document OCR Still Remains a Hard Engineering Problem? What does it take to make OCR useful for real documents instead of clean demo images? And can a compact multimodal model handle parsing, tables, formulas, and structured extraction without turning inference into a resource bonfire? That is the problem targeted by GLM-OCR, introduced by researchers from Zhipu AI and Tsinghua University. The research team presents GLM-OCR as a 0.9B-parameter compact multimodal model for document underst

Model ReleaseMultimodal AIOCREfficient AI

News MarkTechPost Mar 15

Meet OpenViking: An Open-Source Context Database that Brings Filesystem-Based Memory and Retrieval to AI Agent Systems like OpenClaw

By Asif Razzaq

55 score

AI Analysis

OpenViking is an open-source context database from Volcengine that organizes AI agent memory through a virtual filesystem paradigm rather than flat text chunks. It addresses fragmented context, poor retrieval quality in RAG pipelines, and limited memory iteration in long-running agent tasks.

OpenViking is an open-source Context Database for AI Agents from Volcengine. The project is built around a simple architectural concept: agent systems should not treat context as a flat collection of text chunks. Instead, OpenViking organizes context through a file system paradigm, with the goal of making memory, resources, and skills manageable through a unified hierarchical structure. In the project’s own framing, this is a response to five recurring problems in agent development: fragmented c

AI AgentsOpen SourceRAGContext ManagementAgent Infrastructure

News MarkTechPost Mar 15

A Coding Implementation to Design an Enterprise AI Governance System Using OpenClaw Gateway Policy Engines, Approval Workflows and Auditable Agent Execution

By Asif Razzaq

42 score

AI Analysis

A technical tutorial demonstrating how to build an enterprise AI governance system using OpenClaw, featuring risk classification, approval workflows, and auditable agent execution. Combines policy engines with agent capabilities for safe autonomous AI deployment.

In this tutorial, we build an enterprise-grade AI governance system using OpenClaw and Python. We start by setting up the OpenClaw runtime and launching the OpenClaw Gateway so that our Python environment can interact with a real agent through the OpenClaw API. We then design a governance layer that classifies requests based on risk, enforces approval policies, and routes safe tasks to the OpenClaw agent for execution. By combining OpenClaw’s agent capabilities with policy controls, we demonstra

AI GovernanceEnterprise AITutorialsAI Safety

News MarkTechPost Mar 15

How to Build Type-Safe, Schema-Constrained, and Function-Driven LLM Pipelines Using Outlines and Pydantic

By Asif Razzaq

38 score

AI Analysis

Tutorial on building structured, type-safe LLM pipelines using Outlines and Pydantic, covering schema-constrained generation, JSON recovery, and function-calling patterns. Focuses on production reliability and constraint enforcement.

In this tutorial, we build a workflow using Outlines to generate structured and type-safe outputs from language models. We work with typed constraints like Literal, int, and bool, and design prompt templates using outlines.Template, and enforce strict schema validation with Pydantic models. We also implement robust JSON recovery and a function-calling style that generates validated arguments and executes Python functions safely. Throughout the tutorial, we focus on reliability, constraint enforc

Structured GenerationDeveloper ToolsTutorialsLLM Engineering