Understand your degree. See the risks. Plan your next move.
Making the risks hidden in a degree plan understandable and actionable.
Public source migration pendingA degree audit can show what is missing without explaining what to do next. I designed Pathwise around the decision that follows: identify the risk, understand the rule behind it, compare possible plans, and arrive at advising with a clear next step. The product brings academic records, requirements, dependencies, risk detection, scenarios and AI-assisted explanations into one planning experience.
Rico confirms that the product capabilities are built in his private repository. The implementation account is owner-reported; public source and individual screens have not yet been independently reviewed. No measured Pathwise evaluation results have been provided.
Start with a limited set of undergraduate programs and explicit catalog versions. Model applicability and exceptions carefully before expanding coverage. The portfolio examples use fictional courses; they do not assert actual UT Dallas requirements.
External student-success research supports the usefulness of degree-planning tools. My secondary-research synthesis highlights last-minute requirement surprises, confusing audits, prerequisite delays, transfer-credit ambiguity and uncertain course availability.
I used these themes to frame a product opportunity: make academic risk understandable before registration decisions become costly. The qualitative priorities are exploratory, not a representative ranking. The external survey percentages below are not original Pathwise findings.
Very or extremely useful among surveyed tool users. EDUCAUSE, 2019.
Very or extremely useful among surveyed tool users. EDUCAUSE, 2019.
Of 5,025 respondents in the 2024 Student Voice survey.
A helpful warning needs three things: what could happen, why it could happen, and what the student can do next.
Separate credits accumulated from applicable requirements satisfied. Expose course allocation, permitted overlaps, minimum grades and unresolved equivalencies.
An accepted transfer credit may still need confirmation before it satisfies a major requirement.
Connect a risk to its rule, source, applicable catalog and affected course sequence. Compare alternative plans with assumptions visible.
Historical offerings are a planning signal; they cannot establish future course availability.
Use optional AI to explain deterministic findings, then turn uncertainty into advisor questions. Let students review imported data, correct it and control what they share.
Neither an AI explanation nor a feasible scenario is an official academic determination.
Prioritize trustworthy planning, then the workflows that explain it.
| Product capability | Priority | Decision rationale |
|---|---|---|
| Requirement tracking | Core | Establish which rules apply and how completed courses are allocated. |
| Prerequisite engine | Core | Identify the dependency that could block a later course. |
| Risk dashboard | Core | Make the next consequential decision visible immediately. |
| What-if simulator | Core | Compare alternatives against workload and sequencing constraints. |
| Source-grounded AI & advisor briefs | High | Turn rule output into an understandable explanation and a reviewable next step. |
| Weekly scheduling | Secondary | Complement degree planning where supported scheduling data is available. |
| Live seat alerts | Deferred | Depend on a reliable live feed and recurring checks. |
| Automatic registration | Excluded | Requires institutional authorization beyond the independent product scope. |
Explore the product thinking through a small, interactive illustration. The example content below is fictional and is not a screenshot or a live product integration.
Surface the most consequential uncertainty before presenting a reassuring completion score.
ANA 201 must precede CAP 301.
A synthetic student scenario connecting the complete product experience. It is not a completed-pilot result.
A fictional student targets Fall 2027, sets a 12-credit term limit and reviews an imported audit with an unresolved transfer equivalency.
A planning profile with visible unknownsA prerequisite is scheduled after the term needed to unlock an advanced course. The warning connects the dependency to its source.
A specific sequence to reviewKeep the current sequence, move the prerequisite earlier, or consider summer. Each option exposes workload and availability assumptions.
Conditional plans, side by sideThe assistant explains the rule-engine finding and distinguishes a confirmed requirement from an unverified future course offering.
An explanation the student can inspectBring a preferred plan, source references and explicit questions about transfer credit, course eligibility and future availability.
A concrete next action for advisor reviewProgram, catalog year, course history and workload constraints. Manual entry, CSV or supported text-based PDF import, followed by student review.
Prioritized warnings, requirement status, target term and conditional scenario results. Credits accumulated remain distinct from requirements satisfied.
Source-linked rules, catalog applicability, course allocation and unresolved transfer equivalencies. Conflicting interpretations remain visible for review.
Course sequences, bottlenecks, prerequisite alternatives, grade thresholds, concurrent enrollment conditions and recorded exceptions.
