Build 10 AI Agents with Claude Code & Claude Cowork [2026]

Explore Cloud Cowork and Cloud Code as agentic AI reads your data and autonomously builds a PowerPoint deck from Excel, executing end-to-end tasks.

Discover how to extend Claude with agent skills by creating and uploading skill.md files, and source trusted skills from official providers to tailor design, docs, and workflows.

Install the PPT visual skill and upload it to Cloud Co-Work, then enable the skill in a new session to design enhanced PowerPoint slides with better visuals and layouts.

Create an executive PowerPoint deck from quarterly sales data using the PPT visual skill and PPTX, designed for the C-suite with cloud services and consulting insights.

Explore how Claude plugins bundle skills, connectors, slash commands, and sub-agents into finance, data, and HR tools, with the model context protocol enabling standardized access to Gmail, Slack, and more.

Explore data with the data plugin in Cloud Co-Work to visualize SAS monthly metrics and build a dashboard showing MRR, churn, and regional subscriber trends.

Learn to track Claude token usage and limits using the data plugin in Cloud Cowork, visualize sas metrics, and prompt a five-slide powerpoint summary.

Learn to use Claude plugins, including the data plugin, to generate publication-quality visualizations with Pandas and automatically produce a five-slide PowerPoint deck.

Track your Claude plugin usage in the cloud desktop app, monitor plan limits, and switch between Opus, Haiku, and Sonnet models via cloud co-work.

Learn to set up Claude code with the desktop app and VS Code, configure the pro plan, and test by creating a wealth path landing page for personal finance coaching.

Install Visual Studio Code and Cloud Code, configure a project folder, and run Cloud Code to generate a sample HTML page and dashboard.

Explore cloud code, an agentic ai coding tool that reads code, edits files, and runs commands, and build Elevate Performance with Elevate HTML featuring a one-question-at-a-time intake and readiness score.

Learn to use slash commands in cloud code within VS Code to trigger, switch models (opus 4.7, haiku, sonnet), view usage, and manage context and sessions with slash init.

Learn to build a stock dashboard inside VS Code using Claude Code, fetch prices from Yahoo Finance with a python script, and generate an html dashboard with no design skills.

Install the front end design skill in VS Code to empower the Claude Code agent with production-grade dashboards, loaded by skill.md metadata and triggered by prompts or slash commands.

Use the front-end design skill to upgrade a stock dashboard by generating dashboard-version2.html with the same six stocks and data, delivering a professional, client-ready financial dashboard.

Create a Mindful Paws landing page with an interactive pet quiz for type, size, and service, delivering a Claude Code price estimate, and compare raw versus skill-based front-end builds.

Create a single html landing page for a pet grooming business with Claude Code in VS Code, featuring a mindful pause interactive booking quiz and the front end design skill.

Master context window management to optimize Claude for larger apps and personal agents. Discover how system prompts, tools, memory, and Claude.md briefing file shape sessions and token use.

Create a concise cloud.md briefing note from your project root to guide cloud code sessions, detailing the project, tech stack, structure, and commands while avoiding raw data.

Explore the Model Context Protocol (MCP), a standard for plug-and-play connectors that let Claude access Gmail, Slack, calendar, Notion, and more, with email drafting but no autonomous sending.

Create a slash command automation that runs a scheduled stock research workflow and delivers a real-time Apple monitor report with price, sentiment, key metrics, and news.

Develop an Apple monitor Excel report with a multi-tab dashboard, fundamentals, sentiment, and news sheets, embedding formulas and color-coded indicators (green for positive, red for negative, yellow for neutral).

Create a six-slide PowerPoint deck from Apple monitoring data and Excel inputs, including price, analyst sentiment, and social sentiment, via a slash command automation in cloud cowork.

Build a slash command, slash apple monitor, that automates apple research, organizes data into an excel sheet with chart, fundamentals, sentiment and news tabs, and creates a six-slide appleupdate.pptx deck.

Schedule an automation for the Apple Monitor slash command to run every two hours on weekdays during US market hours, generating sentiment analysis, Excel tracker, and a PowerPoint slide deck.

