Package Your AI Recruitment Portfolio

Turn your Bubble AI recruiter prototype into a safe, evidence-backed case study.

Introduction

30 Second Summary

Finished prototypes often disappear into old editor tabs once a demo ends. A reviewer cannot see the thinking behind the work from a tool list or a confident claim.

In this project, you will package your completed Bubble AI recruitment assistant as an evidence-backed product case study without spending more credits. You will turn static prototype evidence into a case file you can present in two minutes.

What You'll Build

You will hand a reviewer one compact case file that takes them from the recruiter problem to inspectable prototype proof within minutes.

By the end of this project, you'll have:

A scannable project card that lets a reviewer grasp your product story in seconds.
Three portfolio-safe screenshots that let a reviewer inspect the interface, workflow, plus API integration. Credentials and candidate data stay out of view.
A concise case study you can deliver as a two-minute interview walkthrough. Its architecture flow makes the product decisions easy to follow. Its limitations keep your claims honest.
Secret Mission: A five-case quality and safety rubric that defines safe output expectations plus human-review triggers for synthetic candidate scenarios.

Are there any prerequisites?

You need a Windows device plus a completed Bubble AI recruitment prototype with the API Connector already configured. Static evidence avoids another Gemini API call, so this project requires no additional spend.

Before We Start

Before any portfolio work begins, lock in the recruiter problem your completed Bubble assistant addresses. Your one-sentence definition should explain what it helps a recruiter do plus why that task matters.

Set Up a Zero-Cost Evidence Workspace

Portfolio files can easily become mixed with private resumes. Credentials or unfinished app material can also slip into the evidence pack.

A separate local workspace keeps every portfolio artifact controlled. You will prepare it with Bubble, Snipping Tool, and Notepad without running another AI request.

In this step, get ready to:

Open the completed Bubble prototype without running it.
Confirm Snipping Tool and Notepad open successfully.
Create a local workspace with an evidence folder and a blank case-study file.

Keep the completed prototype idle

The Bubble editor can display the completed interface without starting the recruitment workflow. Keeping the prototype idle preserves your existing work without using more of the testing allocation.

Open Bubble's project list in your browser.
Select your completed AI recruitment prototype.
Confirm the editor displays the completed recruiter-facing interface.
Leave the Preview control untouched.
Leave every workflow execution control untouched.
Leave every API Connector initialization control untouched.

You have the safest starting point now. Your completed interface is visible while its workflow stays idle.

Why Keep the Editor Idle?

Bubble API initialization sends a real request to the connected provider. Keeping Preview, workflow execution, and initialization untouched prevents another API call.

Confirm the Windows tools

Snipping Tool handles the evidence captures in the next step. Notepad gives your written case study a simple local home.

Press the Windows key to open the Start menu.
Type Snipping Tool.
Select Snipping Tool from the results.
Confirm the Snipping Tool window opens.
Minimize Snipping Tool.

The capture tool is installed and ready for your evidence work.

Press the Windows key to return to the Start menu.
Type Notepad.
Select Notepad from the results.
Confirm a blank Notepad window opens.

Good. Both offline tools are available for building the portfolio pack without running the prototype.

Having Trouble Opening a Tool?

Repeat the Start search with the full product name. Select the installed application if Windows also displays web results.

If either application still refuses to open, help me troubleshoot the Windows application.

Create the local workspace

File Explorer keeps the portfolio draft separate from private project material. The evidence subfolder gives every safe screenshot one clear destination.

Press Windows logo key + E to open File Explorer.
Select Documents in the left navigation pane.
Create a new folder by pressing Ctrl + Shift + N.

You will see a new folder with its temporary name selected.

Type AI-Recruitment-Portfolio as the folder name.
Press Enter.

You will see the new AI-Recruitment-Portfolio folder inside Documents.

Open the AI-Recruitment-Portfolio folder.
Create a new subfolder by pressing Ctrl + Shift + N.

A second folder will appear with its temporary name selected.

