Latest / Business Podcasting: Production Strategy / How do you make a podcast with AI?

Transcript

How do you make a podcast with AI? · Business Podcasting: Production Strategy

• 0:00
  [Synthetic version of Lam Hoang's voice]
• 0:07
  This is Business Podcasting: Production Strategy, produced by WorkHacker.
• 0:13
  I'm your host, Lam Hoang, co-founder of WorkHacker.
• 0:16
  You're hearing a synthetic version of my voice.
• 0:19
  Each episode I take one question about making a podcast for your business,
• 0:23
  tell you what I would do and how solid the evidence is.
• 0:32
  Today's question: how do you make a podcast with AI?
• 0:36
  By one tally's own judgment, about one in four new
• 0:40
  podcasts in September 2026 were AI-made fakes.
• 0:44
  But using AI isn't the problem.
• 0:47
  Putting in less work because you used it is.
• 0:50
  That's what people call AI slop.
• 0:53
  The way to avoid it: you direct the AI, and you own what it makes.
• 0:57
  As of this fall, it can do most of the work of making a show.
• 1:01
  It needs three things from you to get there: guidelines, a vision and oversight.
• 1:06
  So you set the direction.
• 1:08
  You review the work, heavily at first.
• 1:11
  And you answer for what goes out, legally too.
• 1:14
  That's how this show is made, with the help of AI.
• 1:17
  Here's what you'll leave with.
• 1:19
  How this show splits the work, and why it leans in.
• 1:22
  Then what to keep for yourself, and how much to review as the models change.
• 1:27
  First: who does what, you or the AI?
• 1:31
  For this show, I set the direction.
• 1:33
  That means who the show is for and the questions we take on.
• 1:37
  It also means the format of each episode,
• 1:39
  what the show stands for and the rules every episode follows.
• 1:43
  AI gathers the research.
• 1:45
  It keeps a copy of every page it cites.
• 1:48
  So any claim can be checked against its page.
• 1:51
  Then a model drafts the script to my direction.
• 1:54
  A separate model checks every claim in the script against the page it came from.
• 1:58
  Then I listen to every episode from start to finish, with the script beside me.
• 2:03
  I approve each one or send it back with notes.
• 2:06
  Each note becomes a rule for later episodes.
• 2:08
  And before an episode publishes I also go through every one of its sources myself.
• 2:14
  And the voice really is mine.
• 2:16
  I just don't record the episodes.
• 2:18
  I'm not a professional podcaster.
• 2:21
  My work is websites, search and automation.
• 2:24
  Reading on mic every week isn't practical for me, or in my nature.
• 2:29
  Some listeners draw their line at the voice.
• 2:32
  In Edison Research's Evolving Ear study from 2025 data,
• 2:36
  listeners called hosting and narration the human core of podcasting.
• 2:41
  I agree.
• 2:42
  That's why we spend a lot of time making sure the voice sounds right.
• 2:46
  But hosting is also choosing what gets said.
• 2:49
  I still do that.
• 2:50
  For me, it's this voice or no show.
• 2:53
  Whether your show should use a synthetic voice is episode two.
• 2:58
  Second: why lean into AI at all?
• 3:01
  The way I see it, there's more worth doing in
• 3:03
  the world than people with the time to do it.
• 3:06
  So what could we do with more intelligence on hand?
• 3:09
  Having a tutor on call and a PhD to collaborate
• 3:12
  with should make our work better, not sloppier.
• 3:15
  Machines should help with machine tasks, like editing photos.
• 3:19
  The same goes for evening out an episode's sound and timing its chapters.
• 3:24
  Or making the title cards a show shares on social media.
• 3:27
  Those tasks are different from the human part of the work:
• 3:30
  the judgment, the taste and the choices.
• 3:33
  AI will let far more people make things.
• 3:36
  Look at websites.
• 3:38
  When my family got a computer back around 2000,
• 3:41
  people were building their own sites for fun.
• 3:43
  First custom pages, then Xanga and MySpace.
• 3:47
  Since then, a lot of that enthusiasm has faded.
• 3:50
  Most of us now build inside other people's platforms, with only so much control.
• 3:55
  But people are designing their own sites again and building their own tools.
• 3:59
  It feels like the early web again: a small renaissance in website design.
• 4:04
  And websites are just one example of what we see across many fields.
• 4:08
  The upheaval is real, though.
• 4:10
  So are the worries about who owns these tools
• 4:13
  and whether they widen gaps in power and income.
• 4:16
  Those are worries about control, not about the tasks a machine does.
• 4:21
  Still, I see a future where this technology helps.
