How to Save Operating Rules, Checkpoints and Project Context
By Sravanth Thota
·
When AI work continues across multiple chats or tools, a few pieces of saved context can make the next session much easier to start.
Not a full chat history. Not every draft. Just the parts that should keep guiding the work: rules, project background, the latest checkpoint, important decisions, and files or links that matter.
Memside gives you a place to save that reusable context so it is not buried inside one long conversation. The main choice is simple: save durable instructions as operating rules, save current progress as checkpoints, and use notes, tasks, tags, sensitivity, and references to keep the context easy to find later.
1. Save operating rules for instructions that should keep applying
An operating rule is a standing instruction. Use it when you want Memside to carry a preference or constraint into relevant AI workflows.
______________________________________________________________________________________________
Example:
Writing style : Use simple English. Avoid hype, buzzwords, and corporate-sounding phrases.When a sentence sounds too promotional, suggest a plainer version.
______________________________________________________________________________________________
How to save an operating rule manually
In Memside:
1. Open Create Memory and choose Operating Rule.
2. Add a clear Rule Title and Instruction Text.
3. Choose where the rule applies:
- Global: applies broadly across your work.
- Workspace: applies within one workspace area.
- Project: applies within one project area.
4. If you choose Workspace or Project, add a clear name for that workspace or project.
5. Choose priority only if you need one rule to matter more than another.
6. Add tags if they help you find or group the rule later.
7. Save the memory.
For most rules, start with Project if the instruction belongs to one project, and Global only if you truly want it to apply everywhere. This avoids carrying the wrong instruction into unrelated work.
How to ask AI to save one
You can also ask a connected AI assistant to help create the rule:
"Save this as a Memside operating rule for my website launch project: Use simple English, avoid unverified claims, and keep homepage copy focused on the main offer, pricing, and contact flow."
If the rule belongs to a specific project, say that in plain language. For example: "for my personal blog," "for my website launch project," or "when writing chapters for my novel." That helps the AI save the rule in the right place instead of making it too broad.
2. Save checkpoints when work reaches a useful pause
A checkpoint is not a separate memory type. In Memside, it is a Note or Task that you mark with Checkpoint Preset.
Use a checkpoint when you want to return to work later without rebuilding the background. A good checkpoint says what was completed, what is currently true, what decisions matter, what should happen next, and which project it belongs to.
______________________________________________________________________________________________
Example:
Website launch checkpoint
Completed: Homepage headline and first section are drafted. Pricing language has been simplified.
Current status: The page is clearer, but the service descriptions still need review.
Next step: Review the service descriptions and make the call-to-action more direct.
______________________________________________________________________________________________
How to save a checkpoint manually
In Memside:
1. Create a Note or Task.
2. Add the checkpoint details in the body.
3. Add a project tag, such as `website-launch`.
4. Enable Checkpoint Preset.
5. Add a short Checkpoint Summary.
6. Set sensitivity and save.
The checkpoint summary should be short and high-signal. It helps the AI understand the checkpoint quickly without reading a long memory body every time.
How to ask AI to save one
You can ask:
"Create a Memside checkpoint for my weekly newsletter. Summarize the topic, what has already been drafted, the key editorial decisions, and the next step. Tag it as newsletter."
The project or topic name matters. If you tell the AI what this checkpoint belongs to, it can later pull the right context for that work, whether it is "the mobile app redesign," "my weekly newsletter," or "the client proposal," instead of guessing from all your saved memories.
3. Use tags to keep context grouped
Tags are simple, but they matter. Use them to group memories that belong together:
Examples:
website-launch
homepage
pricing
content-review
client-a
Try to use one consistent project tag instead of five variations. For example, use `website-launch` everywhere instead of mixing `website`, `site-launch`, and `new-site`.
4. Choose sensitivity deliberately
Sensitivity controls how a memory should be handled. Private is the default for most project context. Use Secret for anything you do not want AI tools to access through normal sharing flows. Use Public carefully for user-owned context that connected AI tools can more easily notice as a lightweight hint, such as open items, wishlists, or research topics. Public does not mean visible to everyone on the internet.
5. Save references with enough explanation
A reference can be a file, link, draft, note, screenshot, or source. Do not save only the link. Add why it matters.
Weak: 'pricing-notes.pdf'
Better: 'Reference: Pricing notes
Use this file when writing or reviewing pricing page content. It contains current package names, price ranges, and claims we should avoid.'
References are useful inside project notes, checkpoints, and tasks. They help the next AI know what to read first.
A simple habit
At the end of a useful AI session, save three things:
1. Any instruction that should keep applying as an operating rule.
2. The current state of the work as a checkpoint.
3. Any project tag, sensitivity setting, or reference that will help retrieve the context later.
That is enough for most workflows. You do not need a perfect memory system. You only need to save the context that makes the next session easier to start and easier to guide.