Type evidence as the subfolder name.
Press Enter.

The empty evidence subfolder is now ready for portfolio-safe images.

Switch back to the blank Notepad window from earlier.
Confirm the editing area is blank.
Select File.
Select Save As.

You will see the Save As dialog for choosing the file location.

Select Documents in the Save As dialog.
Open the AI-Recruitment-Portfolio folder.

You will see the evidence subfolder in the selected destination.

Enter portfolio-case-study.txt as the file name.
Select Save.

The Notepad title bar now shows portfolio-case-study.txt. Your blank draft is saved inside the portfolio workspace.

Return to the File Explorer window from earlier.
Confirm the evidence subfolder is visible.
Confirm portfolio-case-study.txt is visible.

That is your evidence home ready. The workspace contains one empty evidence folder plus one blank case-study file.

Can't Find the Saved File?

Switch back to Notepad.
Select File.
Select Save As.
Confirm the destination is the AI-Recruitment-Portfolio folder inside Documents.

If the file remains missing, ask for help locating the saved Notepad file.

✔️ My workspace matches

Great. Your local evidence workspace is saved while the completed Bubble app remains idle.

ⓧ I need to double check

Confirm the local AI-Recruitment-Portfolio folder exists in Documents.
Confirm the AI-Recruitment-Portfolio/evidence subfolder exists.
Confirm the AI-Recruitment-Portfolio/evidence subfolder is empty.
Confirm the blank AI-Recruitment-Portfolio/portfolio-case-study.txt file is saved.
Confirm the Bubble editor remains open without a Preview session.
Confirm no workflow execution or API initialization has been triggered.
Confirm Snipping Tool opened successfully.
Confirm Notepad opened successfully.

Before the final check, what should you see in each half of the screen?

Switch back to portfolio-case-study.txt in Notepad.
Drag the Notepad window to the left side of your screen.
Restore the Bubble editor from earlier.
Drag the browser window to the right side of your screen.

You should see the blank case-study draft beside the completed recruiter-facing interface. No Preview session, workflow execution, or API initialization should be active.

Your zero-cost evidence workspace is ready. Next, you will capture proof of the interface, workflow, and API integration without running the app.

Capture Proof of the Prototype

Your zero-cost workspace is ready beside the completed Bubble editor. You can now collect proof that the recruiter-facing prototype exists without using another workflow run.

Screenshots make the completed work inspectable when a live demonstration is unavailable. Raw images still leave the recruiter problem unstated.

They also hide the product decisions behind the interface. This step collects the safe visual foundation that your case study needs.

In this step, get ready to:

Capture the completed interface with Snipping Tool while excluding account details and real candidate information.
Capture the main Bubble workflow event and action sequence without revealing candidate data.
Capture the API Connector configuration while excluding private integration details.

Capture the product interface

The interface capture proves that the recruiter-facing product exists. A tight frame keeps the reviewer focused on the completed experience.

Switch back to the Bubble editor from earlier.
Leave Preview closed.
Position the editor so the completed recruiter-facing interface fills the usable page area.
Check the visible area for browser account details.
Check the visible area for real candidate information.
Press Windows logo key + Shift + S.
Drag the capture frame around the product area.

What Does This Capture Prove?

This image gives a reviewer direct evidence of the interface you completed. The narrow frame keeps private browser details outside the portfolio evidence.

Switch back to Snipping Tool from earlier.
Press Ctrl + S to save the capture.
Choose the AI-Recruitment-Portfolio/evidence folder as the destination.
Enter raw-interface-capture.png as the file name.
Confirm the save.
Inspect the saved image in Snipping Tool at full size.

You should see the completed product area clearly. Browser account details and real candidate information should stay outside the image.

Is the Interface Capture Unclear?

Retake the capture with a tighter frame if browser details remain visible.
Increase the Bubble editor area before recapturing if the interface text is unreadable.
Replace the image if any real candidate information appears.

Help me create a readable Bubble interface capture without exposing private information.

