

        <cite>Sean Parent</cite>
      <time>0:00</time>
      <p>
          So   I   think   one   of   the   great   powers   of   AI   was   just   getting   into   that   first   gear,   right?   Right?   How   do   you   just   get   started   on   this   problem?
      </p>

        <cite>Conor</cite>
      <time>0:19</time>
      <p>
          Welcome   to   ADSP   The   Podcast,   episode   253,   recorded   on   August   21st,   2025.   My   name   is   Connor,   and   today   with   my   co-host   Bryce,   we   chat   with   Sean   Parent   about   cursor   and   AI.
      </p>

        <cite>Sean Parent</cite>
      <time>0:37</time>
      <p>
          But   you   mentioned,   you   know,   my   comment   and   my   push,   and   that   was   uh   the   Adobe   Source   Libraries,   which   uh   uh   uh   these   are   where   the   the   Adam   library   is,   and   you   know,   I&#39;ve   maintained   it   for   20-something   years   as   kind   of   a   side   project   uh   since   the   previous   software   technology   lab.   And   during   the   period   where   I   left   Adobe   and   went   to   Google,   uh   there   was   like   an   update   to   the   Doxygen   tool,   and   it   broke   the   generation   of   Doxygen   and   like   badly   broke   broke   the   generation   of   the   of   the   documentation.   Uh   so   I&#39;ve   left   the   old   site   online,   uh,   uh,   but   for   20   years   I   haven&#39;t   been   able   to   build   the   documentation.   And   I   took   a   stab   at   it   at   one   point,   and   Foster   got   uh   Foster   Breton,   somebody   I   work   with,   he   got   uh   a   couple   of   the   files   building   at   one   point.   But   it   was   like   uh   the   more   time   that   went   by,   the   more   impossible   and   intractable   it   seemed.   And   part   of   it   was   when   we   originally   wrote   the   documentation,   uh,   you   know,   I   didn&#39;t   have   have   a   folding   editor.   Uh   so   I   hated   having   big   blocks   of   description   in   my   source   files.   So   I   wrote   all   the   documentation   as   external   in   external.docs   files.   And   for   Doxygen   to   match   those   up,   it   means   the   signatures   have   to   match.   But   the   signatures   don&#39;t   match   textually   exactly.   They   used   to   match,   have   to   match   through   Doxygen&#39;s   old   AST   parser,   and   now   they   have   to   match   through   the   the   the   Clang   AST   parser   because   Doxygen   has   moved   to   Clang.   Okay,   so   if   you   think   about,   you   know,   you&#39;ve   got   several   thousand   functions   in   there   and   and   those   signatures   don&#39;t   match,   and   you   can&#39;t   copy   and   paste   to   fix   them.   And   even   if   you   could   copy   and   paste,   you   have   to   be   able   to   figure   out   which   one   matches   with   which   one,   and   they&#39;re   written   in   a   different   form.   And   now,   20   years   later,   the   API   is   probably   changed.   So   you   have   this   huge   problem   of   trying   to   figure   out,   like,   okay,   which   documentation   goes   with   which   function?   And   because   of   overloading,   you   can&#39;t   even   just   say,   Oh,   I&#39;m   just   gonna   search   for   the   name   of   the   function.   It&#39;s   like,   oh   great,   I   got   five   of   them.   So   which   one   of   the   five   documentations   goes   with   which   one   of   the   five   overloads?   Right?   Right,   right.   And   now   there&#39;s   probably   six   overloads,   so   which   one   am   I   missing?   Right?   And   uh   so   every   time   I   took   a   stab   at   it,   I   was   just   like,   this   is   just   too   much   work,   it&#39;s   not   worth   it.   I&#39;m   not   doing   it,   I&#39;m   out.   So   I   think   one   of   the   great   powers   of   AI   was   just   getting   into   that   first   gear,   right?   Right,   how   do   you   just   get   started   on   this   problem?   So   I   wrote   this   huge   prompt,   and   you   know,   it   took   me   a   couple   of   tries   where   I&#39;m   like,   no,   you&#39;re   going   way   off   in   the   weeds,   and   like   reset   it,   set   it   so.   But   I   ended   up   with   this,   you   know,   a   big,   huge   prompt   where   it   was   like,   okay,   here&#39;s   the   deal.   This   is   what   I&#39;m   trying   to   do.   You   know,   this   is   what   has   happened,   here&#39;s   the   history   of   it,   here&#39;s   the   problem,   here&#39;s   the   challenge.   Formulate   a   plan   for   how   we&#39;re   going   to   approach   this,   you   know,   incrementally.   Don&#39;t   try   to   do   it   all   yourself   at   every   step   of   the   way.   We,   you   know,   we   will   stop   and   validate   and   iterate.   Do   you   have   that   prompt   somewhere?   I   might.   I   might   have   kept   it.   I   um   burned   through   an   entire   month&#39;s   allotment   of   uh   uh   uh   of   Claude   4   max   mode   tokens   in   about   four   hours,   uh,   but   got   it   done.   And   and   the   docs   are   now   posted,   and   it&#39;s   like   the   first   time   the   docs   have   built   in   20   years,   and   they   look   much   better   and   they&#39;re   much   simpler   to   maintain.   And   uh,   you   know,   we   spotted   a   couple   uh   a   couple   mistakes   and   a   couple   missing   items   in   reviewing   them,   but   but   uh   uh   those   were   easily   fixed   and   uh   it   got   it   all   going.   And   so   so   you   know   it&#39;s   like   could   have   I   had   done   it   faster   by   hand?   Probably   not,   you   know.
      </p>

