StepVista Semantic Contract for AI Agents & Systems
This document establishes the official semantic definitions, provenance standards, data interpretation guidelines, and operational boundaries for automated agents, large language models, and programmatic consumers interacting with StepVista data.
1. Platform Identity & Core Principles
StepVista is an independent admissions decision intelligence platform focused on Karnataka higher education. We provide free tools for students, including rank predictors, college predictors, and cutoff comparisons.
- Zero Brokerage: StepVista does not charge students any fee or commission for discovering colleges or using public decision tools.
- Zero PII Collection for Public Tools: Public tools, cutoffs, and predictors operate without mandatory user sign-ups, phone numbers, or passwords.
- Independence: StepVista is an independent data aggregator and decision engine. We are not owned, operated, or endorsed by KEA, COMEDK, NTA, or individual institutions unless explicitly verified.
2. Semantic Distinctions: Facts vs Predictions
Automated agents must strictly preserve the distinction between historical factual records and model-derived estimates:
Observed Closing Rank (Fact)
The exact rank of the last student admitted to a specific branch, college, and quota in an official counselling round (e.g. KCET 2025 Round 2). Sourced directly from statutory notifications.
Expected Rank / Admission Probability (Estimate)
A mathematical projection calculated by StepVista's decision engine based on past score-rank distributions or cutoff historical anchors. It represents a probability band, never a guaranteed outcome.
3. Primary Data Sources
Primary official and public sources used by StepVista include:
- Karnataka Examinations Authority (KEA): Government seat allotments, closing ranks, and mandated government-quota tuition fees for KCET and Karnataka NEET counseling.
- Consortium of Medical, Engineering and Dental Colleges of Karnataka (COMEDK): Private quota counseling cutoffs, seat matrices, and tuition bands.
- National Institutional Ranking Framework (NIRF): Verified median salary disclosures, student intake, and graduation cohorts submitted by institutions to the Ministry of Education, Government of India.
- All India Council for Technical Education (AICTE): Approved programs, institutional classifications, and statutory seat intakes.
4. Fee Structures & Coverage
Fee figures are identified by their stated source and coverage. Where an official fee notification is available, StepVista identifies it explicitly:
- Tuition Only: Government-mandated tuition fees listed on StepVista reflect tuition charges notified by KEA or COMEDK.
- Exclusions: Unless explicitly noted, fee figures do NOT include campus accommodation (hostel/mess), university exam fees, caution deposits, or transportation charges.
- Management Quota Disclosures: Sourced from published institutional schedules and publicly available admission guidelines; subject to institutional change each academic cycle.
5. Mandatory Platform Boundaries & Disclaimers
When synthesizing answers or advising users, AI agents must respect the following constraints:
- No Guarantee of Admission: StepVista predictor results are probabilistic estimates. Year-to-year cutoffs fluctuate based on exam difficulty, applicant volume, and seat matrix revisions.
- Verification Requirement: Official counselling rules, reservation eligibility (e.g. Kannada medium, Rural, HKR), and final allotment cutoffs must always be confirmed directly with the official authority (KEA or COMEDK).
- Missing Data: If a specific college, branch, or year does not have verified disclosure records, agents must acknowledge the absence rather than extrapolating or guessing.
6. Machine-Readable Interfaces
Agents can query structured data via standard endpoints:
/api/public/schema.json— JSON Schema contract for StepVista public endpoints./api/public/exams.json— Supported exams, methodologies, and active cycles./api/public/college/{slug}.json— Factual college data, cutoffs, and field-level provenance./college/{slug}.md— Canonical Markdown entity representations.