{{first_name | Hey}}, welcome back.
Today’s issue covers ChatGPT turning conversations into interactive tools, Claude building live dashboards and animations from your data, and a green card freeze affecting eight major tech companies.
These are today's updates.
Launch production-ready sites now with Agents*
ChatGPT now builds interactive calculators, maps, charts and diagrams inside the chat
Claude turns your data into live dashboards and animated clips
Vance’s green card freeze hits Microsoft and seven other tech companies
Tools, jobs, resources, and last issue’s winning prompt ⬇️
Together with Framer
You can now build, manage, and optimize your entire website with Framer Agents. Create and design new pages, update your CMS, edit content, and handle SEO, all without jumping between tools or waiting on handoffs. Framer Agents work alongside you while you stay in control of every change. Trusted by teams at companies like Miro and Perplexity.
Agents bring speed and scale; you bring taste, judgment, and control. Together you can:
- Build and iterate faster: generate sections, refine layout and styling, and ship updates in minutes.
- Scale content with confidence: update CMS collections, localize copy, and keep metadata consistent across pages.
- Stay production‑ready: reviewable changes, collaboration, and safeguards that help teams ship without handoffs.
Top News
ChatGPT now answers some questions with working tools inside the chat. Ask about a road trip and you can get a map with stops and detours. It also builds calculators with inputs you can change, adjustable charts, diagrams with selectable parts and forms. OpenAI has shared more examples, from practical calculators to interactive explainers. Check them out here.
Screenshot from OpenAI’s launch page, comparing GPT-5.6 Instant (left) and GPT-6 Instant (right) on the same road trip prompt. Source: GPT-6 and Intelligent UI for everyone 7 Oct 2026.
OpenAI began rolling out the feature, called Intelligent UI, on October 7, making it available for all tiers. No special prompt is needed, though you can ask for a tool directly. Enterprise access depends on admin settings, and it covers the Chat tab only. The Rundown’s tutorial covers a workshop budget calculator, and its advice is to try edge cases.
Anthropic launched Claude Dashboards and Claude Motion in beta. For Dashboards, you connect Snowflake, Databricks, BigQuery, Redshift, ClickHouse or Salesforce, ask a question in plain language, and Claude builds a dashboard that refreshes as the data changes. Click any number to see the query behind it. Motion, for Team and Enterprise plans, turns text, charts and images into an editable animation you export as an MP4. Enterprise admins can turn both on under Organization settings, then Artifacts.
Cowork has merged into Claude chat, rolling out to Pro and Max first, so Claude can take a larger task, keep going after you close the laptop and check in with your phone. Claude Design builds prototypes, slides, one-pagers and campaign assets from a description, works with your design system, and exports to PPTX, PDF or HTML. It is included in every plan, including the free tier.
The Labor Department stopped accepting new PERM applications Thursday from eight companies and will not process their pending ones. Among the companies are Microsoft, Adobe, Cognizant, Infosys, Tata, Wipro, HCL and Capgemini. PERM is the labor certification employers need before sponsoring H-1B workers for green cards.
Vice President JD Vance cited Microsoft’s 6,000 American layoffs against 6,300 H-1B visas. Microsoft says about 80% of its H-1B filings last fiscal year extended or changed the status of current employees. The rule behind the action allows an initial suspension of up to 180 days, extendable while an investigation continues. H-1B status is untouched, and other employers can keep filing unless they receive a notice.
Workers nearing the six-year H-1B limit often rely on a pending labor certification to extend their status, so they feel it first. If a listed firm staffs your AI project, ask which team members rely on a pending green card and who covers if they have to leave.
Don’t miss this: OpenAI is adding an invisible watermark called textGrain to ChatGPT and Codex text in the EU, phased in over the coming weeks, to meet the EU AI Act. API customers worldwide can opt in for select models starting October 5. According to OpenAI, a missing watermark proves nothing about who wrote the text. For images and audio, OpenAI’s separate provenance post covers C2PA credentials and SynthID.
Signals
Tools
Mistral AI: Let’s you decide how AI works in your organization
Le Chonk: Unifies instruction, reasoning, and agentic capabilities in a single model (Mistral Large 4)
Hark: Personal AI assistant to get everyday things done
Resources
Poll
Choose a prompt you want us to publish next
Market
Funding
Arena raised $200M to build the AI benchmark for models and agents
Manus raised $500M to scale its AI agent business and expand hiring
Roles in AI
Embedded AI Engineer at Hark
Expert Take
Andrew Ng helped build Google Brain, co-founded Coursera, and has taught AI to roughly 8 million people. Now he thinks much of the AI fear dominating headlines is misplaced.
He breaks down the claim that AI could automate 30–40% of your job, explains what happens to the other 60%, and tells anxious college students why universities are already falling behind. He also has a surprising take on using AI to learn, and a clear test for who gets hired in marketing, recruiting, and ops over the next few years: can you build with AI?
Prompt of the Day
AI Governance Framework Builder
When to use this?
Use this when you're formalizing AI governance, reviewing your current controls, or preparing to scale AI across the enterprise. It helps turn a collection of AI policies into a practical framework for deciding what needs approval, who owns it, and which controls apply.
Act as an enterprise AI governance advisor. Help me build a practical AI governance framework for my organization that balances responsible AI, security, compliance, and business velocity.
Start by asking me one question at a time about:
1. Organization size, industry, and geographic markets
2. Current AI use cases and where AI is being deployed
3. Types of AI being used, including third-party tools, internal models, and AI agents
4. Data types involved, especially sensitive, customer, financial, or confidential data
5. Key regulations, contractual requirements, and internal policies
6. Current AI approval and review processes
7. Who owns AI governance today
8. Security, privacy, legal, and compliance requirements
9. How AI systems are monitored after deployment
10. Current gaps, incidents, or concerns
Then create a governance framework covering:
- AI inventory: What systems, models, agents, and use cases need to be tracked
- Risk tiers: Classify AI use cases as Low, Medium, High, or Critical risk, with clear criteria
- Approval workflow: Define what requires self-assessment, human review, legal/security review, or executive approval
- Data controls: Define what data each risk tier can access and what restrictions apply
- Human oversight: Specify when humans must review, approve, or override AI decisions
- Security controls: Include access management, monitoring, testing, prompt injection risks, third-party tools, and agent permissions
- Vendor governance: Define what to assess before approving an AI vendor or model
- Monitoring: Define the metrics, logs, reviews, and escalation processes needed after deployment
- Incident response: Create a simple process for reporting, investigating, and responding to AI-related incidents
- Accountability: Assign clear responsibilities across business, IT, security, legal, compliance, and executive teams
- Employee guidelines: Create practical rules employees can follow when using AI
Present the final framework as:
1. Executive summary
2. AI risk classification matrix
3. Governance roles and responsibilities
4. AI approval workflow
5. Required controls by risk level
6. AI inventory template
7. Vendor assessment checklist
8. Monitoring and incident response process
9. Employee AI usage guidelines
10. 30/60/90-day implementation plan
Keep the framework practical and proportionate. Don't create unnecessary approval steps for low-risk use cases. Prioritize controls based on business impact, data sensitivity, autonomy, scale, and potential harm.
Don't invent regulatory requirements or assume controls are already in place. Clearly label assumptions, identify gaps, and flag the 5 governance decisions leadership should make first.Get more such prompts in the Prompting Playbook (free for you)
Stay curious, {{first_name | reader}}
PS. If you missed yesterday’s issue, you can find it here.