Capture the Bubble workflow

The workflow capture proves how a recruiter action reaches the analysis result. The useful evidence is the event and action sequence itself.

Return to the Bubble editor.
Switch to the workflow editor for the completed app.
Select the existing workflow that starts candidate analysis.
Arrange the view so the user-action event is readable.
Arrange the view so the analysis-result action sequence is readable.
Keep any values that may contain candidate data closed.
Leave every workflow execution control unused.
Press Windows logo key + Shift + S.

Drag the capture frame around the event and action sequence.
Switch back to Snipping Tool.
Press Ctrl + S to save the capture.
Choose the AI-Recruitment-Portfolio/evidence folder as the destination.
Enter raw-workflow-capture.png as the file name.
Confirm the save.
Inspect the saved image in Snipping Tool at full size.

You should see the event that begins candidate analysis. You should also see the sequence that produces the analysis result without any candidate values being open.

Is the Workflow Sequence Hard to Read?

Resize the workflow editor area before recapturing if the event name is cut off.
Reposition the sequence before recapturing if the relevant actions fall outside the frame.
Discard the image if an opened value reveals candidate data.

Help me frame my Bubble workflow so the event and action sequence are readable without exposing candidate data.

Capture the API connection

The API connection is implementation proof that Bubble connects to the Gemini API. The collection or call-level view provides enough evidence for this portfolio.

This is the most sensitive capture in the set. You can prove the integration while keeping every credential and payload outside the image.

Return to the Bubble editor.
Select API Connector in the left-hand navigation menu.
Keep the view at collection or call level.
Leave API initialization untouched.
Collapse any area that shows a credential.
Collapse any area that shows request or response values.
Reframe the view to exclude project identifiers.
Reframe the view to exclude candidate details.

Why Capture Less?

API keys should never appear in screenshots. Snipping Tool automatic redaction is documented for email addresses and phone numbers.

Credential safety requires a deliberate visual check before capture. A narrow image proves the external connection without publishing its private contents.

Press Windows logo key + Shift + S.
Drag the capture frame around the collection or call-level connection proof.
Switch back to Snipping Tool.
Press Ctrl + S to save the capture.
Choose the AI-Recruitment-Portfolio/evidence folder as the destination.
Enter raw-api-connection-capture.png as the file name.
Confirm the save.
Inspect every visible part of the saved image at full size.

You should see enough of the API Connector to prove that an external service is connected. Credentials and private request data should remain completely absent.

Did Private Information Enter the Capture?

Discard any image that contains a credential or private value. Retake the capture with that area closed or outside the frame.

Replace the capture if a request value or response value is readable.
Replace the capture if a project identifier appears.
Replace the capture if any candidate detail appears.

Help me identify sensitive information that must stay out of my API Connector screenshot.

Before you inspect the set, do you think the three files can tell the whole product story on their own?

Open the evidence subfolder inside AI-Recruitment-Portfolio in Windows.
Confirm that raw-interface-capture.png is present.
Confirm that raw-workflow-capture.png is present.
Confirm that raw-api-connection-capture.png is present.
View raw-interface-capture.png at full size.
View raw-workflow-capture.png at full size.
View raw-api-connection-capture.png at full size.

What Should I See?

The interface capture shows the completed recruiter-facing product without browser account details or real candidate information.

The workflow capture shows the user-action event and analysis-result sequence without revealing candidate data.

The API connection capture shows the external integration without credentials or private configuration values.

Viewed in sequence, the set proves that the interface and implementation path exist. It also proves that the external connection was configured.

The images still leave the recruiter problem and product decisions unstated. That shortfall is the intended result of collecting raw evidence first.

That is the proof layer secured. Next, you will turn these raw captures into an ordered product story that is safe to publish.

Create a Safe Product Story

You now have three raw captures from your completed Bubble prototype. Together they prove that the prototype exists.

Raw screenshots still leave a reviewer to guess why each screen matters. They also need a careful privacy check before they become portfolio evidence.

