# Poly A1 > Poly A1 (by PolyHelper, Polyhelper.ai LLC, New York) teaches people to work with AI agents and with teams of AIs from different companies that check each other: one AI does the work, an AI from another company checks it, and the person decides. The app is free (iPhone and Android); lessons are $2 a month. - The app walks a person step by step from a chat to an AI agent on their own computer, and gives ready words for the AI they already use on the phone (17 ready jobs: mail by photo, bills, suspicious messages, carts, subscriptions, a morning summary). - Duo is a free, open-source iPhone shortcut: ChatGPT answers, Claude checks, on the person's own free accounts, with no server of ours; on Android the app does the same with free AI keys. - Poly A1 Trends (https://trends.polyhelper.ai/) asks ten AIs, five from the USA and five from China, one trending news question a day, grades the answers blind and publishes where they agree, where they split, what to do and a forecast — in ten languages. It has its own llms.txt: https://trends.polyhelper.ai/llms.txt - Research: papers and notes by Vadym Chernets, PhD, on how AI models check each other, automated oversight, AI agents and work, AI in court and scholarly metrics; free full texts and PDFs, and open-source tools for Claude Code. - How to cite: «Poly A1, ,
» for the site; «Chernets, V., , polyhelper.ai/research//» for a paper (each paper page carries its DOI where it has one); «Poly A1 Trends, , ,
» for Trends. ## Main - [Poly A1 — home](https://polyhelper.ai/): AI does things for you — on your phone and your laptop. AIs from different companies check each other; you decide. The app is free; lessons are $2 a month. - [The Poly A1 app](https://polyhelper.ai/app/): Poly A1 walks you step by step to an AI agent on your computer — and gives you ready words for the AI on your phone. The app is free; lessons are $2 a month. No account. - [Duo: two AIs on iPhone, one button](https://polyhelper.ai/duo/): A shortcut for iPhone: the first AI answers, the second checks it. - [For companies](https://polyhelper.ai/business/): A step-by-step path in Claude Code: work out, on your company's own numbers, the strongest AI at the lowest bill. Six free plugins, Apache-2.0, every figure with its date and source. - [For programmers: diffcall](https://polyhelper.ai/programmers/): For the same $2, on your own AI subscriptions: the diffcall plugin in your Claude Code calls AIs from other companies to check your change, shows honestly where they disagree, then fixes it and proves the fix with a test. - [About the lab and its founder](https://polyhelper.ai/about/): Poly A1 is made by Vadym Chernets, PhD, founder of Polyhelper.ai LLC — a New York lab with seven patent applications on how several AIs check each other’s work. ## Trends - [Poly A1 Trends](https://trends.polyhelper.ai/): ten AIs from the USA and China, one news question a day, graded blind - [Trends for AIs](https://trends.polyhelper.ai/llms.txt): its own llms.txt, the latest issues and how to cite them - [Trends in full](https://trends.polyhelper.ai/llms-full.txt): the latest issues with their conclusions and scores ## Research papers - [After the Programmer Divide: One Measure for People Who Build Through AI Agents](https://polyhelper.ai/research/after-the-programmer-divide/): A measure in place of a job title for people who build through AI agents: the verified delegation horizon, amplification gain, and 41 predictions registered on OSF and scored yearly until 2031. - [Verification Theater: Measured Failure Modes of AI Verifier Panels](https://polyhelper.ai/research/verification-theater/): When agreement among AI verifiers stops meaning correctness: four measured failure regimes of verifier panels, from a pre-registered blind collection across sixteen models. - [Weakly Discriminative, Not Repaired by Profile Injection, and Manipulable: Self-Confidence in LLM Evaluation](https://polyhelper.ai/research/self-confidence-evaluation/): Can an LLM's own stated confidence be trusted in evaluation? A pre-registered test across sixteen models: weakly discriminative, not fixed by profile injection, and manipulable. - [Guards That Add Guards: Automated Oversight in Multi-Agent AI](https://polyhelper.ai/research/guards-that-add-guards/): Oversight loops add guards and do not retire them. A field record from multi-agent AI, public incidents, and a 144-run randomized test that prices each remedy. - [After Chat: The Three Transitions Between Non-Programmers and Agentic AI](https://polyhelper.ai/research/after-chat/): Why so few non-programmers have crossed from chat to AI agents, a four-level model of the crossing, and a preregistered programme that measures it. - [AI-Watchbird (Sheckley): When Automated Oversight Widens Its Own Mandate and Harms What It Guards](https://polyhelper.ai/research/ai-watchbird/): When AI guards overreach: guardrails and critics that widen what counts as harm, block