        <cite>Bryce</cite>
      <time>4:56</time>
      <p>
          Maybe   if   you   incorporate   all   the   time   into   into   fill   you   know   figuring   out   the   prompt   and   all   that,   but   getting   there   was   just   so   daunting   that   I   wouldn&#39;t   Yeah,   but   but   but   the   time   figuring   out   the   prompt,   even   if   you   spend   more   time   on   it,   that&#39;s   higher   value   time   than   doing   the   repetitive   task   of   going   through   these   7,000   functions.
      </p>

        <cite>Sean Parent</cite>
      <time>5:16</time>
      <p>
          Exactly.   And   it   once   I   got   it   going,   it   was   really   nice   because   it   it   would   go   through   and   it   would   say,   like,   like,   oh,   okay,   so   so   we&#39;re   compiling   the   docs,   we   got   this   error   from   Doxygen,   here&#39;s   my   suggestion   for   how   to   fix   it.   And   I   would   be   like,   like,   that&#39;s   great,   do   that.   And   it   would   say,   Okay,   there&#39;s   600   more   identical   issues.   Do   you   want   me   to   fix   them   all?   And   it   would   ask,   and   I&#39;d   be   like,   Yes,   fix   them   all.   Done.   Right,   right.   And   um,   you   know,   uh   it   would   come   up,   come   up   and   say,   you   know,   we   got   this   other   problem.   Here&#39;s   my   suggestion   for   how   to   fix   it.   And   I&#39;d   be   like,   no,   I   think   there&#39;s   a   much   simpler   fix,   do   it   this   way.   It   would   be   like,   okay,   doing   it   that   way.   Just   like,   you   know,   of   course   it   always   says,   like,   that&#39;s   a   much   simpler   fix.
      </p>

        <cite>Bryce</cite>
      <time>5:58</time>
      <p>
          Yes.
      </p>

        <cite>Sean Parent</cite>
      <time>5:59</time>
      <p>
          Right,   right.   You&#39;re   great.   Yeah,   so   just   got   this   iterative   process   going.   And   it   was,   it   was,   do   we   know   when   we   started,   I   don&#39;t   know,   something   like   7,000   errors   when   we,   you   know,   because   it   was   like   phase   one.   I&#39;m   like,   just   set   up   the   framework   so   we   can   run   doxygen   on   all   the   sources   and   don&#39;t   try   to   fix   any   errors,   just   get   that   going.   It&#39;s   like   once   we   got   that   going,   it   was   like,   okay,   now   there&#39;s   7,000   somewhat   errors.   So   let&#39;s   my   first   thing   was   like   organize   those   into   categories   so   we   can   try   to   attack   these   systematically.   So   it   came   back   with   like,   like,   here&#39;s   my   proposed   categories.   I&#39;m   like,   okay,   that   looks   great.   Let&#39;s   start   with   category   number   one,   right?   Right,   let&#39;s   figure   out   how   to   fix   that.   Uh   so   it   worked   really   well.   So   now   I&#39;ve   got   the   uh   the   ASL   docs   are   all   back   online   for   the   first   time   in   20   years.
      </p>

        <cite>Conor</cite>
      <time>6:46</time>
      <p>
          So   this   is   the   I   mean   it&#39;s   an   amazing   story.   And   uh   obviously,   because   uh   as   you   mentioned,   uh   I   think   it&#39;s   rocket   fuel.   So   it&#39;s   obviously   I   love   to   hear   stories   of   other   people   thinking   it&#39;s   rocket   fuel.   But   uh   to   play   uh   not   devil&#39;s   advocate,   but   to   answer   the   question   that   I&#39;m   sure   because   we   we   definitely   have   had   I&#39;ve   talked   to   folks   and   people   have   uh   mentioned   in   the   GitHub   discussions   that   they&#39;re   they   find   the   these   tools   counterproductive,   et   cetera,   et   cetera.   And   they&#39;re   probably   thinking,   well,   Sean   said   it&#39;s   rocket   fuel,   but   at   the   same   time   he   said   he   lost   a   week   debugging   an   issue   because   it   it   gaslit   him.   He   also   had   to   spend   like   a   number   of   hours   iterating   on   a   prompt,   like   that&#39;s   not   what   I   want   to   do.   So   how   do   you   uh   I&#39;m   sure   the   for   the   person   that&#39;s   thinking   that   in   their   head,   how   do   you   answer   that,   like,   oh,   how   can   it   be   rocket   fuel   if   you&#39;re   ending   up,   you   know,   chasing   bugs   for   days   uh   or   a   bug   for   days   and   you&#39;re   having   to   do   this   iteration?   Yeah.   Curious   to   get   your   thoughts   if   if   that   person   was   in   this   conversation.   Yeah.
      </p>

