luis bitencourt-emilio (@luisatlive) header
luis bitencourt-emilio (@luisatlive) profile picture

svp @ github copilot, board member @ wikipedia. ex: microsoft, reddit, rappi and loft. engineer, product leader, founder, latino and father.

Joined January 2009
Photo by luis bitencourt-emilio (@luisatlive)Photo by luis bitencourt-emilio (@luisatlive)Photo by luis bitencourt-emilio (@luisatlive)Photo by luis bitencourt-emilio (@luisatlive)Photo by luis bitencourt-emilio (@luisatlive)Photo by luis bitencourt-emilio (@luisatlive)Photo by luis bitencourt-emilio (@luisatlive)Photo by luis bitencourt-emilio (@luisatlive)Photo by luis bitencourt-emilio (@luisatlive)Photo by luis bitencourt-emilio (@luisatlive)
a fantastic set of folks I'm humbled to even be mentioned alongside. bora!
Forbes' youngest self-made woman billionaire, Zuckerberg's co-founder and Qualcomm's CEO all have one thing in common: Brazil. They're not the only ones. 16 Brazilians at the top 👇 ▪️ Eduardo Saverin @esaverin, co-founded Facebook, now co-CEO of B Capital ▪️ Mike Krieger @mikeyk, co-founded Instagram, now co-leads Anthropic Labs ▪️ Cristiano Amon @cristianoamon, CEO of Qualcomm ▪️ Luana Lopes Lara @luanalopeslara, co-founded Kalshi, Forbes' youngest self-made woman billionaire ▪️ Cristina Junqueira @junqueira_cris, co-founded Nubank (143M customers) ▪️ Hugo Barra @hbarra, ex-VP of Android at Google, ran Oculus at Meta, now at Meta Superintelligence Labs ▪️ Daniel De Freitas @dan_defr, co-founded Character.AI and created Google's LaMDA, now at Google DeepMind ▪️ Pedro Franceschi @pedroh96, co-founded Brex, sold to Capital One for $5.15B ▪️ Henrique Dubugras @hdubugras, co-founded Brex and was its co-CEO ▪️ Lidiane Jones, CEO of Integral Ad Science, ex-CEO of Slack ▪️ Camilla Matias, COO of Monaco (ex-COO of Brex) ▪️ Cesar Carvalho, co-founded Wellhub (Gympass), valued at $2.2B ▪️ 🎧 Luis Bitencourt-Emilio @luisatlive, SVP of Copilot Agents at GitHub ▪️ 🎧 Rodrigo Schmidt, built the systems that personalize Instagram for 2B users, now co-founder of Vetto AI ▪️ 🎧 Gina Gotthilf @ginag, co-founder & CMO of Outsmart, ex-Duolingo (helped grow it from 3M to 200M users) ▪️ 🎧 Nicolas Silberstein @nickscamara_, co-founder & CTO of Firecrawl (raised a $75M Series B in September) 🎧 = I've interviewed them on Startupeable. Since I moved to SF, I’ve been interviewing and mapping the founders and execs from around the world who made it to the top of tech. That map is Diaspora and Brazil is one of 30+ countries on it. Find your country: startupeable.com/en/diaspora
7
189
luis bitencourt-emilio retweeted
Amazing work from our partners at GitHub and all the security maintainers who worked with us. Read their story: github.blog/open-source/main…
Our national cyber defense will fall behind if we can’t run security agents 24/7.
13
34
2
216
49,786
luis bitencourt-emilio retweeted
At GitHub Universe in October, I talked about multi-model solutions like Copilot Fusion will start dominating benchmarks going forward. But the pace of progress has been faster - ~10% better quality at half cost just few weeks later.
3
7
3
43
3,583
luis bitencourt-emilio retweeted
The founders of Kalshi and Vercel weren't born in the US. Neither were the people behind 59% of America's $1B+ startups. 🌎 Introducing Diaspora: An interactive map of 500+ founders, investors and operators who left home to build world-class companies. It took me years to find these now you can find them in one click. There was no list of them, so we made it public. With 500 more you haven't heard of yet! This map shows where each person was born and where they build today. If you're raising or selling in a new market, people from your country could open doors for you. I believe shared roots build trust faster than any warm intro. Pick your country and find your people. 🌎 Explore Diaspora: startupeable.com/diaspora/
48
32
17
219
57,578
luis bitencourt-emilio retweeted
AI is forcing us to make software be an actual engineering discipline, AI generated code behaves similar to physical matter, engineers need to understand durable principles, know how to measure, and know how to control production. Not just assume code will do “the right thing” because we’ve crafted it ourselves, atom by atom.
