Letterman Labs.
AI Weekly · Issue 6

Anthropic Launches Opus 4.7 With Design Tools

New model, visual design suite, and compliance APIs push Claude deeper into enterprise workflows and creative collaboration

№ 6 May 25, 2026 Written by machines, for humans
Hero illustration for Anthropic Launches Opus 4.7 With Design Tools
Anthropic's Claude lineup saw significant expansion this week, with Opus 4.7 rolling out, new compliance APIs, and the launch of Claude Design. Google countered with Gemini 3.5 Flash, now the default in Search's AI Mode, boasting enhanced agentic and coding prowess. GitHub Copilot deepened its IDE integration, introducing a CLI agent for JetBrains, user-level memory, and a new debugger agent workflow. The industry continues its rapid shift towards agentic workflows and deeply integrated AI assistants.

Top of the Week

Top of the Week illustration
Top of the Week

Anthropic officially launched Claude Opus 4.7, marking a notable improvement in advanced software capabilities, available as of April 16, 2026. This release is accompanied by Claude Design, a new Anthropic Labs product that allows users to collaborate with Claude on visual outputs like designs, prototypes, and slides. The company also introduced Claude Compliance API integrations on May 21, 2026, enabling IT and security teams to govern Claude across platforms, integrating it with existing security and compliance tools.

[VIDEO: https://www.anthropic.com/news/claude-opus-4-7]

This push comes alongside a flurry of Q1 2026 updates, including the expansion of Claude Cowork to Pro, Team, and Enterprise plans, and the introduction of Claude Code with multi-agent code review and remote control capabilities. Anthropic is clearly positioning Claude not just as a conversational AI, but as a deeply integrated, secure, and collaborative agent for enterprise workflows, particularly in design and software development. The strategic focus on compliance and visual collaboration signals a move beyond pure text generation, aiming for broader utility across professional domains.

This story matters because it highlights Anthropic’s aggressive strategy to embed Claude into critical enterprise functions, from creative design to secure IT governance, directly challenging competitors with a comprehensive, agent-first approach.

Claude

Claude illustration
Claude

Gemini

Gemini illustration
Gemini

Copilot

Copilot illustration
Copilot

Tools Worth Trying

  1. Midjourney V8.1 — This AI image generator shipped a stability-focused update in April 2026, making HD mode 3x faster and cheaper, and standard resolution 50% faster and 25% cheaper. It also brought back image prompts and introduced new Prompt Shortener and Describe features for quicker iteration. Ideal for creative artwork and concept designs, with an editing model expected soon.
  2. Perplexity Comet — Originally a premium product, Comet is now free on iOS (March 18, 2026), Android, Windows, and Mac. Its browser integrates Perplexity's answer engine directly with web content, offering context-aware tab assistance, voice mode, and multi-step agentic task automation. Best for real-time research with cited answers and agentic browsing.
  3. Superhuman — An AI-powered email client that drafts replies, summarizes threads, and prioritizes your inbox. It's designed to help users drowning in email achieve "inbox zero" more efficiently. Great for professionals seeking to streamline their email workflow with AI assistance.
  4. NotebookLM — Google's powerful free tool for research, allowing users to upload sources (PDFs, audio, websites) and create a grounded AI expert on that specific data. Free users can create up to 100 notebooks, each with 50 sources and up to 500,000 words. Essential for data practitioners needing to focus on information retrieval and synthesis with source citations.
  5. Pine — An autonomous AI assistant that can perform daunting real-world tasks like canceling subscriptions, negotiating bills, or appealing health insurance claims by emailing or calling on your behalf. It operates on a tiered pricing model with new users admitted via a waitlist. Perfect for outsourcing frustrating administrative tasks.
  6. Aesty — A free AI-powered fashion tool that takes images and descriptions of your wardrobe, showing how outfits look on your face and body type. It builds full looks tailored to you and your plans, and can analyze a selfie for makeup and color palette suggestions. Subscription unlocks planning and recommendation features. Great for personal styling and wardrobe management.
Gemini 3.5 Flash outperforms Gemini 3.1 Pro on key benchmarks for coding and agentic tasks, achieving 76.2% on Terminal-Bench 2.1, 1656 Elo on GDPval-AA, and 83.6% on MCP Atlas, while also leading in multimodal understanding with 84.2% on CharXiv.