• 4:25
  Third: what do you keep for yourself?
• 4:28
  The decision, the taste and knowing your context.
• 4:32
  It's not that AI can't help you think about what's best for your business.
• 4:36
  It just doesn't care enough to.
• 4:38
  So you have to tell it what you need.
• 4:41
  Put generic in and you'll get generic out.
• 4:44
  Without your direction, the model falls back on a default.
• 4:47
  A 2026 study gave one AI podcast tool different kinds of documents.
• 4:53
  The tool built the shows on the same fixed template.
• 4:56
  The creative work stays with you, and so does the responsibility.
• 5:00
  Your thinking also sets the bar.
• 5:03
  A show has to be worth hearing for people in your category,
• 5:06
  beyond the buyers you care about.
• 5:08
  Every episode should teach those listeners something, or entertain them.
• 5:13
  That thinking is the work I do for people.
• 5:16
  Last: how much do you review, when the tools keep changing?
• 5:20
  A lot at first.
• 5:21
  Review heavily while you iterate,
• 5:23
  until the episodes follow your vision and your guidelines.
• 5:27
  As your formats settle, you can review less.
• 5:30
  Notes turned into automatic checks make the review lighter.
• 5:33
  One early note: the scripts said "my take" too often.
• 5:37
  Now a check catches that, so the same note doesn't reach me twice.
• 5:41
  Agents and extra steps such as a second model hunting for
• 5:45
  mistakes can raise accuracy before the work reaches you.
• 5:48
  But someone always approves every episode.
• 5:51
  And the review stays your job.
• 5:54
  And the models keep moving.
• 5:56
  This fall, new ones arrive about monthly.
• 5:59
  In February, I wrote that what was a limitation yesterday is a capability today.
• 6:05
  One example I gave was computer use.
• 6:08
  At the start of 2025, agents couldn't use a computer well.
• 6:12
  About a year later, that was routine.
• 6:15
  So what's published about the models' limits wears out fast,
• 6:18
  and this workflow could change tomorrow too.
• 6:21
  The only way to know what the latest models can do is to test them yourself.
• 6:26
  A new model is your cue to test again.
• 6:29
  Can it take on more of the work?
• 6:31
  Should your steps change?
• 6:32
  That's the moment to improve your workflow.
• 6:35
  We're always experimenting at WorkHacker.
• 6:37
  Keeping up can wear you out, though.
• 6:40
  My approach is to watch closely but build deliberately.
• 6:43
  Take what matters for your problems, and let the rest go.
• 6:47
  Still, we should all be experimenting with AI.
• 6:50
  That goes for a podcast and for everyday work.
• 6:53
  It doesn't have to be a major part of your work.
• 6:56
  You don't have to voice your show with a synthetic voice.
• 6:59
  Use it to correct a line, or for a short voice-over.
• 7:02
  But I believe not trying these tools is a mistake.
• 7:05
  They're what comes next.
• 7:07
  They let us fix things and get more done.
• 7:10
  So let the machine do as much of the work as it does well.
• 7:13
  You set the direction, review every episode and own what goes out.
• 7:18
  Pick one step you still do yourself, and test the newest model on it this week.
• 7:22
  Hand the model your guidelines.
• 7:24
  Then review what comes back as a draft.
• 7:27
  I'm not saying everyone should make a podcast, though.
• 7:30
  It's a lot of work.
• 7:31
  And a lot of thinking.
• 7:33
  The people best at it have done it for a while, or really love it.
• 7:38
  If you don't love it now, AI won't make you love it.
• 7:41
  It might just make it sloppy.
• 7:43
  So have people who love the work make it for you.
• 7:46
  Whether and how to start a podcast for your business is the next episode.
• 7:51
  Whoever makes yours, this series looks at how to
• 7:54
  make a podcast that performs for your business.
• 7:57
  We take questions like these.
• 7:59
  Who actually listens?
• 8:00
  Which format fits?
• 8:02
  How do you get found?
• 8:03
  And should you use an AI voice?
• 8:06
  The show is about podcasting for business.
• 8:08
  It's also about what work looks like now that AI does more of it.
• 8:12
  And where work goes from here.
• 8:15
  Our workflow is modeled on what the frontier labs advise for working with AI agents.
• 8:20
  Their advice is on one sheet, free to download with the written version.
• 8:24
  Thanks for starting with us.
• 8:26
  Know someone thinking about a podcast for their business?
• 8:29
  Send them this episode.
• 8:30
  The written version is on workhacker.com, with every source.
• 8:34
  And if you want a podcast like this for your business, WorkHacker makes them.
• 8:39
  See you next time.