What this tutorial solves
Most long ChatGPT tasks lose context after a few sessions because each chat starts fresh and you end up re-pasting the same instructions, files, and tone notes every Tuesday. Projects fix that by binding files, instructions, and chats into one shared space — but only if you treat the Project as a discipline, not a folder. This is for the multi-week author of a thesis, a launch plan, an ongoing audit, or any task that spans more chats than your sidebar can remember.
Who this is for
- People who paste “you are a senior PM helping me launch X, voice is direct…” into 5 chats a week.
- Anyone juggling a thesis, grant application, launch plan, audit, or product spec across many sessions.
- Writers and analysts whose recurring tasks share 80% of their context but 20% of their goal.
When to reach for it
When the same context (files, role, tone, goals) shows up in chat after chat — and your current workaround is copy-pasting from a saved doc.
When this is NOT the right tool
One-off questions, exploratory brainstorming, or tasks where context fits in a single message. Projects add overhead (setup, file pruning, instruction maintenance) that doesn’t pay back on short jobs. If you only need it twice, just paste.
Before you start
- Pick the outcome, not the topic. “Q3 launch plan” works; “ChatGPT for marketing” doesn’t — it’s too broad to give the model success criteria.
- Draft your Project Instructions in a real editor first, not in the in-app field. You’ll iterate on them more than you expect.
- Decide which 3-5 files are your daily references. Everything else stays in your local notes and gets uploaded only when needed.
- Plan the chat-naming convention up front:
2026-05-22-press-release-draftbeats “New chat” five weeks later.
Step by step
- Create one Project per long task. Name it after the outcome (“Q3 launch plan”), not the topic (“marketing stuff”).
- Write Project Instructions: role, voice, success criteria, what to ignore. Keep under 1000 words — long instructions get diluted by the model.
- Upload only the 3-5 files that you reference daily. Old drafts and tangents stay out. Old versions should be deleted, not stacked.
- Start each new chat by stating the sub-goal: “Today I want to draft the press release intro” — this anchors the model and gives you a searchable chat name later.
- When a chat gets long (>30 turns), summarize it into a
running-notes.mdfile and re-upload — Projects do not auto-remember across chats. - Every two weeks, prune Project files: remove what is stale, replace with the newer source-of-truth version. Date-stamp filenames so you know what’s current.
A real Project setup
Project name: Q3 product launch
Instructions:
- Role: senior PM helping me run a Q3 SaaS launch.
- Voice: direct, no hedging, no marketing fluff.
- Success: launch on Aug 15 with 5 press hits and a working demo.
- Ignore: anything about Q4 — that's a separate project.
Files:
- prd-2026-05-22.md (the live PRD)
- target-press-list.md (15 outlets)
- decisions.md (running list of trade-offs we settled)
- voice-samples.md (2 launch posts I'd want to write like)Quality check
- After 3 weeks, open a fresh chat in the Project and ask “summarize what we’ve decided.” If it can’t, your
decisions.mdis out of date. - Spot-check whether Project Instructions are actually being followed in chats. If a chat ignored “no hedging,” your instructions are too long or contradictory.
- Audit the file list monthly: if you can’t justify why a file is still there, delete it.
How to reuse this workflow
- For recurring task types (every launch, every audit), keep an “Instructions template” file. New Project = duplicate template + adjust 3 fields.
- The
decisions.mdpattern is the highest-leverage habit. It works in every Project you’ll ever build. - Snapshot Project Instructions to your notes app. The in-app field has no version history.
Recommended workflow
A grant-writing Project: Instructions describe tone and funder. Files include the call-for-proposals, last year’s funded sample, your CV, and decisions.md. Each chat handles one section. A “running notes” file captures decisions across chats. For Projects centered on web-grounded answers, pair this setup with a ChatGPT web search workflow so source-grounding lives next to your decisions file.
Common mistakes
- Treating Projects like folders — dumping every file in and never pruning. After a month nothing surfaces relevantly.
- Writing 5000-word Project Instructions that ChatGPT then ignores in practice. The model’s recall on instructions drops sharply past ~1000 words.
- Forgetting that Project Instructions apply to every chat, including unrelated ones you start inside it. Don’t ask cooking questions in your launch Project.
- Expecting cross-chat memory — Projects share files and instructions, not chat history. You must summarize forward yourself.
- Stacking 10 versions of the same PRD because deleting feels scary. The model will pick a random one.
- Using one Project for two related but distinct outcomes (launch + post-launch retention). The tone and audience are different; they need different Projects.
Advanced tips
- Pin the 2-3 chats that capture key decisions; treat them as the project’s memory layer.
- Keep a
decisions.mdfile inside the Project and re-upload after each major decision. The diff over time is itself documentation. - Use one Project per audience: a “client-facing” Project and an “internal” Project can share files but use different tones in Instructions.
- For technical Projects, drop a
glossary.mdwith internal acronyms — saves you 20 explanations per week.
FAQ
- Do Project files count against my context window?: Yes — only the relevant chunks are retrieved per message, but heavy file sets still slow responses. Keep it lean (under 10 files, under 500KB total is a reasonable rule of thumb).
- Can I share a Project?: Not directly — share specific chats or export the files. Project Instructions stay private to you.
- What’s the file limit?: 20 files on Plus, more on Team/Enterprise. The practical limit is much lower because relevance retrieval degrades with file count.
- Do Projects use my Memory?: Memory is account-level; it applies inside Projects too. If you don’t want it bleeding in, turn Memory off for that account or use a separate workspace.
- Why does it sometimes ignore my Instructions?: Usually because they’re too long, contradictory, or the chat’s own context has overridden them. Restate the key constraint in the chat.