Superpower Daily
Scottish Government Stops Short of Datacentre Moratorium as Hun...Microsoft Releases MAI-Transcribe-2 at 10 Cents an HourFigure Signs $3.5B Nscale Deal for Up to 100,000 GPUs for Human...HUMAIN Releases Arabic AI Preview Built on MiniMax’s M3 LineagexAI Explains How Grok Bot Is Built to Keep Agents Working Betwe...Cohere Publishes Agent-Tool Dataset, Finds 419 Occupations Have...Scottish Government Stops Short of Datacentre Moratorium as Hun...Microsoft Releases MAI-Transcribe-2 at 10 Cents an HourFigure Signs $3.5B Nscale Deal for Up to 100,000 GPUs for Human...HUMAIN Releases Arabic AI Preview Built on MiniMax’s M3 LineagexAI Explains How Grok Bot Is Built to Keep Agents Working Betwe...Cohere Publishes Agent-Tool Dataset, Finds 419 Occupations Have...

Full archive

Launch: Latest AI News

Search reporting on AI markets, models, products, companies, policy, and culture.

Category

More filters
LaunchClear all
Showing 1-13 of 13Results matching "Agent harness"

Yesterday

1 story
LaunchTools

Google Launches Fairwind to Give Selected Defenders AI Tools for Finding and Fixing Flaws

The program draws a line between broadly available code-security tooling and restricted access to Gemini 3.8 Flash Cyber, reflecting the dual-use risk of AI that can identify and repair software weaknesses. Google has launched the Fairwind Program, a limited-access channel for governments and trusted partners to use AI systems that can find, verify and fix software vulnerabilities.

3 min readRead story

Earlier this week

2 stories
LaunchProducts

CrowdStrike’s SafeMind Pits Red and Blue AI Agents in an NVIDIA Infrastructure Twin

The new Falcon system combines CrowdStrike’s security data and agent harnesses with NVIDIA’s open Nemotron models. Early tests favor the optimized open pipeline on detection quality, but cover only one attack scenario family. CrowdStrike has introduced SafeMind inside its Falcon platform, pairing its cybersecurity models and agent harnesses with NVIDIA Nemotron open models.

3 min readRead story
LaunchProducts

Modellix Adds 28+ Models, Including $0 Options, to DeepSeek Harness Beta

The beta gives users of the rapidly adopted coding agent one model-selection layer and web-tool routing, while its free-call pitch still requires a Modellix account and funded balance to create an API key. Aurora Mobile’s Modellix has released dsh-modellix, a beta plugin that brings its LLM gateway into DeepSeek Harness ’s native model selector. The plugin’s LLM, Design and Web functions can be toggled independently.

3 min readRead story

Last week

5 stories
LaunchProducts

Harness Launches Agent-Ready Repository and AI Review for High-Volume Code Changes

The products pair high-volume repository operations with inherited permissions and merge-time checks, but Harness still places the final decision to ship with people. Harness has launched an agent-ready code repository and AI code-review product built for a delivery problem that emerges after code generation: agents can create changes faster than conventional pull-request workflows can absorb.

2 min readRead story
LaunchInfrastructure

AWS Publishes Two AgentCore Paths to Query Cross-Account Knowledge Bases

The useful constraint is not simply access across AWS accounts: generated answers require an assumed role, while many workloads may not need an agent at all. AWS has published two generally available AgentCore patterns for agents that need answers from a Bedrock Knowledge Base held in a separate AWS account. In AWS’s example flow, a user submits a natural-language question through an API or Streamlit interface.

3 min readRead story
LaunchModels

DeepSeek V4 Pro Lands on Fireworks With $2.50 CyberGym Solves, but Kimi K3 Scores Higher

The provider’s results frame security-agent selection as a three-way choice among raw solve rate, cost per completed task, and whether a model will reliably execute the required tool workflow. Fireworks AI has made DeepSeek-V4-Pro-0813 available on its platform, pairing the release with security-benchmark results that favor the model on cost and tool reliability rather than outright solve rate.

3 min readRead story
LaunchProducts

Perplexity’s Portable Computer Runs Agents Locally—but Needs a 24GB Nvidia GPU

The product turns a difficult DIY local-AI setup into one application, but its hardware floor and Linux-only launch keep it aimed at well-equipped paid users. Perplexity has launched Portable Computer, a local-first edition of its Computer agent platform for Linux machines with compatible Nvidia hardware. The handoff is a product control, not an automatic fallback Portable Computer starts tasks on-device by default.

3 min readRead story
LaunchProducts

OpenAI’s $20 ChatGPT Work Brings Codex Agents to Office Apps, With Access Still a Barrier

The product’s promise is a ChatGPT that acts across workplace systems rather than merely answering prompts. Its immediate test is whether non-engineers will grant access and navigate controls built for more autonomous software. ChatGPT Work asks office workers to give an AI system access to the software that holds their schedules, messages, documents, and customer records.

3 min readRead story

August 2026

5 stories
LaunchProducts

Databricks Says Its New Extraction Mode Beats Frontier Models on the Documents That Break Them

Precision Mode is designed for cross-page reasoning, enormous line-item outputs and deeply nested schemas. Databricks reports a seven-point lead over its best tested chunk-and-merge baseline, but the evaluation blends internal and public datasets. Databricks is promising a more reliable way to turn the documents that overwhelm ordinary AI extraction into structured data.

3 min readRead story