In this step, you will create an ordered evidence set. You will connect each image to recruiter value through captions.

In this step, get ready to:

Create three ordered screenshots from the raw captures.
Write a reviewer-focused caption for each screenshot.
Add an architecture flow to complete the product story.

Crop and save the safe evidence set

A portfolio screenshot needs a clear subject plus a safe publication boundary. You will use Snipping Tool to keep the useful proof inside that boundary.

Leave the Bubble editor on its current screen.
Before you inspect the captures, predict whether a reviewer can explain the recruiter problem from these images alone.
Return to AI-Recruitment-Portfolio/evidence in File Explorer.
Inspect raw-interface-capture.png at full size.
Inspect raw-workflow-capture.png at full size.
Inspect raw-api-connection-capture.png at full size.

You can identify the interface plus its implementation screens. The images alone do not explain the recruiter problem or your product decisions.

Protect the Source Material

Screenshots can expose private information after they leave your computer. API initialization also sends a real request to the provider.

Avoid Preview during this step.
Avoid workflow execution during this step.
Avoid API initialization during this step.
Discard any source image that contains a key or token.
Retake any source image that contains candidate identity or contact details.
Use cropping as the publication boundary for every saved image.

Select the three raw capture files in File Explorer.
Copy the selection with Ctrl+C.
Paste three duplicates with Ctrl+V.

You should see one duplicate of each raw capture. The original files remain available if you need to restart a crop.

Rename the copy of raw-interface-capture.png to 01-interface.png.
Rename the copy of raw-workflow-capture.png to 02-workflow.png.
Rename the copy of raw-api-connection-capture.png to 03-api-connection.png.

The numeric prefixes lock the evidence into the order a reviewer should follow. The sequence moves from the product experience to its implementation proof.

Use File Explorer's Open with option to load 01-interface.png in Snipping Tool.
Select Image crop.
Adjust the crop boundary to keep the completed product area in frame.
Save the cropped image with Ctrl+S.
Inspect 01-interface.png at full size in File Explorer.

You should see the completed recruiter-facing interface. Browser account details plus candidate information should remain outside the frame.

Use File Explorer's Open with option to load 02-workflow.png in Snipping Tool.
Select Image crop.
Adjust the crop boundary to keep the event-to-action sequence readable.
Save the cropped image with Ctrl+S.
Inspect 02-workflow.png at full size in File Explorer.

You should see the user-action event plus the analysis-result action sequence. Candidate data plus sensitive values should remain hidden.

Use File Explorer's Open with option to load 03-api-connection.png in Snipping Tool.
Select Image crop.
Adjust the crop boundary to keep only the safe collection-level or call-level connection proof.
Save the cropped image with Ctrl+S.
Inspect 03-api-connection.png at full size in File Explorer.

You should see proof that the external connection exists. Credentials plus request details plus response details should remain outside the frame.

Found Private Information?

Delete the affected ordered copy.
Create a fresh copy from the corresponding raw capture.
Repeat the crop with the sensitive region outside the boundary.

If you are unsure whether a detail is safe to publish, help me review a screenshot privacy concern.

Write captions that explain the evidence

Your Notepad file gives each screenshot the context it currently lacks. A reviewer should understand the visible proof without opening the prototype.

What Makes a Useful Caption?

The visible screen tells the reviewer what they are looking at.
The demonstrated capability explains what the screenshot proves.
The recruiter value connects that capability to the user problem.

Return to portfolio-case-study.txt in Notepad.
Add your one-sentence recruiter-task problem statement at the top of the file.
Read the statement to confirm it names the recruiter task plus its importance.

Write one caption for 01-interface.png using the three-part caption structure.
Review the caption to confirm it explains the interface's value to a recruiter.
Write one caption for 02-workflow.png using the three-part caption structure.
Review the caption to confirm it explains the workflow's role in candidate analysis.

Write one caption for 03-api-connection.png using the three-part caption structure.
Review the caption to confirm it explains what the external connection proves.
Save the changes to portfolio-case-study.txt.