Term-by-term course movement, workload checks and plan versions. Historical offerings are distinguished from confirmed future schedules.
Compare prerequisite timing, reduced loads, additional requirements and unavailable courses with explicit assumptions and unresolved conditions.
An optional assistant explains rule-engine findings using planning context and versioned catalog evidence, with abstention when support is missing.
Summarize risks, plan comparisons, sources and open questions in a portable brief without requiring an advisor account.
Surface changes across versions while separating a published policy change from confirmed applicability to a particular student.
Local data control, correction, export and deletion; sanitized sharing keeps selected plans separate from grades and student identifiers.
Manual entry, CSV and supported text-PDF import feed a student review step. Provenance and unresolved equivalencies stay visible.
Versioned rules describe requirements, prerequisite alternatives, grade thresholds, credit allocation and recorded exceptions.
Graph relationships, term loads and availability assumptions shape plan comparisons. Incomplete or contradictory rules must remain explicit.
Keyword search and optional local embeddings retrieve relevant catalog passages alongside planning facts and rule-engine output.
An on-device assistant explains supported findings, cites evidence and abstains when a conclusion cannot be verified. Core planning has a non-generative fallback.
This architecture reflects Rico’s product document and confirmation of a completed private build. It has not been independently verified against source code. The portfolio walkthrough is a separate fictional illustration, not the Pathwise application. Official requirement applicability and institutional approvals remain outside the model’s authority.
Implementation details described by Rico; public source verification pending.
The design separates TypeScript rules and scenario calculations from retrieval and optional generative explanations. Model availability should not prevent a student from inspecting a rule or comparing a plan.
WebLLM and Transformers.js are part of the architecture Rico describes. Download size, WebGPU support, device memory and latency make a source-linked, non-generative fallback essential.
IndexedDB keeps planning data on the device. Browser clearing, quotas or device loss still create retention risks, so portable exports and deletion controls belong in the experience.
Static hosting, browser processing and optional local inference target $0 in direct operating charges within provider and model-license limits. This does not imply unlimited scale or free hardware, bandwidth and development time.
DOCUMENTED STACKReact · TypeScript · Vite · Tailwind · Lucide · graph utilities · versioned JSON · IndexedDB · PDF.js · Transformers.js · WebLLM
The evaluation framework covers precision, recall, false all-clear errors and scenario validity using fictional profiles and reviewed rules. Prerequisites, catalog mismatches, transfer uncertainty, workload limits and contradictory inputs belong in the test set.
Evaluate retrieval, source accuracy, catalog-version consistency and unsupported conclusions separately from rule-engine correctness. Include misleading document content and adversarial questions. No live-model accuracy result is reported here.
Compare realistic planning tasks against catalogs, degree audits and spreadsheets. Measure time to identify a blocker, correct interpretation, valid draft-plan creation, scenario completion, abandonment and advisor-brief usefulness.
Assess whether confidence matches correctness, then examine repeat planning, scenario creation, brief exports and resolution of flagged concerns. The supplied document describes the evaluation approach but provides no measured Pathwise results.
Percentage of evaluated planning sessions in which a validated, actionable risk is correctly identified before the relevant academic decision.
A measurement definition, with no reported value. Report risk prevalence alongside it so risk-free sessions are not interpreted as product failures.
Target zero definitive all-clears in the curated release suite when a blocking requirement remains unresolved. This is a target, not a reported result or a real-world guarantee.
For a high-stakes planning tool, admitting uncertainty is a product capability. A convincing answer is less useful than a traceable one.
Publish the implementation and source-backed requirement fixtures, then attach reproducible evaluation results and a product walkthrough. Keep the roadmap focused on valid, understandable decisions as supported program coverage expands.
Reviewed November 10, 2026. Repository documentation is attributed to its author; implementation claims are distinguished from research findings and planned capabilities.
EDUCAUSE 2019 student study (p. 13) (opens in a new tab)Inside Higher Ed — Student Voice survey, July 2024 (opens in a new tab)Rico supplied the full Pathwise product account and confirmed that the described capabilities are built privately. Source migration is pending; private code and screens have not been independently reviewed. Evaluation methods have been supplied without measured results. External survey findings are not original Pathwise research.
Download the content verification record