This is a hypothetical business case from SasaDaily.
A 7-person commercial photography studio regularly shoots for restaurants, brands, and small to medium enterprises:
Product photos.
Brand images.
Advertising materials.
Short social media videos.
The biggest time sink, however, is often not the photo shoots themselves.
It’s:
Constant revisions of client proposals.
Clients initially request 20 product photos.
The next day, they add three short videos.
Shoot dates change.
License scopes are modified.
Finally, they want the entire proposal rewritten to "look more polished."
If the team just tells AI:
"Help me organize the latest version,"
it’s certainly fast.
But this is the studio’s biggest concern.
Because while the text looks nicer,
does the pricing change?
Does the schedule change?
Does the quantity of deliverables change?
Are licensing terms rewritten by AI?
The studio divides the proposal into two layers first
This hypothetical studio does not let Notion Agent freely rewrite the entire proposal from the start.
Instead, they split the content into:
Text that can be optimized,
and
Business terms that cannot be changed autonomously.
Text suitable for AI optimization includes:
Introductions.
Service descriptions.
Shooting process explanations.
Paragraph order.
Removing redundancy.
Clarifying unclear sentences.
Tone adjustments.
But these become protected areas:
Pricing.
Shoot dates.
Delivery quantities.
Video length.
Payment terms.
Number of revisions allowed.
Material licensing scope.
Cancellation and rescheduling rules.
These are not issues of "better wording." They represent:
What the company has actually committed to.
Step 1: Account Manager verifies facts first
Suppose the client’s latest confirmed details are:
Shoot date confirmed.
Product photo quantity confirmed.
Short video quantity confirmed.
Pricing confirmed.
The Account Manager does not immediately ask AI to edit.
Instead, they first cross-check these terms against the client’s latest confirmation.
Once confirmed, instruct Notion Agent:
“The following are confirmed business terms and must not be changed autonomously:
All pricing.
All dates.
All delivery quantities.
All licensing and payment terms.
Only improve clarity and professionalism of other text.
Please use Suggest Edits without directly modifying the original text.”
This prompt is not officially prescribed by Notion.
It is a workflow method SasaDaily designed for this hypothetical case.
Step 2: Notion Agent only proposes changes
With Notion’s August 28 update introducing Suggest Edits, the agent can:
Propose changes first,
instead of directly applying them.
For example, if the original sentence reads:
“Before the shoot, we will confirm requirements with the client, then proceed with shooting.”
The AI might suggest a more concise version.
If such edits only improve wording,
the manager can accept them.
But if the AI sees:
“Delivery expected next Friday,”
and suggests changing it to:
“Will complete delivery next Friday,”
the manager should pause.
Because “expected”
and
“will”
imply different levels of commitment.
The value of Suggest Edits is not that AI won’t make such mistakes,
but that:
It brings these differences to human attention first.
Step 3: Don’t accept all changes blindly
This studio has a simple rule:
For general wording changes,
The Account Manager can approve directly.
But for proposed changes involving four categories:
Pricing.
Dates.
Deliverables.
Business terms.
The changes must be cross-checked against client confirmations.
Even if the edit only removes or adds a single word,
they don’t accept it just because:
“AI probably just changed the tone.”
The real risk often isn’t a big mistake.
It’s:
One sentence becoming more definitive.
One limitation removed.
One date becoming vague.
One “includes one revision” being changed to “includes revisions.”
It may seem like just text,
but it’s actually cost.
Step 4: Create a clean version for the client at the end
After reviewing all Suggest Edits:
Accept necessary changes.
Reject inaccurate changes.
Then reread:
Pricing.
Dates.
Deliverables.
Licensing.
Payments.
Finally, this version becomes the official client proposal.
The whole process is not:
Client requests changes.
AI rewrites the entire document.
Send out.
But rather:
Client requests changes.
Human verifies the latest facts.
AI proposes text changes.
Human approves each item.
Recheck business terms.
Then deliver to the client.
AI is placed in the middle,
not as the final gatekeeper.
Why not let AI rewrite everything for speed?
