Here’s what’s shaping AI right now - without the noise:

Most people use Claude the same way. Open a chat, type a question, get an answer, close the tab. That works. But it leaves most of Claude's actual capability sitting unused.

Here is what the full ecosystem looks like and when each piece genuinely earns its place in your workflow.

Projects - so you stop re-explaining yourself

A Project is a dedicated space for one ongoing thing. You load it once with reference files, background context, and standing instructions. Every conversation inside that Project already knows what you are working on before you type a single word.

If you have ever started a chat by spending the first five messages catching Claude up on your situation, Projects eliminate that entirely. Set it up once. Never repeat yourself again.

Best for: ongoing client work, long writing projects, codebases you return to regularly.

Connectors - so you stop switching tabs

Connectors let Claude read directly from Google Drive, Gmail, Slack, and other tools without you copying and pasting content between apps. Turn it on once in settings. After that, Claude can pull the file, the email thread, or the Slack message directly inside the conversation.

It sounds minor. In practice it removes a surprising amount of friction from information-heavy work.

Best for: anyone who regularly pulls content from external tools into Claude conversations.

Cowork - so you stop supervising every step

Cowork is for when you want something done rather than a conversation about it. Point it at a folder, describe the outcome you want, and it works through the steps on its own reading files, editing them, creating new ones, checking its own output.

The mental shift here is significant. Instead of guiding Claude turn by turn, you hand off the task and check the result. That changes what "using AI" actually feels like for longer jobs.

Best for: batch processing, folder-level tasks, anything with clear inputs and a defined output.

The professionals pulling ahead aren't working more. They're using Claude.

Our free guide will show you how to:

Transform your workflow with AI and stay ahead of the curve with this comprehensive guide to using Claude at work.

Get The Free Guide

Skills - so you stop retyping the same instructions

A Skill is a saved set of instructions stored as a SKILL.md file. You write your preferences once - how you want code reviewed, how you want emails structured, what format you want research in and invoke it by name from then on.

Once you have two or three Skills set up, you notice how much repetitive instruction you were giving Claude every single session. Skills make that repetition permanent on the first pass.

Best for: any recurring task where you have specific preferences Claude keeps needing to relearn.

Subagents - so you stop cramming everything into one conversation

For large tasks - a big refactor, a multi-part research job, anything with a lot of moving pieces - Claude Code can split the work across subagents. Each subagent gets its own context and its own permissions, handles one piece, and reports back.

The practical benefit is a cleaner main thread. Long single conversations lose context as they grow. Subagents keep the work compartmentalized so nothing gets dropped.

Best for: large coding tasks, complex multi-step projects, anything that feels too big for one conversation.

Plugins - so you stop rebuilding the same setup

Once you have a working combination of Skills, Connectors, and Subagents set up for a project, Plugins let you package the whole thing and install it in one step next time. You can also hand it to a teammate so they get the same configuration without any manual setup.

Best for: teams running similar workflows across multiple projects, or anyone who sets up the same configuration repeatedly.

MCP Servers - for the tools Connectors do not cover

Most people never touch this directly. When you turn on a Connector, you are actually connecting to an MCP server without seeing the underlying wiring. But when you need Claude to reach something that does not have a consumer Connector yet - an internal tool, a niche data source, a custom integration - an MCP server is how you get there.

Not something you reach for every day. But when you need it, nothing else fills the gap.

Best for: developers and operators who need Claude to connect to tools outside the standard Connector library.

Routines - so you stop remembering to do things manually

Routines are the newest piece and they solve something none of the others touch: what happens when you are not the one who needs to ask?

A Routine is a saved Claude Code configuration - a prompt, a connected repo, a set of Connectors - that runs on Anthropic's cloud infrastructure on a schedule or a trigger. Without you opening your laptop. You set it up once, and it just runs. You find out what happened after the fact instead of having to remember to kick it off yourself.

Best for: recurring checks, nightly reports, scheduled tasks you kept forgetting to do manually.

How they fit together

Projects hold your context. Connectors and MCP Servers reach your tools. Skills teach a recurring task. Subagents split up the big jobs. Plugins package everything for reuse. Cowork runs tasks autonomously. Routines run all of it without you.

You do not need all eight on day one. Most people start with Projects and Connectors and find those two alone meaningfully change how much time they spend on setup versus actual work. Add the others as the need becomes obvious.

The tools do not become useful just because they exist. They become useful when they fit how you actually work.

Get revenue agents, workflows, and automations across every stage of your motion. Access customer data in real time through Attio's web app, MCP, API, and SDK.

Then Ask Attio anything about your business and get instant answers.

It's the CRM that runs the work behind every win.

Try Attio Now

A few tools quietly worth exploring:

See you next issue.