Arron Bailiss

Principal Engineer @ AWS

ML/AI, robotics, developer tools, and distributed systems

[TheWitcherish @TheWitcherish]

Let's Strands Decider our Role-Playing Game

Forest, dusk. CRACK. Do you roll the dice? 🎲

Strands Decider's confidence tells my AI Game Master when to act AND when to ask :)

#StrandsAgents #AgenticAI #SoftwareEngineering

https://t.co/I3F2cUnfBc
🟣 https://t.co/43ahOwTnGT

[Marc Brooker @MarcJBrooker]

The Strands Harness team wrote a deeper dive on how we compared our harness to alternatives, running over 94,000 trials, achieving 34% lower cost on average.

Don't miss it: https://t.co/XmUucAWMFD

[morgan @MorgantWillis]

Strands harness was launched a few weeks ago showing 28% lower token cost compared to other leading harnesses with comparable accuracy. The Strands team went on to run 94,000 trials and the updated number actually shows Strands harness gets you 34% lower token cost.

Here's a deep dive on how they got these results, which benchmarks were ran and how, and what it means for choosing an agent harness: https://t.co/Dcl7AYNmuD

[Marc Brooker @MarcJBrooker]

Looking for a great new decider model? Today, we're pleased to share four new strands-decider variants: more accurate, and better calibrated.

[Marc Brooker @MarcJBrooker]

Today we're announcing Strands Box, our open source sandbox for developers building AI agents.

New blog post from me, on the Strands blog: https://t.co/UHjn1XA1bE

[Darryl Ruggles @RDarrylR]

Another day, another release from the Strands team! Strands Box was just released as a developer preview.

Most sandboxes are just about what files can be read and if there is network access. This isn't enough granularity for many uses though. Strands Box uses kernel isolation to make a user-space policy layer the only way out of the isolation. This is enforced with Dogwood rules for shell use, Python, MCP and network access. One example shown is for git push being blocked unless npm test recently passed.

Check this out from @MarcJBrooker. If you haven't already checked out Strands, give it a try.

https://t.co/9srnlDjO7H

#Agents #AgentSecurity #Sandboxing #StrandsAgents #AWS

[morgan @MorgantWillis]

back in july, I put an ai agent in a big mouth billy bass bringing billai to life. some of you asked how (or why). well, now that bidirectional streaming is GA in strands, I wrote up a deep dive on the the whole thing 🐟 https://t.co/zaltI19uVq

[Albert Zhao @_AlbertZhao]

excited to see cloudflare's clef! we also released a decision model built from qwen, except we show both the training data and weights on @huggingface.

strands decider, clef, and other decision models gna be huge for agent latency as they replace some LLM calls

https://t.co/Yw6Em0iTWa

[AWS Developers @awsdevelopers]

AWS just got a lot easier to use.

[AWS Developers @awsdevelopers]

01110100 01101111 01101111 01101100 01110011 00101110 00100000 01101000 01101111 01101111 01101011 01110011 00101110 00100000 01100101 01110110 01100001 01101100 01110011 00101110 00100000 01110011 01110100 01110010 01100101 01100001 01101101 01101001 01101110 01100111 00101110 00100000 01110000 01111001 01110100 01101000 01101111 01101110 00100000 00101011 00100000 01110100 01111001 01110000 01100101 01110011 01100011 01110010 01101001 01110000 01110100 00101110 00100000 01101111 01110011 01110011 00001010 00001010 01110100 01110010 01111001 00100000 01110100 01101111 01100100 01100001 01111001 00100000 11110000 10011111 10010001 10001001 00100000 01100111 01101111 00101110 01100001 01110111 01110011 00101111 00110100 01100111 01010010 01000010 01000100 00110101 01100011

[morgan @MorgantWillis]

Claude Opus 5 scores 30% on ARC-AGI-3 on its own. Wrapped in an agent harness built with Strands, it scored 99.95%. I dug through the open source code to see how it works: https://t.co/L2u84HFoMr