Of course, that’s faster.
But the real issue is not:
Saving a couple of minutes per revision.
It’s:
If AI saves you ten minutes but changes a licensing term incorrectly, it could cost hours to fix later.
Errors in business documents have very asymmetrical costs.
For example:
Missing a redundant word
Costs very little.
Giving the wrong price
Costs a lot.
Writing the wrong shoot date
Costs a lot.
Stating a broader licensing scope than promised
Could cost even more.
So, the studio's approach is not:
Have humans write every sentence themselves,
but to spend human time focused on:
The most costly errors.
Can Notion Database pricing be included in Suggest Edits?
This needs to be clarified.
Currently, Notion's Suggest Edits mainly support text-based content.
For example:
Plain text.
Headings.
To-do lists.
Bulleted lists.
Numbered lists.
Database Properties are not included in Suggest Edits support yet.
If this studio keeps:
Formal pricing.
Client status.
Shoot dates.
Payment status,
in database properties,
these cannot be automatically processed through Suggest Edits item-by-item.
This serves as a good reminder:
Reviewing document text and controlling database fields are two separate tasks.
Does Notion Agent have authority to change data directly?
Notion Agent can assist with creating and editing Pages and Databases.
Notion states:
Agents generally have the same content permissions as the user.
So if an employee can edit certain content,
the agent may also assist within that person’s permissions.
This is why a workflow can’t rely solely on:
“We have Suggest Edits, so it’s safe.”
Suggest Edits addresses:
Delaying direct text modification.
Permission management addresses:
Who can access what data.
These should not be confused.
If all 7 members can edit, who’s responsible in the end?
This hypothetical studio adds another rule:
Photographers can propose content changes.
Designers can edit project descriptions.
Producers can update shooting processes.
But pricing, dates, deliverables, licensing, and payment terms
must be finally approved by the Account Manager or person in charge.
So:
The introduction of AI doesn’t blur responsibility.
On the contrary, it requires clearer definitions:
What can AI propose?
Who can approve it?
Which changes must be escalated?
This is the real workflow clarity a business needs when adopting agents.
How much time could this save?
All figures below are:
SasaDaily’s hypotheticals.
They are neither Notion’s official claims,
nor actual results from any photography studio.
Assuming the studio receives 12 client proposals or version changes per week.
Previously, each took about 18 minutes of reading, rewriting, and organizing by staff.
This equals:
216 minutes per week.
If they switch to:
AI generating Suggested Edits first,
and humans only reviewing changes and verifying business terms,
average time per revision drops to 8 minutes.
Then weekly time is:
96 minutes.
Saving about:
120 minutes per week, or 2 hours.
Based on 4 weeks per month,
that’s about 8 hours less spent on repetitive editing monthly.
Again,
these are SasaDaily’s estimates.
Different teams, document lengths, and revision complexities will produce different results.
The real savings may be more than 8 hours
The greater value might be:
Reducing uncontrolled edits.
Previously, after AI rewrote the entire document, staff had to ask:
“What exactly changed?”
Now, it’s closer to:
“Should I accept this particular change?”
It’s a completely different way of working.
Especially when the document contains:
Content.
Pricing.
Timing.
Responsibilities.
Licensing.
What is truly needed isn’t an AI better at writing,
but an AI:
That makes proposed changes visible to humans before applying them.
What can other small companies learn?
You don’t have to run a photography studio.
If your work often involves:
Client proposals.
Quotations.
Collaboration plans.
Project proposals.
Website copy.
SOPs.
Official announcements,
You can first ask:
Which parts are plain text?
Which parts are real commitments?
Then divide tasks:
AI edits text.
Humans manage commitments.
This is more sustainable than simply chasing:
“One-click complete rewriting.”
Because AI can save a small company time on editing,
but determining:
How much to charge.
When to deliver.
What quantities to deliver.
And the extent of commitments,
should always remain human responsibilities.
Today, grow together with AI a bit.
Learn one AI skill every day.
Save a little time daily.
Improve step by step every day.
SasaDaily, growing with you.