The 5-Minute Action Plan

  1. Explore Claude Opus 4.7 & Design: If you're building with Claude, update your API calls to opus-4.7 and experiment with Claude Design for visual asset generation. This could unlock new creative workflows.
  2. Test Gemini 3.5 Flash in Search: Try out Google Search's AI Mode with the new Gemini 3.5 Flash for complex queries or agentic coding tasks. Observe how the "reimagined Search box" handles your requests.
  3. Enable Copilot Memory: If you're a Copilot Pro or Pro+ user, go to your personal Copilot settings and enable Copilot Memory to start personalizing your AI pair programmer's preferences.
  4. Experiment with Copilot CLI Agent in JetBrains: For JetBrains users, download the latest GitHub Copilot plugin and try delegating terminal tasks to the new CLI agent. See how it integrates with your existing editor context.
  5. Leverage NotebookLM for Research: Upload a few PDFs or web articles to Google's NotebookLM and ask it to summarize or extract key insights, creating a grounded AI expert on your specific data.
  6. Review Claude Compliance APIs: For enterprise IT/security teams, investigate the new Claude Compliance API integrations to understand how to govern Claude usage within your existing security stack.
  7. Check for Gemini Science Skills: If you're a researcher, explore the Gemini Science Skills bundle on GitHub or via Google Antigravity to see if it can accelerate your bioinformatics or genomic analyses.
When using Copilot's new user-level memory, explicitly state your preferred coding style or pull request structure in an initial prompt. Copilot will learn and apply these preferences to future interactions.
"When generating code, always prioritize functional programming paradigms and include comprehensive JSDoc comments for all functions. For pull request descriptions, structure them with a 'Problem', 'Solution', and 'Testing' section."

The Prompt Library

[Deep Research Synthesis — Multi-Source]

Use this prompt to synthesize complex information from multiple sources into a coherent, insightful report, complete with citations.

You are an expert research analyst. I will provide you with several documents or links on a specific topic. Your task is to:

1. Read and understand all provided sources.
2. Identify the key arguments, findings, and differing viewpoints across the sources.
3. Synthesize this information into a comprehensive report, structured with an introduction, main body sections (each addressing a specific aspect of the topic), and a conclusion.
4. For every factual claim or argument, cite the source using a clear [Source X] notation.
5. Highlight any areas of contradiction or significant disagreement between the sources.
6. Propose 3-5 open questions or areas for further research based on the synthesis.

Topic: [Insert specific research topic, e.g., "The impact of quantum computing on cybersecurity by 2030"]
Sources:
[List URLs or paste document content for Source 1]
[List URLs or paste document content for Source 2]
[List URLs or paste document content for Source 3]...

[Code Review — Best Practices Focus]

Use this prompt to get an AI to review a code snippet against specified best practices, identifying potential issues and suggesting improvements.

You are an experienced Senior Software Engineer performing a code review. I will provide a code snippet and a set of best practices. Your task is to:

1. Review the code for adherence to the provided best practices.
2. Identify any violations or areas where the code could be improved.
3. For each identified issue, explain why it's a problem and provide a specific, actionable suggestion for improvement.
4. Highlight potential edge cases or security vulnerabilities not covered by the current implementation.
5. Assign a severity (Low, Medium, High) to each issue.

Code Snippet:

Best Practices:


### 3. [Writing Improvement — Persuasive Tone]
*Use this prompt to refine a piece of writing to be more persuasive and impactful for a specific audience.*

You are a master copywriter. I will provide a draft text. Your goal is to rewrite this text to be more persuasive, engaging, and impactful for the target audience. Focus on:

  1. Strengthening the opening hook to immediately grab attention.
  2. Using compelling language and active voice.
  3. Clearly articulating the benefits or value proposition.
  4. Addressing potential objections subtly.
  5. Including a strong call to action (if applicable).
  6. Maintaining a [Specify Tone, e.g., "authoritative but empathetic"] tone.

Original Text:

[PASTE TEXT HERE]

Target Audience: [Describe target audience, e.g., "Small business owners struggling with digital marketing"]

Call to Action (if applicable): [Specify desired action, e.g., "Sign up for a free consultation"]


### 4. [Decision Framework — Pros & Cons Analysis]
*Use this prompt to get a structured pros and cons analysis for a complex decision, including potential risks and mitigation strategies.*

You are an impartial decision-making assistant. I need to make a complex decision. Please provide a structured analysis including:

  1. A clear statement of the decision to be made.
  2. A comprehensive list of pros (advantages) for taking this decision.
  3. A comprehensive list of cons (disadvantages/risks) for taking this decision.
  4. For each con/risk, suggest a potential mitigation strategy.
  5. Identify 3-5 key factors that should weigh most heavily in the final decision.
  6. Ask 2-3 clarifying questions that would help refine the analysis further.

Decision: [Describe the decision you need to make, e.g., "Should our company invest in developing an in-house AI solution or license an existing one?"]

Context: [Provide any relevant background information or constraints]


### 5. [Learning New Topic — Structured Curriculum]
*Use this prompt to generate a structured learning path or curriculum for a new, complex topic.*

You are an expert educator. I want to learn about [New Topic, e.g., "Reinforcement Learning from Human Feedback (RLHF)"]. Please design a structured learning path, broken down into 5-7 modules. For each module, include:

  1. Module Title:
  2. Key Concepts: A bulleted list of 3-5 core ideas to grasp.
  3. Learning Objectives: What I should be able to do or understand after completing the module.
  4. Recommended Resources (Types): Suggest types of resources (e.g., "academic papers," "video lectures," "hands-on coding exercises," "blog posts").
  5. Estimated Time: A rough estimate for completing the module.

Assume I have a [Specify Current Knowledge Level, e.g., "basic understanding of machine learning concepts"].


### 6. [Brainstorming — Innovative Solutions]
*Use this prompt to generate a diverse set of innovative solutions for a defined problem, encouraging out-of-the-box thinking.*

You are a creative innovation consultant. Our team is facing the following problem: [Describe the problem, e.g., "How can we significantly reduce customer churn in our SaaS product?"].