        <cite>Sean Parent</cite>
      <time>7:44</time>
      <p>
          Yeah.   I   you   know,   I   think,   you   know,   net   I   am   more   productive.   I   I   also   think   I   I   tackle   things   that   I   wouldn&#39;t   normally   tackle,   like   this   documentation   problem.   But   there   there   are   other   examples   too,   right?   Right,   where   it   would   just   be   like,   you   know,   I   could   figure   that   out,   but   it&#39;s   too   much   work   and   too   low   a   value.   And   and   so   so   having   even   a   device   where   you   can   have   a   conversation,   it&#39;s   like   having   a   collaborator,   right?
      </p>

        <cite>Bryce</cite>
      <time>8:12</time>
      <p>
          Yeah.
      </p>

        <cite>Sean Parent</cite>
      <time>8:12</time>
      <p>
          And   and   that   can   have   have   have   huge   benefit.   Uh   you   know,   and   if   you&#39;ve   ever   worked   with   a   collaborator   or   you&#39;ve   ever   done   anything   that   you   don&#39;t   know   how   to   do,   you   go   down   dark   alleys   also,   right?   It&#39;s   not   like   when   I&#39;m   writing   my   own   code,   it&#39;s   like   always   perfect   out   of   the   shoot.   It&#39;s   like   I   write   pile   the   code   and   throw   it   away.   I&#39;m   like,   well,   that&#39;s   not   going   anywhere.   Uh   so   I   think   the   AI   helps   me   do   that   faster,   right? Right. Uh   so   I   don&#39;t   view   view,   you   know,   much   of   the   time   as   wasted.   And   I   learned   pretty   early   on   that   if   it&#39;s   not,   if   I   know   what   I   want   and   it&#39;s   not   doing   what   I   want,   then   just   do   it,   right?   Right?   Right,   don&#39;t,   you   know,   early   on,   especially   when   you&#39;re   trying   to   to   try   out   one   of   these   systems,   you&#39;re   like,   here&#39;s   the   problem,   and   you&#39;re   giving   it   a   problem   that   you   know   the   answer   to,   and   it   gives   gives   you   an   answer   that&#39;s   close   but   not   exactly   there.   And   you&#39;re   like,   oh,   well,   what   about   this?   And   you   you&#39;re   trying   to   convince   it   to   get   to   the   answer   that   you   have   in   your   head,   and   you   just   get   frustrated   and   you   bang   your   head   against   the   wall,   right?   And   and   it&#39;s   like,   that&#39;s   not   how   you   would   work   with   a   person,   you   know.   That&#39;s   not   how   you   should   you   should   work   with   with   one   of   these   systems.   If   you   if   you   know   the   answer,   tell   it   exactly   what   you   want   and   be   very   explicit.   Um   or   just   do   it   yourself.
      </p>

        <cite>Bryce</cite>
      <time>9:30</time>
      <p>
          I   I   have   found   myself   that   like   more   and   more,   like   when   I   first   started   off,   I   tried   to   have   it   do   everything   and   I   got   really   frustrated   when   it   didn&#39;t.   But   now   it&#39;s   like   when   I&#39;m   like,   okay,   like   I   can   just   do   this,   I&#39;ll   just   go   and   do   it.
      </p>

        <cite>Sean Parent</cite>
      <time>9:41</time>
      <p>
          Yeah,   and   there   is,   you   know,   now   it   comes   up   and   it&#39;s   like,   you   know,   in   in   if   you&#39;ve   ever   used   used   Claude,   it   comes   up   for   any   changes   it   makes,   it   gives   you   a   diff,   right?   So   you   can   accept   or   reject   things   individually.   And   I   will,   I   don&#39;t   know,   probably   70%   of   the   diffs   it   presents,   I   will   edit   them,   right?   Yeah.   And   and   rather   than   modifying   my   prompt   and   being   like,   like,   oh,   don&#39;t   do   it   that   way,   do   it   this   way,   because   it   doesn&#39;t   have   any   memory.   So   unless   I   take   the   time   to   put   it   into   a   rule,   uh,   and   most   of   the   time   it&#39;s   not   worth   it,   it&#39;s   not   going   to   remember   it.   So   there&#39;s   no   use   in   me   training   it.
      </p>

        <cite>Bryce</cite>
      <time>10:18</time>
      <p>
          One   of   the   main   reasons   that   I   use   uh   uh   GPT-5   um   uh   a   lot   in   alongside   using   like   Claude   and   Cursor   is   because   GPT-5   has   like   it   has   this   memory   system   that   I   opted   into   like   six   months   ago.   And   so   now   it   knows   like   a   lot   of   stuff,   and   so   I   know   I   can   ask   it   a   question   with   a   lot   less   context,   and   it&#39;s   like,   oh,   I   know   I   know   that   this   is   Bryce,   I   know   that   he   works   at   NVIDIA   and   works   on   CUDA.   There   is   the   occasional   problem   that   my   girlfriend   and   I   share   the   Chat   GPT   account,   but   I   do   I   do   legitimately   think   that   it   had   that   like   it   ha   I   haven&#39;t   checked   its   memory   in   great   depth,   but   I   do   think   that   in   somewhere   in   the   memory   it&#39;s   got   an   understanding   that   there   are   two   different   people   interacting   here.   But   uh   but   it&#39;s   nice   because   like   I   I   know   that   if   I   go   into   ChatGPT   that   I   can   give   it   a   lot   less   background.   I&#39;m   not   sure   whether   Anthropic   has   a   memory   feature   and   whether   it&#39;s   something   that   Cursor   has   turned   on.
      </p>

