Anthropic IPO, Microsoft's own AI, Meta Agents in WhatsApp | Weekly Digest
PLUS HOT AI Tools & Tutorials
Anthropic filed its S-1 on June 1 at a $965B valuation — beating OpenAI to the SEC. Microsoft launched 7 in-house MAI models at Build 2026 and said it wants to be a top-4 AI lab. Meta deployed Business Agent globally — WhatsApp, Instagram, and Messenger are now autonomous sales platforms for 1M+ businesses.
Hey! Welcome to the latest Creators’ AI Edition.
Anthropic filed its IPO paperwork with the SEC on June 1 — the first major AI lab to formally begin the process toward a public listing, at a near-trillion-dollar valuation that now puts it ahead of OpenAI. Microsoft dropped 7 in-house AI models at Build 2026 and told the world it no longer wants to be just “the company that resells OpenAI access.” Then Meta launched Business Agent globally on June 3, turning WhatsApp, Instagram, and Messenger into an autonomous sales, support, and operations layer for over one million businesses overnight. Today we have:
Featured Materials 🎟️
News of the week 🌍
Useful tools ⚒️
Weekly Guides 📕
AI Meme of the Week 🤡
AI Tweet of the Week 🐦
(Bonus) Materials 🎁
Keep your mailbox updated with practical knowledge & key news from the AI industry!
Featured Materials 🎟️
Anthropic Files for IPO — and OpenAI Is Probably Next 💰
On June 1, Anthropic confidentially submitted a draft S-1 registration statement to the SEC. This is the formal first step toward a public offering — not a locked-in date, not a share price, not a ticker. But it is the moment the company stopped being a private bet and started becoming a public one.
The mechanics matter: a confidential S-1 lets Anthropic complete the regulatory review process without publicly disclosing financials until 15 days before a roadshow begins. The window between now and the October 2026 targeted listing is roughly four months — long enough for market conditions to change, not long enough for the business to look materially different.
What’s already public from the Series H (May 28):
$65B raised at $965B post-money — the largest funding round in AI company history
ARR exceeds $47B, up 5× in roughly five months
Claude Code run-rate revenue over $2.5B annually
4% of all GitHub public commits worldwide authored by Claude Code
The OpenAI question:
Anthropic beat OpenAI to the SEC. OpenAI is reportedly preparing its own confidential S-1 with Goldman Sachs and Morgan Stanley on standby for a possible fall listing. The race is now explicitly about who gets the better IPO window — not just who has the better model.
What this changes for the industry:
Anthropic preparing a public disclosure means every number it has been hinting at — revenue trajectory, token usage, Claude Code penetration, enterprise share — becomes audited and legally binding within months. For the first time, the AI industry will have a public company whose core product is a frontier model. The metrics that matter for valuation will become metrics that matter for stock price. That changes how every AI lab operates.
Anthropic just decided to let public markets price the question that private investors have been debating for three years: is frontier AI a real business? The S-1 will answer it. OpenAI won’t be far behind.
Source: Anthropic
Microsoft Stops Renting Intelligence 🧠
For three years, Microsoft’s AI story was “we have OpenAI.” At Build 2026 on June 2–3 in San Francisco, that story ended. Microsoft announced a family of seven in-house AI models under the MAI brand — built from scratch by its AI Superintelligence team, not fine-tuned or distilled from existing architectures. The flagship, MAI-Thinking-1, is Microsoft’s first reasoning model.
The MAI model family:
MAI-Thinking-1 — 35B active parameters, 256K context, first Microsoft-built reasoning model. Independent raters preferred it over Claude Sonnet 4.6 in blind tests. Matches Claude Opus 4.6 on SWE Bench Pro. 10× more token-efficient than GPT-5.5.