Your screenshots now carry their own product context. A reviewer can connect each screen to a capability plus a recruiter need.

Do the Captions Sound Vague?

Replace broad claims with details that are visible in the corresponding image. Keep business results out unless you have evidence for them.

For another pass, help me strengthen my screenshot captions.

Map the product architecture and audit the story

An architecture flow shows how information moves through the prototype. The final recruiter review point makes the human decision boundary visible.

Add the product architecture beneath your three captions by pasting this exact line:

Resume PDF -> Bubble upload interface -> Bubble workflow -> API Connector -> Gemini API -> Structured candidate analysis -> Recruiter review

What Does This Flow Show?

The resume enters through the Bubble upload interface.
The workflow passes the request through the API Connector to the Gemini API.
The structured analysis reaches a recruiter for review.

This line documents the completed architecture. It does not execute the workflow or consume another API request.

Save the completed portfolio-case-study.txt file.
Before the final review, predict whether a recruiter can now identify the interface plus the implementation path plus the human review point.
Inspect 01-interface.png at full size.
Inspect 02-workflow.png at full size.
Inspect 03-api-connection.png at full size.
Return to portfolio-case-study.txt in Notepad.
Read the file from top to bottom.

You should see 01-interface.png as the product interface. You should see 02-workflow.png as the event-to-action sequence.

You should see 03-api-connection.png as the external connection proof. Each image should have a matching caption in portfolio-case-study.txt.

No image should reveal a key, token, email address, or phone number. No image should reveal a resume, candidate identity, request value, response value, or project identifier.

The text file should contain your recruiter-task problem statement. The architecture flow should match the line above exactly.

That is the evidence pack transformed into a product story. It now proves the prototype exists without exposing the people or credentials behind it.

Does the Story Feel Incomplete?

Compare each caption with the three-part caption structure.
Restore any missing ordered screenshot from its raw capture.
Copy the exact architecture flow from above again if its stages changed.

If the evidence still feels disconnected, help me find the missing link in my product story.

Your evidence now tells a safe story from interface to recruiter review. Next up, you will turn that story into the full AI product case study.

Write the AI Product Case Study

You handled the careful part. Your recruiter-safe evidence now shows the completed Bubble prototype from three angles.

A polished interface alone cannot show product judgment. This step records the problem framing behind the prototype.

You will also capture the trade-offs. Your draft will state the responsible AI boundaries.

In this step, get ready to:

Structure the case study so a reviewer can follow the product story.
Document genuine product decisions plus a clear limitation.
Verify the draft uses static evidence without claiming unverified results.

Structure the case study

A case study gives each piece of evidence a clear job. Its sections guide a reviewer from the user problem to your decisions.

This writing step stays inside Notepad. The Bubble editor remains reference-only.

The Product Flow section carries the technical path from the API Connector into the Gemini API.

Switch back to portfolio-case-study.txt in Notepad from earlier.
Add these headings in this order: Project Summary, User Problem, My Role, Product Flow, Key Decisions, Evidence, Limitations, Next Steps.
Move your existing one-sentence recruiter-task problem statement under User Problem.
Scan the document from Project Summary to Next Steps.

You should see all eight sections in a clear reading order. Your existing problem statement should now sit under its matching heading.

Write two or three sentences under Project Summary that identify the prototype’s purpose.
Describe your contribution under My Role using only work you actually completed.
Reread both sections to confirm a reviewer can identify the product plus your contribution.

The opening sections should now explain what the prototype does. They should also make your ownership easy to identify.

Move the existing architecture flow under Product Flow.
Move the three existing screenshot captions under Evidence.
Read the Product Flow section from top to bottom.
Read the Evidence section from top to bottom.

The flow should read Resume PDF -> Bubble upload interface -> Bubble workflow -> API Connector -> Gemini API -> Structured candidate analysis -> Recruiter review. The Evidence section should retain one caption for each ordered image.