Verification Engineers, Proof Engineers, State Engineers... Engineering is turning into knowing when things are done right. Back to the world of formal verification, eval verifiers and less vibes. Were going back to basics.
23
87
14
889
202,769
luis bitencourt-emilio retweeted
In October, we launched new ways to work with Copilot: • Copilot automations for issue triage • Projects in the GitHub Copilot app for work across multiple repos (or without a repo!) • Copilot Fusion for routing to frontier models Find out how our dev team is using them ▶️
49
22
7
250
49,699
luis bitencourt-emilio retweeted
Loops, Graphs, Factories, Chains and I'm a little over it. Chatted with @MeaganCojocar, @acolombiadev and @luisatlive about what all this even means, and how you can use Automations to setup working Factories today in the @GitHub Copilot app. youtu.be/39ys3AdPzHo?si=dgn_…
6
5
48
2,794
luis bitencourt-emilio retweeted
Welcome to Hotel Calif. We're in lockdown mode. You can check in, but you can never leave. Broccoli is served around the clock. The Brotli bug apparently affected iOS and macOS too. Apple released version 27.2.1 yesterday, fixing this bug among other security issues. This is an important security update. If you haven't updated yet, now's a good time. The demos show what we were able to do with it.
97
257
71
2,542
502,807
luis bitencourt-emilio retweeted
A vulnerability (worse_query) affecting iOS 26.7.1, 27.0.1, and 27.2 allows malicious apps to read local files from other apps, putting seed phrases, private keys, and login credentials at risk across the entire ecosystem. This is an iOS-level issue, not unique to Binance. Immediate Action Required: - Delete any apps installed from untrusted sources (such as sideloaded or manually signed IPAs). - Update iOS and all apps to the latest official versions. - If you suspect a breach, transfer funds to a new wallet created on a clean device and reset all active login sessions. Never install unauthorized packages, and keep your software up to date.
Readers added context they thought people might want to know
Binance recommends updating iOS to the latest versions against worse_query, but researchers describe it as an unpatched 0day and Apple's iOS 27.2.1 notes omit any such fix. github.com/forcequitOS/wo… support.apple.com/en-us/149715 x.com/0xjohnny/statu…
811
1,724
1,026
8,931
3,937,308
brain surgery... deep skylar vibes from @heroes
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training. We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵 Blog: rikiyat.github.io/blog/weigh…
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math. rikiyat.github.io/blog/weight-sp… github.com/RikiyaT/lm_und… huggingface.co/Qwen/Qwen3.5-0… huggingface.co/RikiyaT/Qwen3.…
1
127
luis bitencourt-emilio retweeted
We're redefining the SDLC for the agentic era. Lots of GitHub updates this morning at Universe: • GitHub Copilot Fusion delivers frontier-level quality, 40% faster and at up to 77% lower cost • Our unified Copilot harness is now in Foundry, a managed service for building long-running agents that reason, use tools, and get work done • Copilot Projects gives you an orchestration loop to build with multiple agents across repos And we’re expanding capacity and strengthening GitHub’s architecture to make the home of agentic development more resilient.
228
206
49
1,622
291,403
luis bitencourt-emilio retweeted
The advantage of running across multiple model families is structural. Agentic iteration converges faster and is more successful. The question was if you can make prompt caching to work well enough - here is the answer:
GitHub has released Copilot Fusion in GitHub Copilot, which orchestrates between Anthropic and OpenAI models. It scores 2 points below Opus 5.5 in Claude Code at half the Cost per Task in the Coding Agent Index GitHub describes Copilot Fusion as a runtime orchestration system, which selects and coordinates models to optimize performance, cost, and latency for each task. Copilot Fusion chooses between a number of models from Anthropic and OpenAI, including Claude Opus 5.5, Claude Sonnet 5.5, GPT-6 Astra, and GPT-6 Sol. We benchmarked its max setting in GitHub Copilot CLI in the Artificial Analysis Coding Agent Index v2.0. Token costs are charged based on the underlying models used to complete a request. Key takeaways: ➤ Cost-efficient for near-frontier performance: Copilot Fusion in GitHub Copilot costs $5.59 per task, versus $11.42 for Claude Code with Opus 5.5 (max). Devin Fusion with Opus 5.5 (max) + GPT-6 Luna (high) scores 62 at $8.00 per task. ➤ Slight lead in Terminal-Bench 4.0: The Coding Agent Index v2.0 weights its three constituent benchmarks equally. Copilot Fusion in GitHub Copilot leads Terminal-Bench 4.0 with a score of 71. It ranks third on FrontierCode in the chart at 51, behind Claude Code with Opus 5.5 (max) at 56 and Devin Fusion with Opus 5.5 (max) + GPT-6 Luna (high) at 54. Its SWE-Atlas-QnA-AA score lags at 57, between GPT-6 Astra in Codex and Gemini 4 Argon in Antigravity CLI. ➤ Half the token use of Opus 5.5 in Claude Code: GitHub Copilot Fusion used 7.2M tokens per task in the Coding Agent Index, approximately half of the tokens used by Claude Opus 5.5 in Claude Code, and 30% less than Devin Fusion with Opus 5.5 (max) + GPT-6 Luna (high).
