Inspiration
A surrogacy journey is shared by intended parents, a gestational carrier, coordinators, attorneys, clinics, escrow, and other professionals. Yet each person usually receives a different fragment: a contract, a clinic calendar, an escrow instruction, or a message thread. Those fragments can disagree, and a quiet date conflict can affect medication, transfer, pregnancy confirmation, money, and trust.
We wanted to build something broader than a company portal: a calm journey guide that can help everyone understand the same underlying journey without pretending everyone should see the same private information.
What it does
Wholeway turns fragmented records and commitments into one evidence-backed journey with five perspectives:
Whole journey: only facts intentionally shared with everyone. Intended parent: funding responsibility, private escrow evidence, and shared schedule impact. Gestational carrier: medication and appointment consequences without restricted financial wording, organization, IDs, or private dates. Coordinator: complete evidence, human authority, plan application, receipts, and Undo. Professional: clinical dependencies without restricted escrow language.
The demo centers on a real kind of coordination failure using entirely synthetic data. A signed agreement makes funding due May 4. Escrow instructions imply May 29. Funding is expected May 10.
GPT-5.6 surfaces the contradiction and computes what each reading means, but it never decides which legal or financial source governs. A coordinator records the confirmed human instruction first. At that point, the dates are still unchanged.
Only after the human decision is on the record can Wholeway apply the proposed plan:
Medication moves from May 14 to May 20. Embryo transfer moves from June 3 to June 9. Pregnancy confirmation moves from June 17 to June 23. Clinic confirmation remains explicitly required.
A visible causal graph connects the source to the funding deadline and every downstream date. Activity keeps separate receipts for the model-detected conflict, the human decision, the applied plan, and Undo. Undo restores the previous dates without deleting how the team arrived there.
Why this is different
Wholeway is not a generic chatbot and not another checklist. The surface is intentionally calm. Underneath is a role-projected, evidence-linked journey state machine with reversible graph memory.
Perspective switching is the product identity. The carrier can understand the shared consequence while the private escrow source disappears. Privacy is enforced through projection, not another notice and not blurred content.
The governing principle is simple:
AI detects. A person decides. The graph remembers.
How we built it
Wholeway is a responsive Next.js, React, and TypeScript application with:
one shared journey state projected through five role lenses; separate proposal, human-decision, application, and Undo states; source-level visibility rules; synthetic rendered agreement and escrow pages; a dependency graph linking evidence to three recalculated dates; local persistence so a refresh does not destroy the judge's position; a deterministic Restart path; responsive desktop and mobile workspace patterns; and provenance-tracked, realism-gated editorial assets.
The demo uses synthetic names, dates, organizations, record IDs, source pages, and journey facts. It contains no customer, patient, payment, contract, or medical data.
How Codex and GPT-5.6 were used
We used GPT-5.6 through Codex as a product, design, engineering, and QA collaborator.
Codex helped us step outside the original company workflow and model the journey from every participant's perspective. It converted those rules into a responsive state machine, implemented privacy-safe projections, created synthetic fixtures, exposed inconsistencies between the proposal and the applied receipt, integrated realism-gated assets, and exercised every perspective across every workspace.
GPT-5.6 was especially useful for reasoning about contradictions and downstream consequences. It helped identify that an earlier prototype moved a pregnancy-confirmation date without showing that change in the receipt—a trust-breaking inconsistency. The final product now shows the complete source → funding → medication → transfer → pregnancy confirmation → clinic confirmation chain.
The submitted interface presents modeled GPT-5.6 reasoning output; it does not pretend there is a production API or autonomous legal/medical decision maker behind the prototype.
Build Week scope
Patriot Conceptions supplied domain experience about surrogacy operations. Wholeway's universal product identity, five-perspective projection model, source contradiction, human-governance flow, three-date replan, causal graph, activity receipts, Undo model, responsive Next.js implementation, synthetic demo journey, visual system, and role-route QA were created or meaningfully extended during Build Week with Codex and GPT-5.6.
The repository README documents that boundary, setup, sample data, the recommended judge path, and the complete Codex/GPT-5.6 build story.
Challenges
The hardest problem was not generating another dashboard. It was keeping one journey internally consistent while changing what each person can see and do.
We found and fixed several subtle trust failures:
the first receipt omitted a real downstream pregnancy-confirmation change; Journey still said "Needs alignment" after the human decision was recorded; a post-Undo replay could have reused the same version identifier; non-coordinator perspectives were being asked to take coordinator-only action; and visible source controls did not originally open the underlying evidence.
Those details matter because a journey product earns trust through its exceptions, not its happy path.
Accomplishments
A complete perspective-safe contradiction flow. A human decision that is visibly separate from state mutation. One confirmed fact driving three inspectable downstream changes. Reversible state without erased history. Thirty verified perspective-workspace combinations with zero restricted-source leaks. A production build that works responsively with realistic synthetic content.
What we learned
The best AI experience in a high-stakes journey is often not chat. It is a calm system that makes provenance, authority, uncertainty, and consequences understandable.
We also learned that privacy can be part of the interaction model itself. The originality is not merely hiding fields; it is letting each person understand the same truth from an appropriate place in the journey.
What's next
The next step is closing the clinic-confirmation loop: a named clinic perspective that can confirm or counter-propose the recalculated medical calendar, producing the next immutable journey version.
After that, we would split Professional into clinic, counsel, escrow, agency, and counselor projections; add authenticated backend storage; support source ingestion; and provide role-specific acknowledgement and handoff actions.
Wholeway's promise stays the same:
One surrogacy journey, understood safely from every perspective.
Built With
- gpt-5.6
- magnific
- next.js
- openai-codex
- phosphor-icons
- react
- typescript