What makes this evidence credible?

The screenshots show that the interface, workflow, plus external connection exist. Their safe crops keep private details outside the portfolio.

The captions explain the product capability behind each image. This combination makes the evidence inspectable without requiring a live demonstration.

Record genuine product decisions

A decision trail is credible when it reflects choices you genuinely made. Each entry needs a reason or a trade-off.

Under Key Decisions, explain why a no-code prototype fit the original testing goal.
State the trade-off that came with the no-code approach.
Read both lines to confirm each one includes a reason.

Your first decision should show why speed or accessibility mattered for the prototype. Its trade-off should reflect a constraint you actually encountered.

Add a separate decision that positions the AI output as support for recruiter judgment.
State that a recruiter reviews the analysis before any hiring decision.
Review these lines to confirm the human-review boundary is explicit.

Why keep a recruiter in the loop?

The prototype organizes candidate analysis for review. A recruiter remains responsible for interpreting that analysis within the wider hiring context.

This boundary demonstrates product judgment. It also gives the case study a concrete responsible-AI safeguard.

Add one decision that names work you deliberately kept out of scope.
Under Limitations, describe one constraint in the current demonstration.
Reread both entries to confirm they describe real project boundaries.

Your decision trail should now cover the no-code approach, recruiter judgment, plus deliberately excluded work. The Limitations section should name a genuine constraint without turning it into an unverified result.

Add the status and verify your claims

A portfolio case study needs a clear status boundary. The reviewer should understand what was completed plus why the evidence is static.

Add this statement under Limitations: Functional prototype built and tested. This portfolio entry uses static evidence because the original testing allocation was exhausted.
State under Evidence that the screenshots demonstrate prototype completion.
Read the updated Evidence section plus the updated Limitations section.

The draft should now distinguish completed prototype work from future validation. Its status should remain clear even when a reviewer only scans the evidence.

Keep the claims evidence-backed

Screenshots provide evidence that the prototype was completed. Production adoption requires separate evidence.

Accuracy, hiring outcomes, plus business results also require their own validation. Leaving those claims out makes the case study more trustworthy.

Check the draft for claims about production adoption, accuracy, hiring outcomes, or business results.
Delete any unsupported claim you find.
Write one concrete future validation step under Next Steps.
Read the completed Next Steps section.

The final section should describe a future test or learning goal. It should avoid presenting that future work as a result you already achieved.

Save your changes to portfolio-case-study.txt.

Your saved draft is ready for a complete review. The final pass checks whether every required element is easy to locate.

Predict whether every required element will be easy to find on the first pass.
Read portfolio-case-study.txt from top to bottom.
Confirm User Problem names the recruiter task plus its importance.
Confirm My Role identifies your genuine contribution.
Confirm Product Flow contains the complete architecture sequence.
Confirm Evidence contains the three matching screenshot captions.

The first half of the draft should establish the problem, your ownership, the working flow, plus the visual proof. Each requirement should have a clearly labeled home.

Confirm Key Decisions contains at least two genuine product decisions.
Confirm one decision keeps recruiter judgment central to the workflow.
Confirm Limitations names one real constraint.
Confirm the exact prototype-status statement appears in the draft.
Confirm the draft contains a responsible-AI safeguard.
Confirm Next Steps contains one future validation step.
Confirm the draft makes no unsupported claims about adoption, accuracy, hiring outcomes, or business results.

You now have a complete case-study draft that makes the prototype inspectable without another API call. Your product reasoning is as visible as the interface itself.

Case study still feels generic?

Generic wording usually hides the decision trail. Specific constraints plus defensible trade-offs make your role easier to understand.

Ask for help to make my case study more specific without inventing outcomes.

Your product story is documented. Next, you will condense it into a scannable project card plus a two-minute interview walkthrough.

Publish a Scannable Portfolio Entry

Your recruiter-safe product case study now explains the completed Bubble prototype. It also records the product decisions behind it.