MAI-Code-1 — coding model built specifically for GitHub and VS Code, already live in Copilot
MAI-Image-2.5 — text-to-image and image-to-image, #3 on Arena AI leaderboard
MAI-Transcribe-1.5 — 43 languages
MAI-Voice-2 — 15+ new languages with voice cloning protection
The honest position:
CEO Satya Nadella said Microsoft’s goal is to become one of the top four AI labs globally — alongside Google DeepMind, OpenAI, and Anthropic. Six months ago, that sentence would have been a stretch. The MAI line, combined with Microsoft’s compute infrastructure and GitHub’s 150M developers, gives it a plausible path. Mustafa Suleyman said the goal is to build “from the ground up” — not to depend on any single external provider.
Also at Build:
Microsoft Scout — a new personal agent for enterprise that runs on OpenClaw and WorkIQ, proactively handles meeting prep, scheduling conflicts, and routine tasks. Majorana 2 quantum processor — qubits 1,000× more reliable than its predecessor, 2029 now the target for a scalable quantum computer.
Microsoft spent years distributing OpenAI’s intelligence. Now it wants to build its own. The MAI line is not a threat to OpenAI today — it’s an insurance policy that becomes a threat the moment OpenAI raises its prices again.
Source: Microsoft Blog
Meta Turns WhatsApp Into Your Sales Team 🤖
On June 3, at Meta’s Conversations event in London, Mark Zuckerberg launched Meta Business Agent globally. Any business — any size, any country — can now deploy an AI agent across WhatsApp, Instagram, and Messenger in minutes.
What the agent does:
Answer customer questions in the customer’s local language. Recommend products from a catalog. Book appointments. Qualify sales leads. Close transactions. Escalate to a human employee at a threshold the business defines. The agent runs 24/7. No IT team, no developer, no custom integration required.
The scale behind it:
Meta spent nearly two years testing Business Agent in India, Mexico, and Brazil before the global launch. Over one million businesses were actively using early versions before June 3. The global rollout extends the product to Instagram for the first time — critical for creator-led commerce and DTC brands. Over one billion messages are sent between people and businesses on Meta’s platforms every day.
The enterprise layer:
Alongside the consumer launch, Meta introduced Meta Business Agent Platform — designed for larger organizations that need to build, customize, and deploy agents at scale. Connects to Shopify, Zendesk, and hundreds of third-party systems. The platform allows agents to take actions inside connected systems, not just answer questions in chat.
Pricing:
Free to start. Paid subscription tiers coming. Enterprise platform pricing not yet disclosed. Token-based billing expected for large-volume deployments.
The creator angle:
For any creator or small business already using Instagram DMs or WhatsApp as a sales channel — this is a free autonomous agent that closes sales while you sleep. The commerce layer that Shopify and Intercom have been trying to build separately just became native to the platforms where your customers already are.
Meta just put an AI sales team inside the apps 3 billion people use to message each other. The barrier between “I saw your product” and “I bought it” just got a lot smaller.
Source: Meta
Share this post with friends, especially those interested in AI!
News of the week 🌍
ChatGPT Hits 1 Billion Users — Claude Is Growing 10× Faster 📈 — Sensor Tower confirmed ChatGPT crossed 1 billion global monthly active app users in May 2026 — the fastest any app in history has reached the milestone, ahead of TikTok, Instagram, YouTube, and Google Maps. The headlining number belongs to OpenAI. The more interesting number belongs to Anthropic: Claude hit 56 million MAUs at 640% year-over-year growth, versus ChatGPT’s 62%. US users who installed Claude in Q1 2026 spent 5% less time on ChatGPT the following month — the first clean evidence of substitution rather than additive usage.
OpenAI Turns Codex Into an Enterprise Platform — Non-Developers Growing 3× Faster 💻 — On June 2, OpenAI expanded Codex beyond coding with Codex Sites (hosted web apps built and deployed from a prompt), 62 business app plugins (Salesforce, Jira, Notion, GitHub and more), and 110 pre-built skills. 5M weekly users, 20% of whom are non-developers — and that segment is growing 3× faster than developers. OpenAI is repositioning Codex from a coding tool into an enterprise automation platform that competes directly with Zapier, Make, and ServiceNow.