4
4
18
1,591
luis bitencourt-emilio retweeted
17.9 billion pushes to GitHub this year. More than 2024 and 2025 combined. And... it’s not slop. 😉 Repos serving mining, construction, health care, and agriculture have each grown several times over since 2022. This year’s report paints a much fuller picture of the times we’re in. Glad to have so many new folks aboard the developer journey. More in this year’s Octoverse Report:
5
10
1
71
6,328
luis bitencourt-emilio retweeted
as bad as things have gotten in the past 4 days, they are going to get a lot worse before they get better. security leaders, never let a crisis go to waste: please go to your boards and ask for help right now. not for ai, for the basics. (we need to hold on for ~18 months.)
The vulnerability tsunami is coming in just a few months and there won’t be enough time to patch critical infrastructure. Network isolation described below (air-gaps and data diodes!) needed now!
454
1,405
765
10,258
4,485,592
Foursquare was acquired last month. The outcome was disappointing. It's taken me a while to write this, but I think I'm finally done grieving it. Here are some thoughts on our final years, an effort to honor the team who helped build & maintain FSQ's legacy, and my attempt to finally get some closure. teendrama.substack.com/p/som…
444
269
324
3,008
738,153
luis bitencourt-emilio retweeted
We still want to find a way to bring the community together. If you planned to be in California for the Grace Hopper Celebration, I invite you to join us this year for GitHub Universe, Oct 28-29. 🌉 We've set aside 75 complimentary passes for folks with cancelled registrations. While this doesn't replace the conference and Universe has far fewer attendees than GHC, my hope is that it does give folks an opportunity to connect with each other. 📷 DM the GitHub team on LinkedIn or email me at [email protected] (first come first serve).
2
25
3
135
31,144
could not agree more. "protocol compatibility" was a key driver of copilot cli adoption internally.
I’m thinking about banning Claude code at Shopify until they change their mind and read AGENTS.md and .agents/skills etc. Insisting on only reading CLAUDE.md sometimes leads to split brain problems when different team members use different tools. Just unnecessary.
3
6
1,033
amen.
We're adding support for AGENTS.md to Claude Code. Starting today in version 2.1.277, if there is no CLAUDE.md in a folder, Claude will check for and use AGENTS.md. You can toggle this behavior in /config.
45
luis bitencourt-emilio retweeted
Given the rate of new model releases and rate of innovation of how different models can be combined on a single task - it is almost impossible to make optimal choices manually. This work is one of key components for our future auto routing which will also incorporate local inference (and immediate 67% cost savings is not bad).
4.9 percentage points higher verified task quality. 67% lower estimated cost. Project HydraFusion delivered those results against Claude Opus 5 on Terminal-Bench 2.1 in controlled offline evaluations. HydraFusion orchestrates the models and workflow for each coding task. You select it like any other model. ⬇️ github.blog/ai-and-ml/github…
1
5
1
71
6,133
luis bitencourt-emilio retweeted
PwC looked into Anthropic's harness primitives (a loop with memory, filesystem, and bash, i.e. the Claude Code shape) and argues that enterprises should stop building a new agent system per use case and instead deploy one identical harness everywhere. This happens mostly because it collapses review, not because it runs better. So they recommend standardize the loop across teams and let the prose in an instructions file carry the difference. If every agent in a company runs the same unmodified harness and differs only in a plain-text instructions file, auditing the whole fleet becomes reading documents instead of reviewing code. – arxiv. org/abs/2608.20622 Title: "Applying Anthropic Primitives at Large Enterprises: Harness Paradigm for Knowledge Work"
31
39
8
318
26,319