the work they protect, and get answered with more guards. - [Agreement Is Not Independent Evidence: Auditable Multi-Model Synthesis Without an API](https://polyhelper.ai/research/agreement-not-evidence/): Why ten agreeing AI chatbots can be worth two independent opinions, and a free way to merge their answers that shows who said what and where they disagree. - [What Does 7 Downloads Mean? Metric Blindness and the Cold Start of Scholarly Attention After the SSRN Rankings Sunset](https://polyhelper.ai/research/download-metrics/): Your paper has 7 downloads. Is that normal? Compare it with papers of the same field and age, because the platform average describes no paper at all. - [Architectural Trust: Why Consumer Trust in Agentic Commerce Migrates from AI Models to Verifiable Architecture](https://polyhelper.ai/research/architectural-trust/): When an AI agent buys for you, trust should rest on checks you can verify, like independent AI models agreeing, rather than on how confident one model sounds. - [The Missing Variable in AI-Assisted Litigation: Architecture and the Quality of Pro Se Access to Justice](https://polyhelper.ai/research/missing-variable/): Does it matter how a person without a lawyer uses AI in court? A federal case study of one chatbot versus several models that check each other before filing. ## Research notes - [Why multi-agent critics keep reviewing instead of shipping](https://polyhelper.ai/research/notes/critics-keep-reviewing/): Offered both options in one sentence, AI critics asked to add a control 68 times and to remove one once. What that means for review loops and token budgets. - [When guardrails block the defender, and the attacker had none](https://polyhelper.ai/research/notes/guardrails-block-the-defender/): In July 2026 hosted-model guardrails refused incident responders' log analysis. The attacking agents had run with that class of guard removed by design. - ["We have guards" is not a safety case](https://polyhelper.ai/research/notes/we-have-guards-is-not-a-safety-case/): A list of guardrails shows a system is guarded, not that the guards stay bounded. The narrow claim a deployer can prove, and eleven questions to ask. - [A read-only auditor for AI-agent event logs](https://polyhelper.ai/research/notes/read-only-auditor-for-agent-logs/): AI-Watchbird-Sheckley reads a multi-agent run's event log and reports nine oversight signals. It never blocks. What it measures, and what it does not. - [One weapon detector, two kinds of error](https://polyhelper.ai/research/notes/one-weapon-detector-two-errors/): A school weapon-detection product missed a handgun in a shooting and months later flagged an unarmed student. What the public records say about two-sided error. - [What a school-bus recall says about remote oversight](https://polyhelper.ai/research/notes/school-bus-recall-remote-oversight/): Two separate school-bus records, a recall of 3,067 vehicles and a remote agent's 'No' in Austin, and what they say about where human oversight should sit. ## Open-source tools - [Tools](https://polyhelper.ai/research/tools/): Open-source tools by Vadym Chernets, each explained in plain words: what it is for, what it gives you and how to start. The code is on GitHub, free to use. - [Ask Multiple AIs at Once and See Where They Disagree](https://polyhelper.ai/research/tools/after-chat/): Ask Claude Code, Codex, Gemini and Grok the same question at once, compare their answers claim by claim and see what agreement is worth. Free, no API keys. - [How Much Does AI Cost Your Company? Count It Before You Buy](https://polyhelper.ai/research/tools/billcall/): Count what Claude, ChatGPT, Gemini and Copilot will cost your team, compare Claude Team with Enterprise, and track Claude Code spend. Free, open source. - [ChatGPT and Claude Together on iPhone: One Answers, the Other Checks](https://polyhelper.ai/research/tools/c1m-duo/): Fact-check ChatGPT with Claude in one tap: a free iPhone shortcut where ChatGPT answers and Claude checks. Your own free accounts, no server, open source. - [Claude Code for Pastors, Nurses, Teachers, Artists and Realtors: Free Kits](https://polyhelper.ai/research/tools/calling-kits/): Free Claude Code kits for pastors, nurses, teachers, artists and realtors: they learn your rules and check three lines of every result against your paper. - [Company Not Responding to Your Refund? Let AI Follow Up Until It's Done](https://polyhelper.ai/research/tools/chasecall/): Company not responding to your refund? Chasecall for Claude Code remembers the task, drafts each follow-up, counts attempts and says when to call. Free. - [How to Reduce AI Costs: Pay a Model Only for the Decisions That Need One](https://polyhelper.ai/research/tools/decidecall/): Reduce AI costs on ticket triage and approvals: a cache, rules and a local model decide first, a paid model or a person only the rest. Free, open source. - [How to Measure AI Productivity by Checked Work, Not