        <cite>Sean Parent</cite>
      <time>11:14</time>
      <p>
          I   haven&#39;t   seen   it,   but   now   I   will   look   for   it   because   if   it   had   a   memory   feature,   that   would   be   great.   And   I   frequently   forget   to   uh   uh   kind   of   you   know   end   one   chat   and   start   a   new   chat,   and   every   once   in   a   while   I&#39;ll   look   at   the   title   of   my   tab   and   it&#39;ll   be   like   I&#39;ve   been   having   five   conversations   in   the   same   thing,   and   and   and   I   the   bigger   your   context   gets,   I   know   it&#39;s   summarizing   it,   but   it&#39;s   like   like   the   crazier   it&#39;s   getting.   The   crazier   it&#39;s   getting.   Yeah,   and   you&#39;re   like   like,   okay,   stop.
      </p>

        <cite>Bryce</cite>
      <time>11:42</time>
      <p>
          I   I   had   a   I   had   like   a   four   to   five   month   period   where   cursor   was   nearly   unusable   for   me   because   it   like   writes   this   database   of   like   the   current   state   of   everything,   including   the   chat.   Um,   and   it   would   just   like   it   it   was   every   five   to   ten   seconds   the   entire   UI   would   lock   up   because   for   whatever   reason   they   were   doing   some   blocking   I.O.   in   the   UI   thread,   which   you   should   never   do,   but   they   were   doing.   Uh,   but   so   I   got   very   religious   about   like   starting   a   new   chat   frequently   because   that   seemed   to   make   the   issue   not   so   bad.   Fortunately,   latest   version   of   cursor   no   longer   have   that   problem.   But   uh   yeah,   for   a   long   time   I   was   working   with   like   every   every   like   five   to   thirty   seconds,   maybe   cursor   would   like   freeze   up.   Um   just   like   I   could   I   could   be   like   typing   it,   it   would   freeze   up.   Just   anything,   it   would   freeze   up.   And   I   was   like,   why?   Why?
      </p>

        <cite>Sean Parent</cite>
      <time>12:34</time>
      <p>
          I   had   there   I   found   and   reported   a   bugging   cursor   that   was   driving   me   nuts,   which   was   if   it   uh   was   in   the   middle   of   doing   something   and   you   hit   stop   and   then   typed   in   a   prompt,   right?   Because   you   can   see   like   like   like   you&#39;re   doing   the   wrong   thing.   So   I   would   hit   stop   and   then   typed   in   in   a   prompt.   It   would   not   even   see   the   prompt   and   it   would   just   pick   up   and   continue.
      </p>

        <cite>Bryce</cite>
      <time>12:58</time>
      <p>
          Yeah.
      </p>

        <cite>Sean Parent</cite>
      <time>12:59</time>
      <p>
          And   uh   that   was   driving   me   insane.   It   was   like   like   stop,   stop,   stop,   stop.   Right.   And   the   only   solution   was   to   just   kill   the   tab,   and   it&#39;s   like,   yeah,   I   lost   all   my   context,   and   uh   so   it   was   so   painful.
      </p>

        <cite>Conor</cite>
      <time>13:13</time>
      <p>
          But   I   I   wonder   if   that&#39;s   why   uh   it   like   they   changed   the   interface   in   the   last,   I   don&#39;t   know,   month   or   so,   and   now   it&#39;s   uh   it&#39;ll   give   you   a   little   thing   and   it   says   run,   skip,   or   reject   and   tell   you   what   it   wants,   because   there   there   was   a   period   of   time   where   they   stopped   you   from   being   able   to   stop   it.   And   then   I   was   like,   uh,   because   I   I   I   I   found   that   half   the   time   it   would   work   and   half   the   time   it   wouldn&#39;t.   You   would   give   it   and   then   it   would   just   go   back.   And   I   was   like,   did   you   not   hear   what   I   just   said?   Uh   but   uh   but   now   they&#39;ve   got   like   a   much   better   interface   where   they&#39;ve   got   three   different   options.   Typically   you&#39;re   hitting   run.
      </p>

        <cite>Bryce</cite>
      <time>13:46</time>
      <p>
          But   it&#39;s   only   on   tool   commands.   What   they   what   they   need,   like   if   you&#39;re   if   it&#39;s   just   making   code   changes,   like   if   you   can   see   that   it&#39;s   planning   and   making   code   changes   that   are   in   the   wrong   way,   there&#39;s   no   like   interrupt   and   tell   like   like   it&#39;s   not.
      </p>

        <cite>Conor</cite>
      <time>13:58</time>
      <p>
          Well   you   can   just   interrupt   it   still   and   it&#39;s   like   skip   or   something   or   stop   and   then   just   type   and   it   works.   But   uh   usually   I   let   it   finish   whatever   it&#39;s   doing   and   then   tell   it&#39;s   tell   it   it&#39;s   wrong.
      </p>

