visible screen context · display 1 of 2
invoked only
export
clickr says
that export button is hiding in the upper-right. yep, the tiny one. i circled it for you.
ctrl+althold to ask anything on screen
interactive placeholder · screen content is illustrative
native windowswpf + .net tray app
multi-monitorthe cursor’s screen comes first
byok-firstwindows credential manager
hold to talkctrl + alt, then release
features
stay inside the work.
hold one shortcut. ask one normal question. keep moving.
01 · show me where
learn any tool without leaving the tool.
stuck inside excel, blender, figma, visual studio, or that one settings panel designed by a committee? keep the problem visible and ask. clickr explains the next step out loud, then points to where you should look.
ctrlalt
live guidance
pointing now
exportthis one. clickr can explain why, then point you to the right control.
02 · skip the screenshot circus
your screen already has the context.
you shouldn’t need a five-paragraph prompt beginning with “there’s a blue button near the upper-right.” hold ctrl + alt, ask, and release. clickr captures the visible context—even across multiple monitors—and lets your chosen vision model reason over it.
invokecaptureforget
visible desktop context
2 displays
the screen under your cursor comes first
invoked only
invokecaptureforget
03 · agent preview
say the task. let an agent carry it.
the next version goes beyond answering. hand a bounded task to an agent by voice—research options, organize notes, or watch a background job—while clickr asks before anything consequential changes.
previewpermission-first
bounded agent queue
roadmap preview
voice requestresearch three options, write a clean brief, and ask before sending.
01research three optionsread-only web taskdone
02write a clean briefworking in draftworking
03ask before sendingexternal write protectedlocked
a note from giog
gio
maker of clickr
i kept doing the same ridiculous loop: get stuck, leave the app, open a browser, explain what was on my screen, upload a screenshot, then forget what i was trying to finish.
i wanted something closer to asking the person sitting next to me—except it lives in the tray, understands both monitors, and can point at the button i somehow missed six times.
so i started building clickr as a native windows app with wpf and .net. it’s fast to summon, byok-first, and designed to help without becoming another giant workspace you have to manage.
this is early software. i’ll keep the useful parts sharp, the weird parts fun, and the permissions understandable.
feedback wall
the wall of tiny wins.
sample early-tester copy for layout only. these are fictional placeholders, not customer endorsements.
sample / placeholder
“it pointed directly at the export button. rude. correct. helpful.”
dsample · designerlayout placeholder · not an endorsement
sample / placeholder
“it finally understands what i mean by ‘this weird thing over here.’”
{ }sample · developerlayout placeholder · not an endorsement
sample / placeholder
“holding ctrl + alt feels like using a walkie-talkie for my pc.”
opssample · operationslayout placeholder · not an endorsement
sample / placeholder
“two monitors, seventeen windows, one question. it kept up.”
vidsample · editorlayout placeholder · not an endorsement
sample / placeholder
“it feels native, not like a website wearing a windows trench coat.”
msample · indie makerlayout placeholder · not an endorsement
sample / placeholder
“the first ai tool that understands ‘no, the other monitor.’”
pwrsample · power userlayout placeholder · not an endorsement
replace this wall with verified feedback before a public launch.
pricing
less than one rage-click.
bring your own model key on every plan. provider usage is billed separately.
what to know before a tiny ai moves into your system tray.
clickr is a native windows tray app you summon with a hold-to-talk hotkey. it captures your screen context, sends the image and question to a vision-capable model, reads the answer aloud, and can point at the relevant place on screen.
clickr targets windows 10 version 2004 or newer and windows 11. windows 11 is the primary testing environment. there is no macos or linux build.
press and hold ctrl + alt, ask your question, then release. clickr processes the visible screen context and speaks the answer. if another app owns that shortcut, close or reconfigure the conflicting app.
no. the microphone opens while push-to-talk is active, and screen context is captured after you finish the request. screenshots and transcripts stay in memory rather than being written to disk. your selected ai provider still processes the request, so avoid showing sensitive data.
clickr captures all windows displays, sorts the monitor under your cursor first, and maps pointer guidance across the full virtual desktop. private information visible on any captured display can become part of the request.
in direct byok mode, provider keys are stored with windows credential manager instead of plain-text settings. optional proxy mode is available for controlled team deployments.
“bring your own key.” you connect an api key from a supported provider and pay that provider for model usage. a future paid clickr plan would license app features; it would not magically include unlimited third-party usage unless clearly stated.
the stable app provides spoken guidance and an animated on-screen pointer. it is not an unrestricted remote-control agent and does not silently click, type, or approve actions for you.
not in the stable windows build. voice-spawned agents are a preview and roadmap feature until they are implemented, permission-scoped, and tested. this page labels them that way on purpose.
clickr.techc
ask your pc.
the source is open today. a signed installer still needs to be shipped.