LoupistGitHubDownload

Loupist Docs

Search Loupist Docs

Browse all pages or type to search

Get startedOverviewLearn what a Loupist keeps, how it uses evidence, and where to begin your first research question.Get startedQuickstartInstall Loupist, connect an AI account, and check the evidence behind your first answer.Get startedInstall and verify the BetaInstall the macOS Beta and check that the download matches the public release workflow.Get startedConnect an AI accountConnect the account or API key that will answer your questions, and understand which work uses it.Get startedCreate a LoupistCreate a separate research companion with its own focus, conversations and Memory.Core conceptsHow a Loupist learnsFollow the path from a retained original to a method used in a later answer, and inspect what changed.Core conceptsMemory recordsTell retained evidence, analysis and your own judgments apart, and find each record in the app.Core conceptsWorld and MapTrace a relationship to its evidence, navigate your Map, and export an image or growth replay.Core conceptsLens and your principlesTeach a research rule, inspect pages built from evidence, and see why a learned method changed.GuidesFollow assets and import holdingsBuild the list your research uses, import a holdings snapshot, and distinguish following from automatic checks.GuidesWatching and alertsChoose automatic checks, inspect their source coverage, and pause or refresh them without an AI call.GuidesBriefings and ScheduleCreate timed research in Bot, inspect the result, and control future runs from Schedule.GuidesBot and research sessionsUse Bot for ongoing work and open a separate research session for a question that needs its own conversation.GuidesRead an answerCheck a finding against its original sources, inspect calculated figures, and distinguish a receipt from proof of correctness.GuidesRun calculations with run_pythonAsk for reproducible numerical work and inspect its code, files and limits.GuidesJournal entries and later reviewsRecord your own judgment with an optional dated check, then compare it with later evidence.GuidesPrivate chatTurn off Memory use and saving for a conversation while understanding what still persists and leaves your Mac.GuidesData connectionsSet up a built-in source or a custom read-only service and choose which Loupists may use it.GuidesMove or back up a LoupistPreserve a Loupist folder, reopen its saved identity, and know which shared settings need separate setup.ReferenceData sourcesFind the source for a research task and identify the setup or coverage to check before using it.ReferenceKeyboard and command paletteNavigate the app and find research without the mouse, with the current shortcut list available in Settings.ReferenceSettingsFind the right settings scope for accounts, defaults, appearance and a single Loupist.ReferenceWhat leaves your Mac and securityCheck which requests leave your Mac, where credentials stay, and what the app permits research tools to do.ReferenceUpdates and Beta limitsCheck for an update, understand version compatibility, and confirm the supported Beta platform.ReferenceTroubleshooting and FAQRecover from common setup, account, source and schedule problems without treating missing evidence as a successful check.ReferenceLicenceFind the terms for the app, distinguish them from data-provider rights, and keep private research separate from code licensing.

↑ ↓ navigate · Enter opens · Escape closes

Reference

Troubleshooting and FAQ

Recover from common setup, account, source and schedule problems without treating missing evidence as a successful check.

macOS will not open the Beta#

This build is not notarized. Follow the one-time System Settings → Privacy & Security → Open Anyway steps in Install and verify the Beta. If the button is gone, repeat the blocked launch first. If macOS says the app is damaged, download again and verify; do not bypass that message.

I cannot get an answer#

Open Settings → AI accounts. Reconnect an expired account, check an account limit or credit notice, or use an eligible API key. Look around first does not connect a model. If models could not load, choose Reload models. Your saved research remains readable without an AI account.

For a failed reply, inspect the preserved steps and text, then use Try again. It repeats the original request with current settings. Files written by the first attempt can remain.

A Source is saved but the Map is empty#

Retention and Map building are separate. Check the Source's coverage and processing status, then Loupist settings → Map building. Readiness can require an account, a supported model or resuming processing. A binary original without readable text cannot supply relationship extraction.

Map building stopped for this source can offer Try again. This uses another allowance of model requests. Recover map building restores the last confirmed Map without a model; it does not authorize retrying an uncertain paid attempt.

I receive no alerts or briefing#

Following alone starts no checks on a new install. Enable Watch all or an applicable Watch for choice. The first read sets a starting point without alerting. Check Data connection coverage and failures.

Open Schedule to check whether the task is paused and inspect its last run. Briefings need an AI account; Watching does not. Both need the app running. Inspect Settings → General → Keep running in the menu bar. Three consecutive recurring failures can pause a task.

My folder moved#

Choose Reconnect folder and locate the original or copied folder including .loupist/. An empty folder cannot replace it. A newer saved format requires a newer app. Removing a registration leaves the folder intact.

Does Private chat leave no record?#

No. It skips Memory reads and saves. Its transcript is still saved and messages still go to the AI provider. Use Private chat to check the full behavior.

Does learning mean training a model or promising returns?#

No. Learning retains evidence and methods for later relevant research. Loupist can share its view but is not a licensed adviser and never trades. Check a later answer's actual evidence and principle use rather than relying on counts.

Based on the app’s source documentation and English interface labels. Report a docs issue.