Git Maxd
@GitMaxd
Followers
2K
Following
31K
Media
558
Statuses
3K
Linux Dev | AI Rev Early Adopter | Next | Gen AI\UX | Deep Agents
Phoenix, AZ
Joined July 2023
A great new guide just dropped by @FactoryAI “How to become a (Droid) Power User” In-depth coverage on: 📓Fewer tool calls 📓 Better first attempts 📓 Consistent Quality 📓 Self-Correcting code
4
8
157
Coincidences, I’m sure https://t.co/KVITFCgEmW
@AnthropicAI @FTC @DavidSacks Hey US @WhiteHouse admin, since quite a few of you follow my account, @sriramk @pmarca can you please investigate @DarioAmodei and @AnthropicAI for fradulent behaviour & mental health damage to their users by swapping out what they sell us without notice? (p.s. opus 4 works fine)
0
0
1
@OranAITech @AnthropicAI @claudeai @FactoryAI Skills make you capable •Skills let you execute •Skills compound over time •Skills travel with you everywhere Skills are not Claude Exclusive- it’s a markdown file with progressively disclosed instructions. I would debate that skills are less valuable in @claudeai Code than
1
1
2
Skill issue, I’m sure… Probably doesn’t have any intuition about what is happening. I’d take @TheAhmadOsman’s vibes over just about any benchmark at this point. The house is on fire, @claudeai
anthropic should refund people for this poor performance i am not gonna say opus 4.5 is quantized or nerfed or whatever but you owe people their money back if you're serving buggy software, especially since this is a repeated offense now
1
0
0
What is a “perceived drop” here is not the harness. The model does the exact same thing in Droid. Unless the coordinated to provide a shitty service, im out of this debate. It’s way easier to run GPT Models and know what I am getting in the morning.
0
0
1
I’m committing to two UI’s 1. Chat 2. TUI Ok 3, Also OP’s UI they like for whatever smart reason Why? I have seen this before and you are almost certainly wrong if you think you’re reinventing the birth of AI UI. Familiarity > Novelty I don’t want to learn a new UX, if you
0
0
1
I wonder why @AnthropicAI @claudeai killed Opus? No I don’t rly - I just switched to GPT 5.2. It’s the players, not the game. The game is fine @claudeai “We don’t nerf” - that leaves scaling failure worth scolding for 2 bills a month - so… Opus is 1/2 half the model it
I really don’t want to leave @claudeai Claude Code is fully broke. Nothing from @AnthropicAI to address concerns growing like fire? @alexalbert__ are you all seeing this? Not a brush fire anymore.
2
0
3
I can’t stop looking at this image This stat has intrigued me the most and it’s something I felt in my soul Is it the Model Quality? Is it the agent harness? Or, is it the Context? My feels are that we are just getting started with Context Engineering; We can solve Quality
Quality remains the biggest barrier to production. Making sure agent outputs are reliable is still the hardest part. Factors like cost were cited less frequently as blockers than in previous years, with more orgs focused on making agents work well and fast.
0
1
7
It’s out! @LangChainAI’s 🦜 “State of Agent Engineering” survey results are now available and it’s packed full of goodness provided straight from the Community More than half of respondents now have agents in Prod 🤯 A giant leap from last year
We surveyed 1,300+ developers and business leaders to understand how AI agents are actually being built, tested, and shipped to real users. What are the most popular agent use cases? Biggest challenges when building agents? And who's really using observability or evals tooling?
0
4
24
Such a fun night at the @claudeai Phoenix Meetup tonight! I met some some very talented ASU students & @LangChainAI Engineers, especially @panguin6010 🔥 thanks for the great chat! There is no shortage of AI enthusiasm, or LangChain, in PHX! ☀️
2
1
6
LangSmith Agent Builder, built on the @LangChainAI Deep Agents library, is now generally available in public beta! This is a super neat way to- ✅ Build Agents with tools ✅ Give Agents MCP access ✅ Remotely access the agent Great project - Save this & give it a try 🚀
🚀 LangSmith Agent Builder in Public Beta Agent Builder (now available for every LangSmith user) opens the gates to everyone to build agents without code Video on why no-code agents can open up agent building and how I built a recruiting agent with it: https://t.co/707bTgb7M4
0
3
11
These are always great - I wouldn’t miss one Tons of valuable, real-time insight directly from the Dev leaders at @LangChainAI talking about the latest AI tech Share your feedback on LangChain v1, Deep Agents, any pain points and soak up the valuable info
💬 Join our Community Jam Session on Dec 15 (Dec 16 for some APAC zones)! Share your feedback on LangChain 1.0 & 1.1, what’s working, what’s not, and what you love. Your input shapes LangChain’s future! 🚀 👉 RSVP: https://t.co/Uuu6z6m0FQ
0
0
0
Want to learn about these CLI’s everyone’s talking about? @LangChainAI’s OSS Deep Agents CLI is a fun way to start Built on LangChain’s OSS Deep Agents library, the CLI will introduce you to subagents, task lists, memory, *skills* + more uv pip install deepagents-cli
✒️Skills in DeepAgents CLI We recently added support for skills in DeepAgents CLI Skills let you define prompts/tools that you load on command Check out @RLanceMartin breaking them down here: https://t.co/NxCBVfZReG And try out the CLI here: https://t.co/RNZYzGPFUU
1
3
10
We all leveled up today @FactoryAI 🤝 @claudeai Opus 4.5 Droid and Opus 4.5 locked in
3
1
48
Very exciting news! Deep Agents demand excellent instruction following These sorts of early results are promising. That’s not a little beat 🔥
Ran our internal agentic evals on Opus 4.5 and it did not disappoint! These prompts are tuned for Sonnet 4.5, but Opus still managed to beat it out: 🥇Opus 4.5 - 91.5% 🥈Sonnet 4.5 - 86.2% 🥉Gemini 3.0 Pro - 83.6% GPT 5 - 67.8% Our evals look at the model's ability to extract &
0
0
2
Useful learnings: 1. Deep Agents Blog post: https://t.co/x4gChAWKfm 2. @LangChainAI Deep Agents Docs: https://t.co/ssqfgMX3Is 3. LangGraph Starter Template: https://t.co/esCQtnJosP 4. Deep Agents Repo: https://t.co/aPoceLWY6e 4. Deep Agents UI: https://t.co/lf1RL9F2xE OSS?
1
0
2
The prompt in this demonstration was "Create a brief lesson explaining what subagents are in LangGraph DeepAgents framework" Here is the Final Result: https://t.co/mrFE6uWASI
1
0
1
A @LangChainAI Deep Agent to learn stuff Subagents - intention-analyst - Extracts learning requirements - outline-planner - Creates lesson structure - content-synthesizer - Writes complete lesson - quiz-maker - Generates assessments - parity-auditor - Fact-checks
1
0
6