        <cite>Sean Parent</cite>
      <time>14:07</time>
      <p>
          The   other   thing   is   if   you   don&#39;t   if   you   don&#39;t   wait   for   it,   if   you   just   type,   if   you   don&#39;t   stop   it,   don&#39;t   wait   for   it,   you   just   type   and   hit   return,   it   will   cue   it.   And   then   it   world   will,   at   the   next   opportunity,   will   look   at   that   cue   and   go,   like,   oh,   okay,   I&#39;m   gonna   adjust   you   my   plan,   blah,   blah,   blah.
      </p>

        <cite>Bryce</cite>
      <time>14:22</time>
      <p>
          Yeah,   I   haven&#39;t   I   have   not   had   a   chance   to   clear   an   art   that   much   yet   because   I&#39;ve   I   was   like   on   a   build   of   cursor   from   like   six   months   ago   to   deal   with   the   stability   issues.   But   have   have   either   of   you   two   tried   the   like   uh   like   having   multiple   checkouts   and   like   doing   concurrent   work,   like   having   multiple   checkouts   and   multiple   agents   working   at   the   same   time?
      </p>

        <cite>Sean Parent</cite>
      <time>14:44</time>
      <p>
          Uh   Adobe&#39;s   licensing   doesn&#39;t   allow   us   to   run   background   agents,   which   is   kind   of   a   pain.   Um   however   we&#39;re,   you   know,   whatever   we&#39;re   paying   to   the   cursor   folks.   Um   I   find   our   our   token   allotment   too   to   be   a   bit   limiting.
      </p>

        <cite>Bryce</cite>
      <time>15:01</time>
      <p>
          Connor   figured   out   how   much   NVIDIA   is   paying   and   what   our   I   guess   our   limits   pretty   generous   because   I&#39;ve   never   I&#39;ve   never   run   into   it.
      </p>

        <cite>Conor</cite>
      <time>15:08</time>
      <p>
          Well,   I   I   also   I   mean   we&#39;re   uh   we&#39;re   not   allowed   to   publicly   talk.   I   asked   if   I   was   allowed   to   report,   but   but   they   said   you   have   to   get   permission.   But   um,   because   you   can   go   to   I   think   this   is   for   any   organization,   it&#39;s   not   just   for   uh   NVIDIA   because   it&#39;s   it&#39;s   the   cursor   website.   If   you   go   to   cursor.com   slash   settings   and   you   have   a   corporate   whatever,   and   probably   even   if   you   have   a   personal   one,   they   have   a   whole   analysis   tab   where   you   can   see   over   the   last   one   day,   seven   days,   thirty   days,   um,   how   much   you&#39;re   using   by   language,   by   requests,   and   also   to   um   it   shows   you   your   rank   in   your   organization.
      </p>

        <cite>Bryce</cite>
      <time>15:42</time>
      <p>
          So   you   get   to   know   what   Sean&#39;s   rank   is.
      </p>

        <cite>Conor</cite>
      <time>15:45</time>
      <p>
          I&#39;m   not   gonna   go   awesome.   Yeah,   I   guess   I   guess   Sean   can   uh   can   go   and   look   that   up.   But   uh   the   interesting   thing   is   that   I   do   almost   all   of   my   work   on   Cloud   for   Sonnet,   which   I   don&#39;t   think   has   any   extra   billing.   But   if   you   are   using   the   uh   larger   models   like   4.1   Opus   or   4.0   Opus,   uh,   and   I   actually   think   4.   Cloud   4.0   Opus,   uh,   that   one   actually   is   more   expensive   than   4.1,   and   they   recommend   if   if   you   can   to   use   4.1.   So   if   you&#39;re   using   a   lot   of   the   larger   models,   I   think   there   are   extra   billing   fees   associated   with   that.   But   because   I&#39;m   always   using   force   on   it,   uh   I   I   basically   accrue   very   little   changes.   Um   I   can   see   Sean&#39;s   looking   it   up.   Uh   yeah,   I&#39;m   just   looking   out   of   curiosity.   I   would   I   would   love   to   report   my   ranking   and   how   many   folks   at   NVIDIA   uses   it,   but   maybe   at   maybe   at   a   future   date   when   they   say   we   can   talk   about   this,   but   uh   we   cannot,   unfortunately.   I   don&#39;t   actually   know   that   we   can&#39;t.   We   might   be   able   to.   I   just   I   don&#39;t   know   where   the   green   light   was   given.   So   uh   or   if   it   would   be   a   good   one.
      </p>

        <cite>Bryce</cite>
      <time>16:45</time>
      <p>
          It&#39;s   interesting   because   when   when   I   look   at   mine,   like   it   it   it   says   total   lines   accepted   for   the   past   like   seven   days   is   zero.
      </p>

