In partnership with
TL;DR
Anthropic signed a SpaceX compute deal and doubled Claude Code rate limits
GPT-5.5 Instant is now ChatGPT's default, with half the hallucinations
CopilotKit raised $27M to put AI agents inside your app UI
Google is secretly testing "Remy," a personal AI agent for Gemini
Vercel open-sourced an AI security scanner that runs on your own infra
BoldSign's migration skill converts your DocuSign integration without breaking anything
AI agents can call & pay for APIs with no account or API key required
Attio is the AI CRM for high-growth teams.
Connect your email, calls, product data and more, and Attio instantly builds your CRM with enriched data and complete context. Whether you’re running product-led growth or enterprise sales, Attio adapts to your unique GTM motion.
Then Ask Attio to plan your next move.
Run deep web research on prospects. Update your pipeline as you work. Find customers and draft outreach emails. Powered by Universal Context, Attio's intelligence layer, Attio searches, updates, and creates across your data to accelerate your workflow.
Ask more from your CRM.
ANTHROPIC IS RENTING ELON'S SUPERCOMPUTER.
@dbonneville on X
Anthropic signed a compute deal with SpaceX, giving them full access to the Colossus 1 supercomputer in Memphis. That's 300+ megawatts of additional capacity coming online within the month. Over 220,000 NVIDIA GPUs, essentially overnight.
The timing is not subtle. OpenAI and xAI are direct competitors. Grok and ChatGPT are fighting for the same users. And yet Anthropic just handed Elon's infrastructure team a contract while Sam Altman watches from across the street.
The practical impact hit immediately. Claude Code rate limits doubled for Pro, Max, and Team plans. Peak-hour throttling is gone. API limits for Opus models jumped significantly. If you've been hitting walls building with Claude, you'll feel this.
Zooming out, Anthropic already has deals with Amazon, Google, Microsoft, and NVIDIA. They're running on basically every major compute stack simultaneously. The SpaceX deal just added rockets to that list.
There's a SpaceX clause worth noting too. They can reclaim capacity if needed. So Anthropic doesn't fully own this relationship. They're renting Elon's supercomputer while both sides quietly discuss orbital AI data centers.
Strange times.
OUR TAKE
Anthropic didn't pick this partner for ideological alignment. They picked it because they needed compute and Colossus 1 was available.
That's actually the most honest thing about this deal. In the infrastructure arms race, principles take a back seat to megawatts.
Builders benefit either way. Doubled rate limits are real. Ship something 🫡
QUICK HITS
GPT-5.5 Instant is now ChatGPT's default model: OpenAI's latest daily-driver model cuts hallucinations by 52%, gives shorter answers, and uses your memory better. Rolling out to everyone now.
CopilotKit raises $27M to put AI agents inside your app: Their AG-UI protocol lets agents respond with interactive UI components instead of walls of text. Google, Microsoft, and Amazon are already on board.
Google is secretly building "Remy," a personal AI agent for Gemini: Internally tested by employees, it monitors things that matter to you, handles complex tasks proactively, and learns your preferences over time. Think OpenClaw but deeply wired into every Google product you already use.
Vercel open-sourced an AI security scanner that finds bugs your team missed: DeepSec runs coding agents across your entire codebase, traces data flows, and exports findings as ready-to-fix tickets. Works with your existing Claude or Codex setup. False positive rate sits around 10-20%, which is good for this category.
Are you tracking agent views on your docs?
AI agents already outnumber human visitors to your docs — now you can track them.
YOUR MIGRATION SHOULDN'T BREAK EVERYTHING ELSE
Move your signing workflows to BoldSign using an AI migration skill that handles the messy parts for you.
⏱️ 30–60 min | 🔧 Needs: AI coding agent (Claude Code, Cursor, Copilot), existing DocuSign integration
Why bother? DocuSign's pricing is manageable when you're small. Once you're automating at scale, costs get hard to forecast fast. BoldSign does everything DocuSign does (e-signatures, embedded signing, webhooks, templates, bulk send) with SDKs for .NET, Node, Python, Java, and PHP. The switch is worth it. The tricky part is doing it without quietly breaking the logic that everything downstream depends on.
Steps:
Clone the BoldSign migration skill from GitHub and drop the skill folder into your agent's skills directory. This is a context file that tells your AI agent exactly how BoldSign works before it writes a single line of code.
Open your agent and paste in your existing DocuSign code with this prompt: "Using the skill, convert this to BoldSign while preserving existing functionality." The agent reads the skill and outputs an informed translation, not a guess.
Start with one document send flow only. Test it end to end in sandbox. Field placement, signer access, the full path. Do not move on until this works.
Set up your webhook handler with HMAC SHA256 verification. BoldSign is event-driven, not polling-based. Confirm every lifecycle event is firing correctly: sent, viewed, signed, completed, declined, expired.
Migrate embedded signing, then templates, then everything else. Sequential signing, reminders, revocation, and signer auth all come last once the foundation is solid.
What you end up with: Your downstream systems keep working exactly as before. Same events, same order, no changes on their end. The only thing that changes is the platform underneath.
Full tutorial:
TOOL OF THE DAY
Let your AI agents pay for APIs on their own.
Right now, most APIs still require a human to set up an account, grab a key, and manage billing. That's a bottleneck when you're building autonomous agents. Pay.sh fixes it. One line of code, no sign-up, no account, no subscription. Your agent finds the API it needs, pays for exactly what it uses, and keeps moving.
Works with Claude and Codex. Backed by Google Cloud.