Please brainstorm 10-15 distinct and innovative solutions. For each solution, provide:

  1. A concise title.
  2. A 1-2 sentence description.
  3. The core mechanism or idea behind it.
  4. One potential challenge in implementation.

Encourage solutions that are unconventional, leverage new technologies, or rethink existing paradigms. Think broadly across product, marketing, operations, and customer experience.


### 7. [Summarization — Executive Brief]
*Use this prompt to condense a lengthy document or conversation into a concise executive brief, highlighting key takeaways and actionable items.*

You are an executive assistant. I need you to summarize the following [Document Type, e.g., "meeting transcript" or "research paper"] into a concise executive brief. The brief should be no more than 300 words and must include:

  1. Purpose/Context: A brief overview of the original document's goal.
  2. Key Findings/Decisions: The most critical information or outcomes.
  3. Actionable Items: Any tasks, next steps, or recommendations.
  4. Implications: What this means for our team/organization.

Original [Document Type]:

[PASTE DOCUMENT/TRANSCRIPT HERE]


### 8. [Task Breakdown — Project Planning]
*Use this prompt to break down a large project into smaller, manageable tasks, estimating effort and identifying dependencies.*

You are a project manager. I need to plan the project: [Project Name, e.g., "Develop a new mobile application for expense tracking"].

Break this project down into logical phases, and then further into detailed tasks. For each task, provide:

  1. Task Name:
  2. Description: A brief explanation of what needs to be done.
  3. Estimated Effort: (e.g., "1-2 days," "4-8 hours")
  4. Dependencies: (e.g., "Requires Task A completion," "Can run in parallel with Task B")
  5. Owner (Placeholder): [Assign a generic role, e.g., "Frontend Dev," "Backend Dev," "Designer," "QA"]

Assume a team of [Number] developers, [Number] designer, and [Number] QA.


### 9. [Debugging — Error Analysis]
*Use this prompt to analyze an error message and code snippet, providing potential causes and specific debugging steps.*

You are a debugging expert. I'm encountering an error in my code. Please analyze the provided error message and code snippet to:

  1. Identify the most likely cause(s) of the error.
  2. Explain why these causes lead to the error.
  3. Provide a step-by-step debugging strategy or specific code modifications to resolve the issue.
  4. Suggest any additional information I might need to provide for a more precise diagnosis.

Error Message:

[PASTE ERROR MESSAGE HERE]

Code Snippet:


Context/Environment: [e.g., "Python 3.9, Django 4.2, PostgreSQL"]

[Data Analysis — Hypothesis Generation]

Use this prompt to generate hypotheses and suggest analytical approaches for a given dataset or business problem.

You are a data scientist. I have a dataset related to [Describe Dataset, e.g., "customer purchase history and demographics"] and I want to understand [Business Problem, e.g., "why our customer retention rate has declined"].

Please propose:
1. **3-5 Testable Hypotheses:** Specific statements that can be proven or disproven with data.
2. **Key Metrics to Analyze:** Which metrics from the dataset are most relevant to these hypotheses.
3. **Analytical Approaches:** Suggest 2-3 statistical or machine learning methods that could be used to test these hypotheses.
4. **Potential Data Visualizations:** Recommend charts or graphs to illustrate findings.

Assume the dataset contains columns like: [List relevant column names, e.g., "customer_id, purchase_date, product_category, lifetime_value, churn_status, region"].
📚 Sources 17
  1. support.claude.comhttps://support.claude.com/en/articles/12138966-release-notes
  2. aimaker.substack.comhttps://aimaker.substack.com/p/anthropic-claude-updates-q1-2026-guide
  3. anthropic.comhttps://www.anthropic.com/news/claude-opus-4-7
  4. builder.iohttps://www.builder.io/blog/claude-code-updates
  5. blog.googlehttps://blog.google/innovation-and-ai/technology/ai/google-io-2026-all-our-announcements
  6. blog.googlehttps://blog.google/products-and-platforms/products/search/search-io-2026
  7. cloud.google.comhttps://cloud.google.com/blog/products/ai-machine-learning/innovations-from-google-io-26-on-google-cloud
  8. gemini.googlehttps://gemini.google/release-notes
  9. docs.cloud.google.comhttps://docs.cloud.google.com/gemini/enterprise/docs/release-notes
  10. releasebot.iohttps://releasebot.io/updates/github
  11. github.bloghttps://github.blog/changelog/2026-04-30-github-copilot-in-visual-studio-april-update
  12. github.bloghttps://github.blog/changelog/label/copilot
  13. github.comhttps://github.com/features/copilot
  14. datanorth.aihttps://datanorth.ai/blog/top-10-ai-tools-for-2026
  15. askglitch.comhttps://www.askglitch.com/blog/best-ai-tools-2026
  16. datacamp.comhttps://www.datacamp.com/blog/free-ai-tools
  17. techradar.comhttps://www.techradar.com/ai-platforms-assistants/im-an-ai-expert-here-are-my-5-favorite-unsung-tools-that-could-change-your-life-in-2026