Hiring reviewers often scan before they read deeply. This step turns your full case study into a portfolio entry they can understand quickly.

In this step, get ready to:

Create a project card that summarizes the prototype for a quick review.
Arrange the three evidence images before the architecture flow.
Prepare a rehearsed two-minute interview walkthrough.

Why Static Evidence Works Here

The completed screenshots already prove that the prototype exists. Your case study supplies the product reasoning behind that proof.

This packaging approach preserves the exhausted testing allocation. It also avoids an unnecessary API call.

Build the project card

A project card is the skim layer for your full case study. It answers a reviewer's first questions before they decide whether to read further.

Return to the top of portfolio-case-study.txt in Notepad from earlier.
Add a clear project title above the existing case study.

You should now see the project title before the existing Project Summary section.

Write the first summary sentence around the recruiter problem.
Write the second summary sentence around the prototype solution.

The two-sentence summary should identify the user immediately. It should also explain what the prototype helps that user do.

Add a role line that names your contribution to the prototype.
Create a tools line starting with Bubble.

Your card now separates your personal contribution from the product summary.

Add Windows Snipping Tool to the tools line.
Add Windows Notepad to the tools line.

The tools line should now name Bubble, Windows Snipping Tool, and Windows Notepad.

Add No-code prototyping as the first demonstrated skill.
Add API integration as the second demonstrated skill.

These first two skills connect the visible interface to the external service integration.

Add Responsible AI product judgment as the third demonstrated skill.
Add one trade-off line explaining why static evidence suited the exhausted testing allocation.

The card now shows technical delivery plus the judgment behind your evidence strategy.

Add an evidence line that points to the ordered screenshots.
Mention the architecture flow on the same evidence line.

A reviewer should now know where to find proof of the interface, workflow, integration, and system design.

Copy the existing prototype-status statement into the card exactly: Functional prototype built and tested. This portfolio entry uses static evidence because the original testing allocation was exhausted.
Save portfolio-case-study.txt.

You should now see a complete project card at the top of the file. Its status statement sets an honest boundary around the evidence.

Review the finished card for unsupported outcome claims.
Remove any claim that your evidence cannot prove.

A publishable card makes no claim about production adoption. It also avoids claims about accuracy, hiring outcomes, or unverified business results.

Arrange the evidence sequence

Evidence order controls what a reviewer understands first. Your sequence moves from the product interface into implementation proof.

The architecture flow closes the sequence by connecting the full system.

Return to the Evidence section in portfolio-case-study.txt.
Move 01-interface.png to the first evidence position.

The interface image should now lead the sequence. This gives the reviewer immediate product context.

Position the matching interface caption directly below its image reference.
Move 02-workflow.png below the interface caption.

The workflow image should now follow the interface. This shows how the user action reaches the analysis sequence.

Position the matching workflow caption directly below its image reference.
Move 03-api-connection.png below the workflow caption.

The API connection image should now appear third. This proves that the prototype connects to an external service.

Position the matching API connection caption directly below its image reference.
Move the existing architecture flow below the API connection caption.

The architecture flow should now close the evidence sequence. It connects the screenshots into one end-to-end product story.

Save portfolio-case-study.txt.
Read the evidence sequence from top to bottom.

You should see 01-interface.png first. 02-workflow.png should follow it.

You should then see 03-api-connection.png. The architecture flow should appear last.

Write and rehearse the interview walkthrough

The interview walkthrough turns the written case study into a spoken decision trail. A two-minute limit keeps the story focused on the details a reviewer can verify.

Add an Interview Walkthrough heading at the bottom of portfolio-case-study.txt.
Draft the opening around the recruiter problem.

Your walkthrough now has a clear user need as its starting point.

Follow the opening with the working product flow.
Add the hardest genuine product decision you made.

The first half should now move from the problem into the workflow. It should then reveal the judgment behind the prototype.

Add the static evidence that proves prototype completion.
Add the main limitation from your case study.

The middle of the walkthrough should now show what the evidence supports. It should also state the current product boundary.

