Knowledge base
Everything about PyGar.AI.
How the engine reasons, every personality and strategy, your back office, billing, privacy, and what to do when something looks wrong.
Platform overview
What PyGar.AI is, who it is for, and the principles that shape every answer.
- What is PyGar.AI?A reasoning engine for individuals and companies who want advanced reasoning without shallow thinking.
- Who is it for?Decision-makers, professionals and individuals who need rigour and want to see how an answer was reached.
- Core principlesShow the work. Name the cost of being wrong. Refuse easy fluency. Keep the human in the loop.
- The pillars of the engineMethodology, Personalities, Strategies, Fingerprint, Mirror and Meta-Brain — each plays a specific role.
Methodology
Classification, routed panel roles, mode-specific panel shapes, synthesis and safeguards.
- How the engine reasonsEvery question is classified, assigned a mode-specific panel, then synthesised under explicit safeguards.
- The five routed panel rolesLead, Depth, Test, Ground and Human shape the Standard category-routed panel.
- Classification and routingSixty categories, each with a fixed routing recipe that names which personality fills which seat.
- Loop safeguards and mitigationsSynthesis is terminal, panels cannot self-summon, and Test cannot also occupy Lead.
Personalities
The council of 23 cognitive agents, what each one is for, and how they are summoned.
- The council of 23Twenty-three cognitive agents, each mapped to a region of cognition.
- How personalities are summonedCategory routing composes Fast, Standard and Philosophical panels; User Defined lets eligible users choose 2–10 voices directly.
- Bias is a featureEach personality has a deliberate bias. Synthesis weighs them; it does not average them.
Strategies
Twelve structured thinking moves the Mirror prescribes when it spots a gap.
Fingerprint
Two separate layers: how you describe your thinking and how it appears in the questions you ask.
- Your Fingerprint has two layersBaseline records how you describe your thinking; Evolution tracks how it appears in your real questions.
- How to read your fingerprintRead Baseline as self-description and Evolution as question-derived evidence; agreement or difference between them is information, not a verdict.
- Continue or re-take your baselineProgress is saved, readings lock at four milestones, and you may continue or begin a new baseline whenever you choose.
- How to read pattern profilesReader-facing pattern names describe repeated question structures and show evidence, possible value, possible cost and a question to test next.
- Your locked weekly reportOne write-once report per completed UK week, generated after at least five new questions and downloadable as a PDF.
Mirror
Private question-history readings and one evidence-backed recommendation at a time.
- What the Mirror doesBuilds private readings from your question history and may surface one evidence-backed strategy recommendation at a time.
- Recommendations and pattern readingsA recommendation offers one questioning technique; a pattern reading explains repeated evidence in your question structure.
Meta-Brain
Your searchable memory of every question, every panel, every synthesis you have ever produced.
Asking questions
How a question moves from prompt to classification to panel to synthesis, step by step.
- How to ask a good questionThe engine rewards specific questions with stakes. Vague prompts get vague panels.
- The ask flow, step by stepFrom prompt to classification to panel confirmation to synthesis.
- Ask modesFast, Standard, Philosophical and User Defined have different panel shapes and hard run limits.
Workspace & repositories
Dashboard, thinking projects, saved questions and repository status controls.
- The dashboardWhere you start questions, open thinking projects, and review private Fingerprint and Meta-Brain entry points.
- RepositoriesThinking projects that group questions under a title and an active, dormant or archived status.
- Opening saved questionsOpen a repository question to review its saved panel and synthesis, or use Meta-Brain to search across question text and syntheses.
Account & billing
Plans, launch pricing status, subscription controls and account-specific access.
- TiersPersonal Taster, Full Personal and Enterprise — three user plans on the same engine.
- How to subscribeThe public Pricing page shows launch readiness; checkout begins only when a price is configured and presented to your account.
- Manage your subscriptionUse the customer portal to change card, update billing details or cancel.
- Plan access and allowancesHow the Personal Taster cap, Ask-mode access and reasoning-model allowances affect a run.
- Custom pricing for individualsAdmins can set a per-user monthly price that overrides the tier default.
Privacy, data & trust
Application data, payment handling, model providers, deletion and exports.
- What we storeYour questions, panel outputs, syntheses, repositories and account metadata.
- Payment and credential boundariesStripe handles full payment-card details and the authentication provider manages credentials outside Pygar application tables.
- Model providersReasoning may use configured Anthropic, OpenAI, Google or xAI integrations, depending on the task and selected reasoning model.
- Export and deleteDownload a machine-readable account export or schedule account deletion with a 30-day grace period.