FP
FUNDPATH
FundPath
AI searches and explains.
Families decide everything involving money.
Arjith · AI Builders Hackathon 2026
PROBLEM STATEMENT
Free money goes unclaimed.
Found money goes unmanaged.
$4.4B
in U.S. Pell Grant aid went unclaimed in 2024 because 47% of eligible high schoolers never completed the FAFSA. The same discovery-and-paperwork friction FundPath targets. (EducationData.org, 2026)
01
Discovery Is Scattered
Scholarship and grant listings are spread across sites full of jargon students can't easily parse eligible students never apply.
02
Awards Sit Idle
Money received usually just sits in an account or gets spent, with no guidance suited to a small, one-time sum.
03
Trust Is Fragile
Families are wary of anything that nudges them toward investment decisions they didn't ask for so most tools get ignored.
FP
SOLUTION OVERVIEW
FundPath
AI searches and explains.
Humans decide everything involving money.
✕
Never
auto-submits an application
✕
Never
auto-allocates an award
✕
Never
auto-invests a dollar
The Loop
Search, explain,
decide, repeat.
1
Search & Match
AI ranks scholarships with pros vs. cons.
2
Family Decides & Confirms
✓
An explicit tap-to-confirm gate nothing is applied to without it.
3
Allocate
Family sets the loan / buffer / invest split.
4
Investment Comparison (Optional)
Only if chosen, no default nudge.
FP
TARGET USERS
Who FundPath is built for
Students and families managing the real cost of education who want help finding money, not someone else making decisions about it.
S
Students
Actively searching for aid
Ready to apply but short on time to compare options
Want AI help, not AI control
P
Parents & Guardians
Co-deciding on education money
Want full visibility into options not automation
Need to trust the platform before engaging
F
First-Gen & New Investors
Curious about growing a scholarship award
Wary of being pushed into unfamiliar products
Need clear risk & liquidity info, then choice
FP
PRODUCT FEATURES
Six features, one continuous flow
AI does the searching and comparing in steps 1–3. From step 4 on, the family is in the driver's seat.
1
Profile & Snapshot
Academic info, income, loans, field of study
2
Scholarship Search & Compare
AI ranks matches with pros & cons for each
3
Application Drafting
Pre-filled forms + first-pass essay, on request only
✓
HUMAN CONFIRMS
before control passes from AI to family
4
Family-Controlled Allocation
Family sets their own loan / buffer / invest split
5
Investment Opt-In & Compare
Small-ticket options shown with pros & cons
6
Decision Log & Dashboard
What the AI suggested vs. what the family chose
FP
TECHNICAL ARCHITECTURE
An agent that calls approved tools but never financial execution tools
AGENT
FundPath Orchestrator
Observes profile + phase → selects ONE tool → acts → logs.
Deterministic rule engine.
LLM-based reasoning is the Phase 2 upgrade (see Roadmap).
score_eligibility()
Ranks scholarships by fit
compute_allocation()
Applies family's own split
filter_investments()
Matches options to amount
log_decision()
Records AI vs. family choice
Design constraint: no submit(), allocate(), or invest() tool exists in the codebase the agent can only recommend, and a human confirmation gate sits before every tool call with financial effect.
Data
Curated dataset (demo)
→ Licensed feed (prod)
Storage
JSON files (demo)
→ Database (prod)
Reasoning
Rule engine (now)
→ LLM agent (Phase 2)
Interface
Vanilla JS dashboard
FP
AI TECHNOLOGIES USED
Rules decide. AI compares. Humans confirm.
RULE ENGINE
Eligibility scoring
Deterministic checks against GPA, income, field, location same input always gives same output.
Investment filtering
Compares allocated amount against minimums, lock-in, and risk tier.
AI REASONING
Comparative pros/cons
Generates in plain language advantages/disadvantages for each option so families can weigh trade-offs.
Plain-language explanations
Turns rule-engine output (scores, filters) into readable reasoning phase 2 swaps this to a live LLM call.
HUMAN DECISION
Apply / skip each scholarship
Nothing is submitted without an explicit family choice, logged either way.
Allocation split & invest opt-in
Family sets every percentage and can decline investing entirely.
FP
IMPACT & VALUE PROPOSITION
Trust, made visible.
20/20
Eligibility Checks Verified
4 synthetic profiles × 5 scholarships rule-engine output manually audited against each eligibility rule.
<0.01ms
Avg. Match Latency
10,000-run local benchmark of the rule engine across the demo dataset: 0.005ms average per profile match.
0
Auto-Decisions
No submit(), allocate(), or invest() tool exists in the codebase verified by inspection, not just policy.
Evaluated on the hackathon demo dataset (5 scholarships, 5 investment options) not a production-scale benchmark.
Sample Decision Log
Decision Point
AI Suggested
Family Chose
Scholarship shortlist
Ranked “Regional Merit Grant” highest (₹30K)
Applied to Regional Merit + one other
Allocation split
Family enters, not suggested
₹15K loan payoff · ₹10K buffer · ₹5K invest
Investment choice
Recurring Deposit (best liquidity/risk fit)
Chose Recurring Deposit (RD)
FP
FUTURE ROADMAP
From Hackathon to Production
Phase 1
Now (Hackathon)
Curated scholarship & investment datasets
Rule-based matching engine
Vanilla JS demo dashboard
Full decision log
Phase 2
Post Hackathon
Phase 3
Scale
Phase 4
Ecosystem
AI explanation audit trail
Live scholarship data feed
Real-time market-data API
Persisted user sessions
Mobile-responsive UI
Multi-region scholarship data
Family collaboration mode
Personalized essay memory
Observability & audit logs
Financial-aid office integrations
Fintech partner APIs
Regulatory compliance layer
FP
FP
FundPath
AI searches and explains.
Families decide everything involving money.
Repo
github.com/your-team/fundpath
Submitted to
AI Builders Hackathon 2026