End the walkthrough with your next validation step.
Save portfolio-case-study.txt.

Your complete walkthrough should cover the problem, workflow, hardest decision, evidence, limitation, and next validation step.

Start a two-minute timer.
Read the full walkthrough aloud.

Your delivery should finish within two minutes. Each section should sound like part of one product decision trail.

Trim repeated phrases if the timer ends before your last line.
Run one more timed rehearsal.

Your walkthrough is ready when it finishes within two minutes. It should still include all six required beats.

Return to the top of portfolio-case-study.txt in Notepad.
Open AI-Recruitment-Portfolio/evidence in File Explorer.

Your project card and evidence folder are now ready for the final scan.

Enlarge the evidence folder until all three image thumbnails are readable.
Arrange Notepad beside File Explorer.

You should now see the final project card beside the three ordered images.

Before the final scan, do you think a hiring reviewer can find every essential fact within 90 seconds?

Start a 90-second timer.
Scan the project card as if you are a hiring reviewer.

What Should the Scan Reveal?

You can identify the recruiter user.
You can identify the candidate-analysis problem.
You can explain the prototype solution.
You can identify your contribution.
You can identify the main trade-off.
You can locate the supporting evidence.
You can state the current prototype status.

That is the packaging complete. A reviewer can now find the prototype story within one short scan.

Secret mission

Add an AI Quality and Safety Rubric

Future candidate analysis needs a repeatable quality test before a recruiter can trust it. You will add an offline five-case rubric. It will assess four quality dimensions plus clear human-review triggers without running the Bubble workflow.

Clean Up Your Resources

Your portfolio pack is stored locally. It creates no ongoing costs.

Decide whether to keep the evidence pack for applications, pause your work, or delete the local workspace.

Resources you used:

A local AI-Recruitment-Portfolio workspace.
An AI-Recruitment-Portfolio/evidence folder containing three ordered portfolio-safe screenshots plus three raw captures.
A AI-Recruitment-Portfolio/portfolio-case-study.txt file containing your written portfolio package.

Keep everything running

No action needed. Choose this if you want to use the evidence pack for applications or interviews.

Retain the AI-Recruitment-Portfolio folder for future portfolio updates.
Keep your completed Bubble app available as the source prototype.
Use portfolio-case-study.txt as the guide for your two-minute interview walkthrough.

The static evidence remains ready without running the workflow. No new Gemini API call is needed.

Pause - I'll come back to this later

Pausing this project means closing the open tools while keeping every local file for later editing. No workflow or background process keeps running.

Close the Bubble editor window.
Close Snipping Tool if it is open.
Close Notepad if it is open.

Your AI-Recruitment-Portfolio folder stays on your device. You can continue editing it later without triggering the AI workflow.

Delete - I don't want to use this again

Deleting your work can feel drastic. This removes only the local evidence pack while leaving your completed Bubble app untouched.

Select the AI-Recruitment-Portfolio folder in the location where you saved it.
Remove the selected folder using the delete option in its context menu.
Leave your completed Bubble app in place unless you separately decide that you no longer need it.
Confirm the AI-Recruitment-Portfolio folder no longer appears in its saved location.

That clears the local evidence pack. Your completed Bubble prototype remains in place.

Nice Work!

You did it! You turned your completed Bubble AI recruitment prototype into an honest portfolio case file that proves your product judgment without claiming unverified results.

You've learned how to:

Created a recruiter-safe evidence trail that shows the interface before the workflow. It then shows the API Connector before the architecture flow.
Explained the recruiter problem through a concise case study. Documented responsible AI boundaries for human review. Kept unverified results out of the story.
Prepared a scannable project card for fast review. Rehearsed a two-minute interview walkthrough that honestly presents the Bubble prototype's Gemini API integration.
Completed the optional Secret Mission by designing an offline five-case quality and safety rubric. The rubric sets pass conditions for each quality dimension. It also records recruiter review triggers.

Ready to quiz yourself?