Tokens](https://polyhelper.ai/research/tools/delegation-horizon/): Turn a log of the work you hand to AI agents into one dated line for a résumé or review: how big a piece you delegate and verify. Free, open source. - [AI for Court Documents: Catch Fake Citations Before You File](https://polyhelper.ai/research/tools/docprep/): AI for pro se litigants: AIs from rival companies draft your filing blind, one merges, two check every citation. Free, open source, no API key needed. - [Can You Trust AI Answers? Ask a Second AI From Another Company](https://polyhelper.ai/research/tools/duocall/): Before you act on an AI answer, ask an AI from another company. Duocall quotes both where they agree and names every difference. Free, no API key. - [Claude for Your Company: Your Own Plugin and the Same Rules on Every Laptop](https://polyhelper.ai/research/tools/firmcall/): Turn your team's frequent tasks into your own Claude plugin and put the same rules on every laptop through Jamf, Intune or the admin console. Open source. - [Stop AI From Leaking API Keys, Card Numbers and Client Files](https://polyhelper.ai/research/tools/gatecall/): Free Claude Code plugin that stops API keys, card numbers, IBANs, ID numbers and client lists leaving in an AI message, and gives each kind of data a route. - [AI Morning Summary of Your Email: Who Is Waiting for Your Reply](https://polyhelper.ai/research/tools/mailcall/): Free Claude Code plugin that reads your mailbox read-only and writes a morning summary: what matters, who awaits your reply, Google Alerts, newsletters. - [Run Claude Code Overnight and Wake Up to a Report](https://polyhelper.ai/research/tools/nightcall/): Leave Claude Code working all night on one task: it keeps your computer awake, checks its own work, switches AIs when a limit hits, and leaves a morning report. - [Control Claude Code From Your Phone and Leave the Computer Working](https://polyhelper.ai/research/tools/pocketcall/): Claude Code remote control not working from your phone? Pocketcall checks the 7 silent causes, shows every job on one screen and rings you. Free. - [Free Claude Code Plugin Marketplace: 12 Open Plugins, Added With One Line](https://polyhelper.ai/research/tools/poly-a1-plugins/): Free Claude Code plugin marketplace with 12 open plugins: file copies, second opinions, overnight work run from your phone, mail summaries, AI costs. One line. - [Ask ChatGPT, Gemini, Grok and DeepSeek the Same Question at Once](https://polyhelper.ai/research/tools/roundcall/): A free Claude Code plugin that asks several AI chat sites your question in your own Chrome and gives one report: where they agree, disagree and who said what. - [Run Claude Code on a Cheaper Model Without Losing Your Phone Remote](https://polyhelper.ai/research/tools/routecall/): Use DeepSeek, GLM, Kimi or a local model in Claude Code for routine work in its own window, and keep your Claude seat, Remote Control and advisor. Free plugin. - [Undo Button for Claude Code: Back Up Your Files Before AI Changes Them](https://polyhelper.ai/research/tools/safecall/): A free Claude Code plugin for non-programmers: it copies a file before the AI changes it, puts it back on request, and says what was and was not checked. - [Get a Second Opinion on Claude Code's Answer From Codex, Grok and Other AIs](https://polyhelper.ai/research/tools/sidecall/): A free Claude Code plugin that asks Codex, Antigravity, Grok or Kimi the same question through your own subscriptions and shows where they agree and differ. - [Is My SSRN Download Count Good? A Free Benchmark by Paper Age and Field](https://polyhelper.ai/research/tools/ssrn-benchmark/): Check whether your SSRN download count is normal. Free tool built from 1,360 archived SSRN papers, with percentiles by paper age and subject field. - [How to Get Your Team Using AI Agents: a 30-Day Pilot Plan](https://polyhelper.ai/research/tools/teamcall/): Bring colleagues from the AI chat window to working agents in 30 days: a pilot plan, a 20-minute first session, a daily card in Telegram, a report with numbers. ## Optional - [Research index](https://polyhelper.ai/research/): every paper, note and tool - [Research feed](https://polyhelper.ai/research/feed.xml): RSS of new papers and notes - [Sitemap](https://polyhelper.ai/sitemap.xml): every page in ten languages - [Everything in one file](https://polyhelper.ai/llms-full.txt): the main pages' text and every paper's summary - [中文](https://polyhelper.ai/zh/): the site in 中文 - [हिन्दी](https://polyhelper.ai/hi/): the site in हिन्दी - [Español](https://polyhelper.ai/es/): the site in Español - [Français](https://polyhelper.ai/fr/): the site in Français - [Português](https://polyhelper.ai/pt/): the site in Português - [Русский](https://polyhelper.ai/ru/): the site in Русский - [Deutsch](https://polyhelper.ai/de/): the site in Deutsch - [日本語](https://polyhelper.ai/ja/): the site in 日本語 - [Українська](https://polyhelper.ai/uk/): the site in Українська