Skip to content

Submit Vote Certificates

Community favorite

Vote for the project you'd actually use

One vote per email. You can't vote for your own team.

Ballot

Pick a team

Website

Your name *

Your email *

One vote per email. You can't vote for your own team.

Pick a team *

Community favorite — the project you'd actually use.

AutomationInterns.comrecordingalert.comHelp rich ppl make more moneyContourContourContour tracks what a public company said, filed, and changed — then surfaces where the three disagree. An initial scan builds a baseline from primary sources (SEC EDGAR, NHTSA, declared feeds); later scans report the delta, not the document. Every figure cites the accession it came from; no model decides what is true. Three layers of checks run in order of judgment involved: six fixed checks for any registrant, a 25-item sector catalogue, and model-authored specifications executed deterministically. A core constraint shapes the design: "we found nothing," "this doesn't apply," and "we couldn't check" never look the same.DISSENTDISSENT - Concern LoopWe built Concern Loop, a system where every workplace concern gets independently researched — one AI track tries to refute it, a second checks it from an outside angle, and an arbiter cross-validates the two by evidence, never by title — before a ranked, evidence-backed report reaches a human who must issue a real verdict, with safeguards so it can't be stalled forever or quietly buried. It matters because every existing tool in this space (NAVEX, AllVoices, Vault) just moves a concern from an inbox to a case file without ever checking if it's true, and that gap is exactly how real disasters happen: Challenger, Pentium FDIV, and the OceanGate Titan all had a written internal warning that reached exactly one desk and died there — Concern Loop is built to be the second opinion that stops that from happening again.Exceptional HandlersRioGuardian AIAn AI-powered student support navigator starting at Rio Salado College, designed to scale for Phoenix, America, and the world. It understands intent, returns verified information, and learns from unanswered questions.GhostlessUntitled projectPitch DeckUntitled projectPrescribeSafePatient-Specific Medication Risk CopilotPrescribeSafe is a clinician-facing medication safety copilot that helps identify patient-specific risks before a medication is prescribed. A clinician can enter a patient’s conditions, allergies, current medications, and a proposed drug in natural language. The system converts that information into structured clinical context, then uses deterministic, source-backed rules to check for contraindications, allergy risks, and drug-drug interactions. Medication facts are grounded in real public sources such as FDA drug labels and RxNorm, while the patient scenarios used for the demo are synthetic for privacy. The AI layer is used to understand and explain clinical context—not to invent safety facts. Risk findings come from the structured medication data and rules, and the system returns potential risk to that patient. PrescribeSafe matters because medication safety alerts are most useful when they are patient-specific, explainable, and grounded in evidence rather than generated freely by an LLM.ScoutScoutScout matches people to teams on three layers, mission alignment, team personality fit, and AI-era skills, instead of one opaque score, and explains every match in writing. It starts with a 12-question assessment that uses everyday framing rather than work scenarios, so answers aren't filtered through "what would a manager want to hear." It matters because the platforms companies hire on get paid when hiring fails: turnover is just another placement fee. Scout charges for people management instead, so a match that doesn't hold costs us money. Recruiting is the wedge. People management is the business.Team CampusProject HarnessProject Harness is a local control center that sits between you and the AI agents working on your projects — your whole portfolio, running at once. Give an agent unrestricted control and it ignores your rules, skips the checks that matter, and overwrites decisions you already made. Multiply that by every project you have running. Project Harness separates AI reasoning from project control. You define the rules; the agents reason inside them and the harness enforces them. A sub-agent that declares itself done with your checks unmet is refused. And when a decision is genuinely yours, that project stops and asks you — while everything else keeps moving.Tenant Talk TeamUntitled projectThe Travel CoupleUntitled projectUnstuckUnstuckWhen a 5th grader fails fractions, the real gap is usually a 3rd-grade skill nobody caught. UnStuck diagnoses across a Common Core prerequisite graph spanning grades 1–6, then generates practice at the student's true level — themed around their own interests, and shaped by notes from their teacher and parent. The AI writes the questions; a deterministic math engine grades them and verifies every item before a student sees it. No hallucinated answer keys. Open source, MIT.ValenceUntitled projectVerdeanUntitled projectVibe TenderingTenderMindTenderMind is an AI-powered tender intelligence platform that helps construction companies decide whether they should bid, what financial exposure they are taking on, and what risks are hidden inside hundreds of pages of tender documents.

Workuity Chandler, 3133 W Frye Rd, Suite 101, Chandler, AZ 85226 · Arizona AI Agent Hackathon