workflow

what Claude Projects actually are — and why you should be using them

by chaz johnson · method with ai

if you start a fresh Claude conversation every time you need help with the same ongoing work, you're doing it the hard way. Projects change that.
most people using Claude daily have never touched them.

the problem with fresh conversations

every time you start a new conversation with Claude, you're starting from zero. Claude doesn't know who you are, what you're working on, how you like things formatted, or what context matters for your situation. so you explain. and then you explain again the next day. and the day after that.

this isn't just tedious — it actively limits what Claude can do for you. the best results come from Claude having full context on your situation. when you have to re-establish that context every conversation, you're leaving most of that potential on the table.

Projects solve this directly.

what a Project actually is

a Project is a persistent workspace inside Claude that holds three things across every conversation you have within it:

custom instructions

a set of standing instructions Claude follows for every conversation in this Project — your role, your preferences, your context, how you want Claude to behave. you write it once, it's active forever.

uploaded knowledge

documents, files, and reference material that Claude can draw on in any conversation inside the Project. brand guides, past work, style references, data — whatever context is relevant to what you're building.

conversation history

all the conversations you've had inside the Project stay in one place. you can reference them, continue from them, or pick up where you left off without re-explaining the situation.

together, these three things mean Claude shows up with context every time. you stop briefing and start working.

what to put in your custom instructions

this is the most important part of a Project and the part most people underinvest in. your custom instructions are the brief Claude reads before every conversation. the more useful they are, the less work you have to do each time you open a chat.

a strong set of instructions covers four things:

who you are — your role, your industry, your level of expertise. if Claude should assume you understand technical concepts in your field, say so. if there's specific vocabulary you use, include it.

what this Project is for — the specific work, business, or ongoing task this Project is built around. be specific. "i run a solo web design business" is better than "i'm a designer."

how you want responses formatted — length, structure, tone. if you hate bullet points, say so. if you want concise answers with no preamble, say that. if you always want three options instead of one, include it.

what to avoid — the things that consistently make Claude's responses less useful for you. vague language you don't want, topics that aren't relevant, formats that don't work for your workflow.

example: Project instructions for a freelance designer

i'm a solo web designer based in michigan. i build custom-coded websites for local small businesses — primarily one-page sites and google business profiles. my clients are business owners, not tech people, so when writing client-facing copy i keep it simple and conversational. i charge $500 for a one-page site and $20/month for hosting.

for this Project: help me with client outreach copy, website copy for client builds, and business strategy. keep responses direct and concise — no bullet points unless i ask for them, no preamble, no "great question." write in lowercase throughout. if i ask for outreach copy, lead with the value to the client, not with what i do.

instructions like this mean Claude already knows your situation before you say a word. you stop re-explaining and start getting directly useful responses from the first message.

what to put in your Project files

the files you upload to a Project become part of Claude's working knowledge for every conversation inside it. think of this as the reference shelf Claude can pull from at any time.

what's worth uploading depends entirely on your work, but common high-value additions are: your brand voice document or style guide, examples of writing you want Claude to match, pricing and service documentation so Claude can answer client questions accurately, past proposals or deliverables Claude can use as templates, and research or reference material relevant to your field.

you don't need to upload everything at once. start with the one or two documents that would be most useful to have in context, and add more as you notice gaps.

the practical difference

the easiest way to understand Projects is to compare what the same request looks like with and without one.

without a Project: "write me a follow-up email for a potential client who hasn't responded to my proposal." Claude writes something generic. you spend four follow-up messages refining it toward something useful.

with a Project that has your instructions and your brand voice uploaded: same request, first response. Claude already knows you're a solo web designer, knows your pricing, knows the tone you use with local business clients, and knows you don't want it to sound like a sales email. the output is closer to usable on the first try.

"a Project doesn't make Claude smarter. it makes Claude more informed — which in practice amounts to the same thing."

how to set one up

in Claude, look for the Projects section in the left sidebar. create a new Project, give it a name that describes the work it's for, and then go into the Project settings to add your custom instructions. upload any relevant files from the same settings panel.

once it's set up, every new conversation you start inside that Project will have access to your instructions and files automatically. you don't have to do anything differently — just start the conversation.

fifteen minutes upfront. what you get back is every conversation starting from context rather than zero.
for ongoing work, that compounds fast.