Active · exploratory · single participant
Case Study 001
An ongoing record of one person building Think Smarter AI with AI assistance while trying to preserve human judgement and develop practical capability.
Important Limitation
This is not a controlled experiment and does not establish general effects.
Current question
Can AI support the build without replacing the learning?
The project is observing how one founder can use AI to help design, build, manage and improve an educational website while testing whether the process strengthens his own knowledge, judgement and practical capability.
Evidence design
What Is Being Recorded
Useful evidence includes mistakes, reliance on assistance and completed work, not just polished outcomes.
01 · Before
Starting Understanding
What Jacob already knows, what he is trying to learn and which parts are uncertain.
02 · During
Decisions and Disagreements
Where AI advice was accepted, rejected or changed after evidence and human judgement.
03 · After
Checks, Errors and Repeatability
What worked, what needed rework and whether the task can later be explained or repeated.
Verified milestones
What the Record Currently Supports
These milestones demonstrate completed website work and traceable versions. They do not by themselves demonstrate learning gains or wider impact.
18 July 2026
The Project Began
Jacob W. founded Think Smarter AI around the principle that AI should improve human thinking, not replace it.
29 July 2026
The Live Site Moved from Carrd to Coded Source
The change used reviewed source, preview deployment, human approval and a preserved rollback path.
30 July 2026
Website v0.3 Became the Human Signal Release
The design system, Founder page and public explanation of the human–AI workflow were released and checked.
30 July 2026
The First Interactive Practice Activity Was Prepared
The Thinking Check translated the mission into five practical scenarios with no answer submission or account required.
By 2 August 2026
Two Reviewed Learning Releases Were Live
Website v0.7 added AI project-management guidance and a local Project Triage activity. Website v0.8 added the Learning Hub and a fictional, local-only ten-decision Claim Check Lab, expanding the sitemap to 15 public routes. These releases provide traceable implementation evidence, not proof of learning gains or wider impact.
Working-method lessons · August 2026
What the Later Work Changed
These are evidence-supported changes to the project's working method. They are not proof of personal learning gain, reconstructed starting knowledge or wider educational impact.
3–20 August 2026
Accounts Made Ownership and Privacy Part of the Product
Anonymous learning, private account history and optional research sharing had to remain separate. On 11 August, Jacob directly checked production Google sign-in, full account deletion, account recreation, result saving, history review and CSV export. On 20 August, he reported completing individual-result deletion, ordinary sign-out, sign-back-in, retained/deleted-state confirmation and final account deletion as one continuous live journey. That sequence is user-verified rather than independently observed. Isolation using two genuinely separate production accounts and Back/Forward Cache acceptance remain unverified.
2–17 August 2026
Young Thinkers Changed Status When the Evidence Changed
The public boundary moved from Adult Review, to a supervised preview, and then to Available after Jacob confirmed review by multiple adults. That owner-reported clearance does not establish a learner age range, formal child testing, independent safeguarding validation, curriculum alignment or learning outcomes.
15–17 August 2026
AI News Separated Announcements from Independent Proof
An official source establishes what an organisation or regulator announced. It does not independently prove every benchmark, safety claim or outcome. When an automated reader received a 429 response, the record treated the source as not checked in that run—not broken or false—and added directly relevant official fallbacks.
17 August 2026
v0.13 Prioritised Coherence Before More Features
The release aligned evidence wording, added shared Human Thinking Loop orientation, clarified homepage routes and made detailed history easier to inspect. Reviewed source and tests establish that implementation. They do not establish improved visitor comprehension, usability or learning. Major feature development is now temporarily paused while unprompted human usability sessions are observed and reviewed.
Current interpretation
Known, Unknown and Still Being Tested
Uncertainty is part of the record, not a gap to hide.
Supported now
Completed and Traceable Website Work
Earlier versions, reviewed releases and current public pages provide direct evidence of the website's development.
Unverified
Independent Repeatability and Learning Gain
The project has not yet demonstrated that Jacob can repeat key work with less assistance or explain every technical decision independently.
Unknown
Time, Cost and Wider Impact
Exact time saved, exact project-attributed AI cost, visitor learning outcomes and general applicability are not currently established.
Limitations
One Project Cannot Answer Every Question
The participant is also the founder, the work is self-observed, there is no control group and the project is still very young. Future claims must remain proportionate to the evidence.