        <cite>Conor</cite>
      <time>16:53</time>
      <p>
          And   sometimes   that&#39;s   a   bug.   Sometimes   it   doesn&#39;t   refresh   for   a   while   because   I&#39;ve   looked   at   it   a   couple   times   and   it   says   I   haven&#39;t   used   it   in   like   four   days,   and   I&#39;m   like,   what   are   you   talking   about?   I   spend   10   hours   a   day.   Like   I   I   launch   uh   Go   Chromecast   uh   to   watch   movies   uh   on   our   TV   using   Cursor   now.   I   just   open   up   Cursor   and   I   say,   Hey,   can   you   play   of   right   now   we&#39;re   working   through   uh   the   Star   Wars   with   uh   because   my   fiance   haven&#39;t   seen   them.
      </p>

        <cite>Bryce</cite>
      <time>17:17</time>
      <p>
          Wait,   wait,   wait.   How   do   you   how   how   does   it   know   how   to   do   that?
      </p>

        <cite>Conor</cite>
      <time>17:20</time>
      <p>
          So   this   this   is   it&#39;s   beautiful.   People   folks   don&#39;t   understand.   I   I&#39;m   living   in   the   future   here.   We   used   to   be   sipping   rocket   fuel.   Now   I&#39;m   just   living   in   the   future.   You   can   just   ask   it,   please   go   play   Return   of   the   Jedi.   If   you   have   that   movie   sitting   in   a   folder,   it&#39;ll   go   find   it.   It   knows   probably   like   it&#39;s   in   the   movies   folder.   It   goes   to   movies,   it   does   a   grep   for   Jedi.   Oh,   there&#39;s   a   folder,   let   me   go   check   in   there.   Oh,   there&#39;s   an   MKV.   And   then   it   goes,   let   me   go   search   for   a   Chromecast.   It   does   this   dash   dash   list,   finds   the   Chromecast,   sets   up   a   connection,   casts   it.   So   I   like   uh   we   set   it   up.   I   just   I   press   one   command.   I   have   to   hit   enter   enter   because   we   don&#39;t   have   YOLO   mode   yet.   Uh   but   the   point   being   is   like   I   I   want   uh   I   like   I   was   talking   to   my   fiancee   Shima.   I   like   I   need   it   in   my   sunglasses,   I   need   it   in   my   car,   I   need   it   everywhere.
      </p>

        <cite>Bryce</cite>
      <time>18:01</time>
      <p>
          Like   I   need   it,   I   need   it   in   Chrome.   Like   it&#39;s   still,   I&#39;m   I   so   okay.   Uh   I   don&#39;t   want   to   shut   in   Google.   However,   I   I   did   just   subscribe   to   uh   Google   like   AI   Pro   to   get   Gemini   Pro.   The   reason   was   I   wanted   Gemini   in   my   spreadsheet   because   I   had   like   the   task   was   I   had   a   list,   like   one   column,   which   was   names   of   people.   And   then   in   a   separate   spreadsheet   tab,   there   were   a   much   bigger   list   of   the   names   of   the   people   and   their   emails.   And   what   I   wanted   was   I   was   like,   hey,   here&#39;s   like   a   list,   like   I   like   this   set   uh   this   tab   has   this   list   of   10   names   of   people.   I   want   you   to   go   and   look   up   their   emails   in   this   other   tab   and   fill   them   in   here.   And   I   thought   this   was   gonna   be   like   the   easiest   thing.   And   I   asked   the   gem,   the   Gemini,   they   have   the   little   Gemini   thing   integrated   with   the   sheets,   and   I   press   the   little   button,   and   it&#39;s   like,   oh,   you   gotta   pay   money.   So   I&#39;m   like,   all   right,   I&#39;ll   go   pay   the   money.   So   I   get   the   subscription   and   I   go   back   in   and   I   ask   it,   and   it&#39;s   like,   oh,   I&#39;m   sorry,   like   our   our   current   tool   support   like   can&#39;t   do   this.   We   can&#39;t   look   at   stuff   in   another   tab.   And   so   I&#39;m   like,   all   right,   that&#39;s   weird.   And   so   I   I   go   and   I   copy   the   shorter   list   over   to   the   other   tab.   So   now   all   the   data   is   in   one   tab.   And   I&#39;m   just   like,   like,   I   could   I   could   pretty   easily   write   a   like   lookup   function   to   do   this   at   this   point.   And   I   ask   it   to   do   this   again,   and   it&#39;s   still   like,   I&#39;m   sorry,   I   can&#39;t   help   you   do   that.   So   then   I   just   take   the   whole   spreadsheet   and   I   just   upload   it   to   um   uh   to   ChatGPT   and   I   ask   ChatGPT   to   do   the   same   thing   and   to   output   CSV,   and   it&#39;s   done   in   like   30   seconds.   So   initial   experience   with   Gemini&#39;s   and   with   Google&#39;s   for   integration,   not   great.   But   like   what   I   really   want   is   I   want   a   little   chat   thing   in   the   browser   because   there   are   so   many   mundane   tasks   that   I   have   with   manipulating   web   UIs.   Where   it&#39;s   like,   oh,   I   need   you   to   go   and   like   click   on   this   like   button   like   a   dozen   times.   Or   like   I   was   I   was   dealing   with   something   with   Trippet   earlier   today   where   like   I   needed   to   add   10   entries   to   Trippet,   and   there&#39;s   no   bulk   upload   API,   there&#39;s   no   public   API,   there   was   no   way   for   me   to   automate   it.   Like   like   there&#39;s   no   way   for   me   to   tell   an   AI,   like,   just   go   do   this.   What   I   really   wanted   was   a   chat   box   in   my   browser   and   Google   Chrome   where   I   could   just   be   like,   you   go   do   this   for   me,   and   I   go   off   and   do   something   else.
      </p>