GitHub Copilot Switches to Token Billing on June 1 — Bills Jump 10×–50× for Power Users 💸 — GitHub retired flat-rate Premium Request Units on June 1, replacing them with GitHub AI Credits — token-based billing at published API rates per model. Base subscription prices are unchanged: Pro at $10/month, Pro+ at $39. But developers running agentic sessions against large codebases are reporting cost increases of 10×–50×. One Reddit user projected a jump from $29/month to $750. Code completions and Next Edit Suggestions remain free — the billing change only affects chat, agent mode, and code review. The backlash landed within hours of launch.
SoftBank Commits €75B to French AI Data Centers — Europe’s Largest-Ever AI Infrastructure Deal 🌍 — At the Choose France summit on June 1, SoftBank announced up to €75B ($87.5B) to build 5 gigawatts of AI data center capacity in France. The first phase commits €45B for 3.1 GW in the Hauts-de-France region by 2031. SoftBank cited France’s low-carbon electricity grid and engineering talent. This is the largest single AI infrastructure investment ever announced in Europe — and a direct signal that the hyperscaler buildout is no longer US-only.
OpenAI Is Preparing Its Own Confidential IPO Filing 📋 — Anthropic’s S-1 filing on June 1 immediately triggered reporting that OpenAI is preparing its own confidential submission in the coming weeks, working with Goldman Sachs and Morgan Stanley. If both companies list in October-November 2026, it will be the largest simultaneous AI company IPO event in history. The combined implied market cap would exceed $2 trillion before either company has filed audited public financials.
Trump Signs Executive Order on AI Innovation and Security — June 2 🏛️ — On June 2, President Trump signed a new AI Executive Order directing federal agencies to accelerate AI adoption and remove regulatory barriers to deployment. The order expands access to government-held datasets for AI training, streamlines procurement for AI tools in national security contexts, and formally designates the Department of Commerce as the lead agency for AI standards coordination. The administration explicitly frames the move as a counter to what it calls the EU's "stifling" regulatory approach.
Useful tools ⚒️
⭐ Mina Meeting Assistant — The AI teammate that speaks during your calls, not just after them. Mina joins Zoom, Google Meet, or Teams, pulls context from your connected tools in real time, and answers questions, books follow-ups, and updates your CRM while the meeting is still happening. Works with Skills you define — sales call mode, standup mode, customer success mode. Unlike passive note-takers, Mina can be interrupted, redirected, and prompted mid-call.
SocialEcho 2.0 — AI social media copilot for teams and agents. Manages multiple accounts, drafts posts in your brand voice, schedules and monitors performance, and routes agent workflows across LinkedIn, X, Instagram, and TikTok. Built for teams that run more than two channels simultaneously and are tired of switching between tools. Free trial available.
Databox MCP — Chat with your business data inside Claude, ChatGPT, or any MCP-compatible tool. Databox connects your metrics — from Google Analytics, HubSpot, Shopify, Salesforce, and 100+ sources — and makes them queryable in plain English from your AI interface. Ask “how did our paid acquisition perform last week vs. the month before” and get an answer directly in Claude without switching tabs. Free plan available.
folk — The lightweight CRM that now lives inside your texts. folk plugs into Gmail, Outlook, and calendar, builds a contact graph from your conversations, and surfaces the right follow-ups at the right moment. AI assistants added in January 2026 include a Follow-up Assistant (stalled conversations), Research Assistant (auto-enriches with Perplexity), and Recap Assistant (summarizes relationship history). Starts at $25/user/month.
Typeahead — AI autocomplete for every app on your Mac. Typeahead watches what you’re typing and surfaces completions based on your writing style, past documents, and connected knowledge bases — in any app, not just your browser. Works in Notion, Linear, Slack, email, and code editors without a plugin. Learns your voice over time. Free tier available.
