Co-Planning Lessons in Shared Projects

Every department has a version of the same ritual. Three teachers who teach the same grade, the same subject, sit down once a week to "co-plan," and within twenty minutes it has collapsed into one person's Google Doc that the other two will glance at on Sunday night and quietly rewrite in their own style. The intention is real — shared planning genuinely lowers everyone's load — but the tooling fights it. A chat window forgets. A doc doesn't reason. And the context that actually makes co-planning valuable, the accumulated sense of what this class is like and what we've already tried, lives in three separate heads.

ChatGPT for Teachers has a feature that maps onto this problem more precisely than it first appears: shared projects. According to OpenAI's own description of the workspace, teachers can use shared projects to co-plan lessons and presentations with colleagues, alongside custom GPTs that work as shared templates across a school or district (OpenAI Help Center). That sounds like a filing feature. It is closer to a shared brain for a planning team — if you set it up deliberately.

What a "project" actually is, for planning purposes

Strip away the marketing and a project is a container that holds three things at once: a set of files, a running set of conversations, and a persistent instruction about what all of it is for. That third part is the one that changes co-planning.

In a normal chat, every session starts cold. You re-explain that you teach 9th-grade biology, that your district uses a particular pacing guide, that this unit is on cell transport, that your third period has four students on IEPs and a wide reading range. A project lets you say that once — in the project's instructions and its uploaded files — and then every conversation inside the project inherits it. Your colleague opening the same project the next morning gets the same context without you having to brief them.

The practical effect is that the project becomes the source of truth instead of any one person's memory. Upload the unit's standards, the pacing calendar, last year's version of the lessons, the common assessment, and a short note on the classes' needs. Now the model isn't reasoning in a vacuum; it's reasoning against the actual constraints your team is planning within.

A shared project turns "help me write a lesson" into "help me write the next lesson for this unit, for these students, consistent with what we did Tuesday." That specificity is the whole game.

A realistic PLC workflow, start to finish

Here is how a professional learning community of three teachers might actually run a week inside one shared project, rather than the idealized version.

Before the meeting. The team lead creates a project called "Grade 9 Bio — Unit 4: Transport." Into it go five files: the state standards for the unit, the district pacing guide, the Unit 3 common assessment (so the model can see the assessment style students are used to), a one-paragraph note on the three sections' composition, and last year's Unit 4 slide deck. The project instructions say, in plain language, You are helping three 9th-grade biology teachers co-plan a shared unit on cellular transport. Lessons run 50 minutes. Prioritize hands-on and visual explanations. Always align activities to the uploaded standards and flag which standard each activity hits.

During the meeting. Instead of starting from a blank page, the team asks the project to draft a five-lesson arc for the unit. Because the standards and pacing guide are in the project, the draft comes back already mapped to specific standards and already fitted to the calendar. The teachers argue about it — which is the point of co-planning — and edit directly. When they decide lesson two needs a diffusion lab, they ask for three lab options at different prep levels, and the model proposes them against the 50-minute constraint it already knows about.

After the meeting. One teacher takes the agreed arc and asks the project to expand lesson one into a full plan with a warm-up, a mini-lecture outline, and an exit ticket. Another asks it to generate a parent-facing summary of the unit. A third asks for a differentiated version of the reading for the lower-Lexile students. All three are working in the same project, so the outputs stay consistent with each other and with the shared standards — nobody is silently drifting into their own private version.

The difference from the Google Doc ritual is that the context compounds. By Unit 6, the project has a history of what worked, what got cut, and how these particular classes respond. That institutional memory is exactly what usually walks out the door when a teacher changes schools.

Custom GPTs as the department's shared templates

Projects hold the context for one unit. Custom GPTs hold the method your department wants to reuse across every unit. This is the second half of the co-planning story, and it's where a school or district can standardize without flattening.

Say your department has a house style for lesson plans — a specific warm-up structure, a required standards alignment, a preferred exit-ticket format. Instead of every teacher re-teaching that format to ChatGPT each time, someone builds a custom GPT that encodes it once. OpenAI describes custom GPTs in the teacher workspace as shared templates that can be distributed across a school or district (OpenAI Help Center). A new teacher joining in October inherits the department's planning method on day one by opening the shared GPT, rather than reverse-engineering it from a veteran's example.

The pairing is the useful mental model: custom GPTs carry the method, shared projects carry the context. A department-standard "Lesson Plan Builder" GPT used inside a unit-specific project gives you consistency across teachers and fidelity to this particular unit's needs at the same time. That's the combination that has been genuinely hard to achieve with a shared drive full of templates nobody follows the same way.

If your department doesn't yet have that house style codified, that's the prerequisite work — and it's worth doing before you build the GPT, because a custom GPT is only as good as the method you pour into it. This is the same reasoning behind building a custom GPT for your department, which walks through encoding a shared method in more depth.

Where the honesty has to come in

Two cautions keep this from becoming another abandoned tool.

The first is that co-planning software has never failed for technical reasons — it fails socially. A shared project only stays useful if the team agrees on who curates it. If all three teachers dump conflicting files and rewrite the instructions in different directions, the project degrades into the same mess as the shared drive. Assign one person per unit to own the project's setup, exactly as you'd assign a facilitator to a meeting. The tool doesn't remove the need for that agreement; it just makes the agreement pay off more.

The second is that the model drafts; it does not decide. A five-lesson arc that comes back beautifully aligned to standards is still a proposal. The judgment about whether a diffusion lab is right for your third period is yours, and it draws on things no uploaded file captures — the kid who shut down last week, the fire drill that ate Tuesday. Treat the project as a faster first draft and a shared memory, not as the planner. The teachers are still the planners.

Used that way, shared projects do something the Friday co-planning ritual rarely manages: they make the second week easier than the first. Once the unit context is loaded and the department's method is encoded, next week's plan starts from everything you already know, not from a blank page and three separate memories.

The next thing that shared context unlocks is grading — because the same uploaded rubric that shaped your lessons can shape how you prep feedback.

---

Part 81 of 100 in the ChatGPT for Teachers series. Previously: Building a Custom GPT for Your Department. Next: Turning File Uploads Into Faster Grading Prep. Browse more builder insights or explore AI skills for education at aiskill.market.