Explore building a personal AI agent with OpenClaw and Claude Code, via Cloud Code and Cowork, featuring a personality, memory vault, and automations with Gmail, Calendar, and Notion.

Explore the personal ai agent architecture with cloud code, dispatch mode, soul.md, cloud.md, a vault visualized in obsidian, brand templates, ten automations, and mcp.

The Karpathy wiki pattern uses a large language model to build a persistent, compounding personal wiki, contrasting with RAG's per-query retrieval.

Explore Obsidian as a private, interconnected wiki that visualizes your personal notes and vaults, linking pages, goals, and tags in a local markdown knowledge base.

Explore the personal OS folder structure to set up your personal AI agent, including Claude.md, Sol.md, slash commands, brand assets, and a vault with index.md and log.md.

Discover how the vault, powered by Obsidian, becomes your agent's long-term brain by storing your business data, people, and goals, with setup-driven ingestion turning raw sources into wiki pages.

Discover how to give your personal ai agent a personality with the sole.md file, encoding identity, role, writing style, priorities, and voice rules to generate authentic outputs.

Set up Claude.md as the master orchestrator playbook for your personal AI agent. Define the vault, three-layer knowledge, brand protocol, and self-correcting workflows to produce consistent PowerPoint and Excel outputs.

Set up your brand folder with logos, colors, fonts, and tone for consistent outputs. See how the setup populates brand-config.md and generates on-brand PowerPoint and Excel templates from your assets.

Explore the pre-defined slash commands in Claude Code & Cowork, including setup, new automations, status, ingest, lint, and brand, and learn how Cloud.MD levels organize onboarding and per-automation playbooks.

Learn to set up your personal AI agent, create the Vault and Soul.md, connect Gmail, calendar, and Notion via Cloud Code, and visualize with Obsidian inside the Carpathian wiki.

Set up your personal AI agent by placing a resume and a social media, goals, and writeups dossier in inbox for Cloud Code, then link LinkedIn, Maven, and YouTube data.

Connect Gmail, Notion, and Google Calendar with MCP connectors, verify the setup, then ingest inbox documents to build soul.md in the vault and configure the brand and Cleo persona.

Create and configure your personal artificial intelligence agent vault in Notion, enabling automations, a personal OS, and trackers for tasks, expenses, customer relationship management, and meeting intelligence.

Install Obsidian to visualize your vault and relationships between markdown pages, using graph view to map your agent’s memory, goals, and identity within a private wiki.

Build and test 10 automations for your personal AI agent, from sprint tracker and morning brief to market pulse and meeting intel, all integrated with Notion.

Build sprint tracker automation that reads the Notion progress board, updates stand-up status, and creates clod.md files with slash commands like sprint-tracker and slash new.

Automate a sprint tracker in Notion using Claude Code, updating the vault and status.md, generating daily stand-up summaries, and tracking velocity as tasks move from to-do to done.

Build a morning brief automation using the /morning brief command to pull unread emails, calendar, and Notion context, filter by soul.md priorities, and publish a daily brief to Notion.

Test the morning brief automation by reading emails and calendar data to extract a new teaching assistant, David Chen, and update vaults and obsidian with urgent daily briefs.

Use dispatch mode to run Claude code and Claude co-work sessions from your phone, triggering the morning brief automation that extracts Gmail and calendar data, updates Notion, and drafts emails.

Build a market pulse automation in cloud code to deliver competitive intelligence by scanning a watch list for pricing, products, careers, news, funding, and leadership moves, and update Notion.

Learn to run the market pulse automation on demand with cloud code and cloud co-work on Maven, using prompts to identify top competitors and differentiate your bootcamp.

Advance personal AI agents by building a research team and automations, including expense wrangler and CRM, integrated via a persistent vault and Notion dashboard.

Build research team automation with adaptive multi-agent design that forms sub-agents on demand, approves plans before spending tokens, and delivers branded PowerPoint or PDF reports.