Share this post with friends, especially those interested in AI!
Weekly Guides 📕
Claude MCPs for Finance, Bookkeeping & Taxes: Stripe, Mercury & QuickBooks in Claude — Our own deep-dive published June 4. How to wire Claude to Stripe, Mercury, and QuickBooks for a daily cash brief, subscription audit, cohort retention, and a tax reconciliation that caught $8K a CPA missed. Includes copy-paste prompts for every workflow, the 4 things Claude is banned from touching, and the full setup in 30 minutes.
GitHub Copilot Token Billing: Full Cost Guide and Alternatives — The complete breakdown: what is and isn’t billed, exact model pricing per token, real-world cost scenarios for different workflow types, and the five optimization moves that cut agentic session costs by 60–80%. Published June 3, the day developer backlash peaked. Read before your next billing cycle closes.
Meta Business Agent on WhatsApp: Setup Guide — Official WhatsApp Business walkthrough: go to Tools → select Meta Business Agent → authenticate → link account → confirm. How to train the agent on your past chats and catalog, configure reply settings, manage escalation to a human, and adjust behavior over time. The actual setup takes under 10 minutes. Published June 3 by WhatsApp for Business.
New MAI Models in Microsoft Foundry: How to Start Building — Microsoft's official Azure AI Foundry blog published June 2: MoE architecture breakdown of MAI-Thinking-1, how to access it via Foundry and GitHub Models, what each MAI model does and which use cases each covers, and how they slot into existing Azure and GitHub Copilot Enterprise workflows. The practical entry point for any developer evaluating Microsoft's own models after Build.
Claude Code How-To: Visual, Example-Driven Guide from Basic to Advanced Agents — Community-maintained GitHub guide synced with every Claude Code release. Now updated for Claude Code v2.1.145 with the new Dynamic Workflows feature: how to configure parallel subagent sessions, set budget limits, and run codebase-scale migrations. The practical complement to Anthropic’s official docs.
AI Meme of the Week 🤡
AI Tweet of the Week 🐦
Bonus Materials 🎁
Anthropic Confidentially Files S-1 — What It Actually Means — Fortune’s full breakdown of the filing mechanics, the October 2026 listing window, what “confidential” means in practice, and why Anthropic filing before OpenAI matters for how both IPOs get priced. The clearest explanation of what happened on June 1.
ChatGPT vs. Claude: The Real Growth Story Behind the 1 Billion User Headline — TNW’s analysis of the Sensor Tower data. ChatGPT owns consumer scale. Claude is growing at 640% YoY from a developer and enterprise base. US users who install Claude spend 5% less time on ChatGPT one month later. The market is splitting into two different products for two different use cases — and the split is accelerating.
Suno Raises $400M at $5.4B — With UMG and Sony Still Suing It — Suno doubled its valuation in 7 months despite active copyright lawsuits from Universal and Sony. Bond Capital led the round. 7M songs generated daily, $300M ARR, Warner settled and signed a licensing deal — UMG and Sony haven't. The fundraise is a direct bet that fair use wins or settles favorably. For any creator building on AI-generated music — this is the case that decides the legal ground you're building on.
Uber Caps Claude Code Usage After Bills Spiral Out of Control — Uber's engineering leadership sent an internal memo capping Claude Code and other AI tool usage after costs exceeded projections significantly. The same week Microsoft canceled its internal Claude Code licenses. Two of the largest tech companies in the world are now actively managing AI tool spend — the GitHub Copilot billing story in the context of enterprise reality.
If you missed our previous updates, don’t worry, here they are:
Opus 4.8, Anthropic $965B, GPT-5.5 wins DeepSWE | Weekly Digest
Your take: Anthropic just filed for IPO — the first major AI lab to formally begin the process toward going public. Is this the moment AI stops being a research race and becomes a public market? Or does the trillion-dollar valuation make the IPO the riskiest bet in tech history? Drop it in the comments 👇