At this point, this file got shared and another AI agent took over and created the feature and deployed it to production. In that process a QA agent was also involved. You might have a wildly different idea or need, I’m certain that Perplexity Computer can solve that for you.
AGENTS
-
Perplexity Computer Agent Completes Tasks in Under Ten Minutes
By
–
Now you just wait, it can take 1min or 10min depends on the task. When Perplexity Computer is done, it will look something like this.
-
Production Intelligence as Competitive Edge in Manufacturing
By
–
Production intelligence is becoming a competitive edge. What is your read on this? @IIoT_World @CRudinschi @agentic_factory @MasterofIoT @survivingwithan @joannefriedman
-
Free Webinar: AI Evolution from Machine Learning to AI Agents
By
–
Free 90-minute webinar: https://
luma.com/gligxe4f hosted by @PacktPublishing @PacktDataML on April 4 The AI Evolution: A Foundational Guide from ML to AI Agents — Playbook for Applying Machine Learning and Agents -
Claw Beta Improves Reliability and Task System for AI Agents
By
–
New Claw beta bits are up! Lots of reliablility+security improvements in there + a new task system for more reliable subagents/crons/etc
-
GAAMA: Graph-Augmented Memory for Long-Term Agent Learning
By
–
// Graph Augmented Associative Memory for Agents // Long-term memory for agents is still an unsolved problem. Flat RAG loses structural relationships, and knowledge graphs miss conversational associations. New research proposes combining both through a hierarchical approach. GAAMA is a graph-augmented associative memory that constructs a concept-mediated hierarchical knowledge graph through episode preservation, LLM-based fact extraction, and higher-order reflection synthesis. It uses four node types connected by five edge types, with retrieval combining semantic search and graph-traversal ranking. On the LoCoMo-10 benchmark, GAAMA achieves 78.9% mean reward, outperforming HippoRAG and tuned RAG baselines. Multi-session agents need memory that captures both facts and their relationships across conversations. GAAMA demonstrates that graph-augmented retrieval consistently beats semantic-only methods, and that higher-order reflections, not just raw fact storage, are key to reliable recall. Paper: arxiv.org/abs/2603.27910 Learn to build effective AI agents in our academy: academy.dair.ai/
-
Codex Desktop App Discovery Significantly Improves Development Workflow
By
–
You can see very clearly in my Codex usage where I discovered the Desktop App (and determined it was a better workflow than the CLI). That's a pretty hard cutover. Romain Huet (@romainhuet) Introducing the Codex app, one place to build with agents. We can’t wait to see what you build with it. — https://nitter.net/romainhuet/status/2018387622809522579#m
→ View original post on X — @romainhuet, 2026-03-31 19:42 UTC
-
Coding Agents Automate Cross-Platform Workflows with Intelligence
By
–
Agents that can code will equally be able to use tools exceedingly well. This allows you to start to automate tasks across a workflow that requires both a component of non-deterministic intelligence but also deterministic system interaction.
— Aaron Levie (@levie) 31 mars 2026
An example would be using something… https://t.co/ziNaeIUR8TAgents that can code will equally be able to use tools exceedingly well. This allows you to start to automate tasks across a workflow that requires both a component of non-deterministic intelligence but also deterministic system interaction. An example would be using something like Codex to automate a workflow connecting data from Box to multiple other systems. The coding agent can interact with systems like an engineer via CLI/MCP/APIs, or write code on the fly when new problems are encountered in a workflow. This will also be one of the reasons you’ll see more technical and engineering roles start to help automate work in non-engineering domains. Marketing, finance, supply chain, pharma research, and other areas where there’s a large amount of data and systems to talk to all have these properties. Box (@Box) Codex just turned an upcoming meeting into a fully automated cross-platform workflow. Box. Gmail. Slack. It researches across all three, synthesizes the context, and delivers a pre-meeting brief without anyone lifting a finger. This is what personal productivity looks like when agentic automation does the work for you. See it in action.👇 — https://nitter.net/Box/status/2039056257282449696#m
→ View original post on X — @romainhuet, 2026-03-31 19:39 UTC
-
Terminal-Bench 2 Scores Jump to 75-80% in 4 Months
By
–
Top scores on Terminal-Bench 2 went from ~25% → 75-80% in just 4 months.
— Snorkel AI (@SnorkelAI) 31 mars 2026
For Benchtalks #1, @vincentsunnchen sat down with @alexgshaw to dig into what happens when your benchmark gets solved before you're ready for the next one.
Key takes:
→ The terminal is the right… pic.twitter.com/XkWlJT8SKTTop scores on Terminal-Bench 2 went from ~25% → 75-80% in just 4 months. For Benchtalks #1, @vincentsunnchen sat down with @alexgshaw to dig into what happens when your benchmark gets solved before you're ready for the next one. Key takes: → The terminal is the right abstraction for agentic AI → Harbor exists because benchmarking and RL at scale are infra problems → "Benchmaxxing" is real; the defense is shipping harder tasks faster → TB3 is coming, and they want your hardest unsolvable problems "We need 1000x more benchmarks than we have right now" — @alexgshaw
→ View original post on X — @snorkelai, 2026-03-31 19:21 UTC
-
Claude’s rapid evolution toward autonomous agent capabilities
By
–
They've released a number of features already heavily hinting on nudging Code into Claw directions, i.e. it's a speedrun of