Test research team automation demonstrates running ai agents to research top humanoid robotics companies and generate a five-slide PowerPoint deck in brand colors.

Learn autonomous research with patch and dispatch modes. Start a cloud code session from a phone to perform sentiment analysis on the Tesla Cybertruck and generate a pdf report.

Build a personal CRM automation that syncs Notion databases with Vault people, Gmail contacts, and calendar attendees, generates Monday follow-ups, and drafts emails in Ryan's voice.

Build a two-mode meeting intel automation with pre-meeting prep and post-meeting processing. Pull calendar events, look up attendees in the CRM, search docs, and generate a one-pager dossier.

Test meeting intel automation by parsing calendar invitations and transcripts to extract decisions and action items, update CRM and Notion, and prepare a prep dossier for pre- and post-meeting workflows.

Build an email triage automation that classifies unread Gmail into act now, read later, or archive. Operate in interactive and scheduled modes, with Gmail drafts and human-in-the-loop approvals.

Demonstrates running the email triage automation to read Gmail, classify messages, archive promos, update the Notion database, and draft prioritized replies for review.

Build expense wrangler automates receipt capture from a photo or batch mode, extracts vendor, date, amount, and category, and updates a Notion database and branded Excel sheet.

test the expense wrangler automation by starting a cloud code session in the personal os folder, processing a receipt, updating notion and the excel sheet with meals category.

Build a content machine automation to plan a 30-day content calendar and generate platform native posts in your voice using a multi-agent workflow with research, writing, and editing pipelines.

Explore how to test and run a content machine and content plan using slash commands to auto-generate cross-platform posts for LinkedIn, YouTube, Twitter, and more with Claude code and co-work.

Build a weekly executive reporting automation that ingests data from nine automations and outputs a branded seven‑slide PowerPoint plus a Notion summary. Schedule Fridays to update progress trackers.

Test weekly executive reporting automation that pulls data from multiple agents to generate a branded seven‑slide deck and a Notion summary page, exportable as PowerPoint or PDF.

Explore how tokens and the context window shape interactions with large language models, including tokenization basics, token IDs, cost, and managing context when orchestrating multiple agents.

Use the OpenAI tokenizer to compare prompts about supervised and unsupervised learning, record token counts, and reflect on an article by Sam Altman to explore token context in ai prompts.

Discover how tokenization and context drive AI agents, using the OpenAI tokenizer to count tokens and IDs, and learn that politeness can increase token usage and costs.

Explore prompt engineering fundamentals to optimize language model outputs by structuring prompts with context, instructions, input data, and output indicators, plus system prompts, avoiding hallucinations, and few-shot anchoring.

Learn prompt engineering techniques such as zero-shot and few-shot prompting, plus chain-of-thought prompting, to guide Claude and other LLMs with in-context examples and stepwise reasoning.

Learn to structure a persistent cloud.md file, manage context size, and build real-world apps with Cloud Code without coding, enabling rapid MVP development.

Manage Claude Code's context window by understanding token limits, system prompts, tools, and persistent memory; use slash compact and slash context to preserve project instructions and history.

Build a no-code calorie tracker app with cloud code, starting phase 1 by setting up a Gemini api key, .env file, and plan mode workflow.

Build a calorie tracker app from scratch with Next.js and TypeScript, using a local SQLite database to log 20 foods with calories, protein, carbs, fats, and serving size.

Advance the calorie tracker to phase two by using Gemini AI to convert text or image inputs into nutrition data, with three add options: type, snap, and quick add.

Extend the calorie tracker by integrating Gemini API calls in phase two. Build two back-end routes for text and image nutrition lookups and update cloud docs.

Wire the calorie tracker app backend to the frontend by connecting Gemini-powered API routes for text and image food queries, enabling quick add, description, and photo analysis.

Install and run Ralph Loop in Cloud Code to improve the calorie tracker through three iterations, delivering a responsive front end with macros charts, meal categories, and loading states.

Implement Ralph loops and the front-end design skill in Claude Code to iteratively review and polish a calorie tracker app with a daily calorie goal, progress bar, and meal categories.