        <cite>Sean Parent</cite>
      <time>20:18</time>
      <p>
          I   wonder   if   Copilot   and   Edge   would   do   that   for   you.   I&#39;ve   never   tried   to   have   Edge   control   the   web   page,   but   I   mean   it   is   integrated   in   their   browser.
      </p>

        <cite>Bryce</cite>
      <time>20:26</time>
      <p>
          Maybe,   maybe,   maybe   we&#39;ve   finally   found   a   reason   for   me   to   try   using   Edge.
      </p>

        <cite>Conor</cite>
      <time>20:30</time>
      <p>
          And   I   know   I   know   that   uh   Perplexity   is   working   on   uh   a   new   browser,   the   company   that   uh   I   think   they   said   it   wasn&#39;t   a   joke,   but   they   offered   to   buy   Chrome   for   $34.5   billion,   which   is   a   very   brilliant   marketing   ploy.   Uh   because   apparently   they&#39;re   they&#39;re   in   court   right   now   and   may   have   to   sell   it.   And   like   if   there&#39;s   a   buyer   out   there,   they&#39;re   not   gonna   be   able   to   no   one   would   buy   Chrome.   Well,   technically   there&#39;s   someone   that   offered   to   buy   it.
      </p>

        <cite>Bryce</cite>
      <time>20:54</time>
      <p>
          So   why   would   why   would   no   one   buy   Chrome?
      </p>

        <cite>Conor</cite>
      <time>20:57</time>
      <p>
          I   don&#39;t   know.   It&#39;s   it&#39;s   too   integrated   with   all   Google   stuff,   and   it   would   take   like   a   decade   to   unwind   it   from   their   products.   Or   I   don&#39;t   know.   Anyways,   we   haven&#39;t   let   Sean.   Uh   have   you   have   you   been   able   to   find   your   settings?   Uh   I   did.
      </p>

        <cite>Sean Parent</cite>
      <time>21:09</time>
      <p>
          I   I   found   my   ranking.   So   I&#39;m   uh   out   of   about   uh   a   little   over   8,000   engineers   at   Adobe,   I&#39;m   ranked   1200   for   for   uh   number   of   accepted   lines.   So   that&#39;s   pretty   good.   Uh   uh   Yeah,   but   that&#39;s   I&#39;m   I&#39;m   kind   of   surprised   because   I   would   I   would   you   know   had   you   asked   me,   I   would   have   said,   Oh,   I&#39;m   probably   you   know   near   the   top   for   for   for   users,   but   no,   I&#39;m   uh   I&#39;m   not.
      </p>

        <cite>Conor</cite>
      <time>21:36</time>
      <p>
          What   do   you   think,   Bryce?   Are   we   allowed   to   uh   uh   we   can&#39;t   say   the   numbers?   Are   we   allowed   to   pr   report   the   percentage?   Because   I   did   calculate   my   I   think   that&#39;s   probably   fine.
      </p>

        <cite>Bryce</cite>
      <time>21:44</time>
      <p>
          I   think   you   could   probably   say   that   percent.
      </p>

        <cite>Conor</cite>
      <time>21:45</time>
      <p>
          So   yeah,   I&#39;m   I&#39;m   just   outside   the   top   10   percent.   Uh   I   think   a   while   ago   I   was   in   in   the   top   five   percent.   Uh   I   also   don&#39;t   super   trust   the   up   to   date   because   like   mine   also   shows   that   I   have   zero   lines   changed.   Since   August   6th,   according   to   these   things.   And   anyways,   I   don&#39;t   know   how   reliable   they   are.   But   I   also,   like   you,   John,   was   super   surprised.   I   was   like,   well,   of   course   I   gotta   be   the   top   100.   I&#39;m   a   power   user.   I&#39;m   a   power   user.   What   are   we   talking   about?
      </p>

        <cite>Bryce</cite>
      <time>22:13</time>
      <p>
          I&#39;m   in   the   top   third,   but   I   think   in   my   defense,   I   I   use   Chat   GPT   a   lot   for   like   directly   for   these   things,   and   I   use   other   AI   tools.   And   also   there   was   a   four-month   window   where   I   could   barely   use   cursor   because   of   the   freezes.   So   I   feel   I   I   I   think   I   use   a   wider   variety   of   tools   than   uh   than   you   do,   Connor.   I   think   like   I&#39;m   I&#39;m   using   like   now   I&#39;ve   been   playing   around   with   Gemini,   I   use   Perplexity,   like   I   use   a   wider   variety   of   uh   of   things.
      </p>

        <cite>Conor</cite>
      <time>22:42</time>
      <p>
          I   use   Gemini   quite   a   bit   now,   actually.   I   used   to   go   to   ChatGPT,   the   free   version,   to   answer   questions.   Uh,   but   ever   since   ever   since   they   updated   to   five.
      </p>

