Skip to content
All updates
Weekly AI digest

Cursor courts India's 27M developers as ChatGPT becomes a login button

2 min readMultra AI Team

Cursor launched a ~$7/month India plan ahead of its SpaceX acquisition closing, OpenAI began rolling out 'Sign in with ChatGPT' across six launch partners, and Microsoft shipped its first cyber model inside an agentic security system.

The last week of July was about distribution: the biggest AI coding tool chasing the world's second-largest developer pool, and ChatGPT quietly becoming a front door to other people's apps.

Cursor's biggest international push yet

On July 27, Cursor launched "Cursor Start," an India-only plan at ₹649/month (~$7) — a third of the $20 Pro price — bundling Composer 2.5 and Grok 4.5 (no OpenAI or Anthropic frontier models), cloud agents, MCP, hooks, and skills. India is already Cursor's #3 market and its user base there tripled in a year, in a country with 27M+ developers on GitHub. The timing matters: Cursor's all-stock acquisition by SpaceX — valued at $60B — is expected to close in Q3. Emerging-market pricing as a land-grab is a template other coding tools will copy.

'Sign in with ChatGPT' arrives

On July 29, OpenAI began rolling out "Sign in with ChatGPT" in beta with six launch partners — Airtable, GitLab, HubSpot, Notion, Supabase, and Vercel. It shares only name, email, and profile picture, but the strategic point is bigger than the scope: ChatGPT is positioning itself as an identity layer for third-party software — distribution leverage over the very apps users pick between. One more reason the chat assistants race is about far more than chat.

A $52M seed for expressive voice

On July 28, Fish Audio raised a $52M seed led by Coreline Ventures and Capital Today, on $21M ARR and 8M+ users across its open-source and hosted voice models; customers include HeyGen and Sanas. A seed that size on that revenue shows how hot steerable, expressive voice remains — compare the voice AI board.

Model watch: Microsoft's first cyber model

On July 27, Microsoft unveiled Project Perception, an agentic cybersecurity system built on multi-agent red/blue/green team loops, alongside its first security-specialized model, MAI-Cyber-1-Flash — which Microsoft says does ~95% of the work of its internal vulnerability-finding harness at roughly half the cost. Public preview via Azure AI Foundry followed on August 3. Autonomous find-investigate-harden loops, not copilots, are becoming the enterprise-security product.

What it means if you're picking tools

Distribution was the theme — cheaper regional plans, identity-layer land grabs, and compute megadeals (Recursive Superintelligence signed a $410M multiyear AWS compute agreement on July 28). Meanwhile the EU's "Digital Omnibus on AI," in force July 27, deferred the AI Act's high-risk obligations from August 2026 to December 2027, giving enterprise AI vendors 16 extra months of runway. Practical upshot: expect more aggressive pricing and bundling in coding and chat assistants — vote for the tools that earn it.


Sources: Cursor Start India plan (TechCrunch) · Sign in with ChatGPT (Vercel changelog) · Sign in with ChatGPT partners (OpenAI Help Center) · Fish Audio $52M seed (TechCrunch) · Microsoft Project Perception + MAI-Cyber-1-Flash (TechCrunch) · Microsoft security blog · Recursive Superintelligence AWS deal (TechCrunch) · EU AI Act high-risk deferral (Norton Rose Fulbright)

Get the weekly dispatch

One email a week: the AI tools that shipped and what they mean for picking the right one. No spam, unsubscribe anytime.