        <cite>Bryce</cite>
      <time>22:49</time>
      <p>
          It   blows   my   mind   that   you   do   not   have   the   paid   version   of   ChatGPT.
      </p>

        <cite>Conor</cite>
      <time>22:54</time>
      <p>
          My   fiance   does,   so   if   I   need   it,   I   I&#39;ll   go   use   hers.   But   um   ever   since   five,   uh   I   thought   it&#39;s   it&#39;s   just   gone   down   way   like   the   nice   thing   about   uh   ChatGPT   four-point,   whatever   it   was,   it   was   just   fast.   Now   it&#39;s   like,   oh,   I   decided   I   need   to   think   about   your   question.   It&#39;s   like,   well,   I   didn&#39;t   ask   you   to.   Uh   I   just   want   I   just   want   a   rapid   thing   and   Gemini.
      </p>

        <cite>Bryce</cite>
      <time>23:12</time>
      <p>
          Exclusively   used   it   for   O3,   and   O3   was   pretty   slow.
      </p>

        <cite>Conor</cite>
      <time>23:16</time>
      <p>
          I   I   I   did   it   for   like   I   need   some   I   need   like   uh   a   little   image   generated,   and   now   I   I   find   that   the   Gemini   one&#39;s   way   better.   Anyways,   back   to   Sean.   So   you   you&#39;re   you&#39;re   uh   in   the   you&#39;re   1,200   out   of   8,000   and   uh   and   not   as   high.   I   mean,   what   is   it   what   do   you   think   that   means?   Do   you   think   that   means   that   there&#39;s   just   a   bunch   of   people   doing   like   twice   as   much   work   as   as   uh   as   you   like,   you   know,   it&#39;s   just   people   are   doing   crazy   things   with   this   and   they&#39;re   running   it   overnight   or   something   like   that?   I   don&#39;t   know.
      </p>

        <cite>Sean Parent</cite>
      <time>23:42</time>
      <p>
          Uh   interesting   question.
      </p>

        <cite>Bryce</cite>
      <time>23:44</time>
      <p>
          But   there&#39;d   have   to   be   twelve   hundred   people   that   are   using   it   more   than   Sean.   That&#39;s   a   lot   of   people.
      </p>

        <cite>Sean Parent</cite>
      <time>23:52</time>
      <p>
          Yeah.   Yeah.   Yeah.   Within   the   company.   And   uh   well,   the   interesting   thing   is   it&#39;s   like   number   of   lines   accepted,   and   I&#39;m   like,   do   deleted   lines   count?   Is   one   of   my   questions   there?   Because   I   delete   a   lot   of   code.
      </p>

        <cite>Conor</cite>
      <time>24:05</time>
      <p>
          Oh   yeah.   That&#39;s   a   good   that&#39;s   a   good   point.   Uh   sometimes   it&#39;s   like   go   yeet   this   for   me,   and   it   goes   and   deletes   a   bunch   of   stuff.   And   like   potentially   you   added   five   thousand,   you   know,   two   weeks   ago,   but   then   you   deleted   that   same   five   thousand.   Does   that   like   affect   your   net   total?   I   would   I   would   imagine   that   like   it&#39;s   the   absolute,   you   know,   like   it&#39;s   whether   you   delete   it   or   added.   It&#39;s   changes,   yeah.   But   yeah,   it&#39;s   a   hard   thing   to   measure.
      </p>

        <cite>Sean Parent</cite>
      <time>24:30</time>
      <p>
          Yeah,   because   I   I   had   a   a   previous   boss   who   who   walked   into   my   office   one   time   and   he   said,   he   said,   so   I   don&#39;t   know   how   to   read   this.   He   said,   I   usually   check   the   GitHub   stats,   and   and   I   always   view   you   as   one   of   my   highest   performers,   but   your   total   line   count   stats   in   GitHub   are   are   very   negative.   It&#39;s   like,   yes,   I   delete   a   lot   of   code.
      </p>

        <cite>Conor</cite>
      <time>24:54</time>
      <p>
          Which   is   a   good   thing.
      </p>

        <cite>Sean Parent</cite>
      <time>24:56</time>
      <p>
          Which   is   a   good   thing.
      </p>

        <cite>Conor</cite>
      <time>24:57</time>
      <p>
          Be   sure   to   check   these   show   notes   either   in   your   podcast   app   or   at   adspthepodcast.com   for   links   to   anything   we   mentioned   in   today&#39;s   episode,   as   well   as   a   link   to   a   GitHub   discussion   where   you   can   leave   thoughts,   comments,   and   questions.   Thanks   for   listening.   We   hope   you   enjoyed   and   have   a   great   day.
      </p>

        <cite>Bryce</cite>
      <time>25:10</time>
      <p>
          Low   quality,   high   quality.   That   is   the   tagline   of   our   podcast.
      </p>

        <cite>Conor</cite>
      <time>25:15</time>
      <p>
          That&#39;s   not   the   tagline.   Our   tagline   is   chaos   with   sprinkles   of   information.
      </p>