I counted the software on my edit machine last month and it came to nineteen applications. Then I looked at what actually touched an episode that week. Four. Everything else was there because I bought it during a sale, or because someone on YouTube told me it would change my life, and it did not.
So this is the honest list. Not the aspirational stack, not the affiliate link stack. The four or five things that carry every episode we deliver, why each one earned its slot, and the tools I keep almost buying and never do.
The one rule that decides everything
Before any tool goes in, it has to survive one question: does a human still watch the whole episode?
That is not a marketing line for us, it is a constraint on the toolchain. We do not auto-cut by silence detection. An editor watches the episode end to end, which means every tool in the stack has to be built for a person who is scrubbing, listening, and making judgment calls, not for a person who is queuing a batch job and going to lunch.
The moment a tool's core promise is "you will not have to watch this", it is out. Those tools are optimized for throughput, and throughput is not what a host is paying for.
Most editing software is now sold on how little of your own show you have to sit through. That is a strange thing to want, and it is the exact thing I refuse to buy.
Premiere Pro, and the boring reason why
Premiere is the spine. Not because it is the best NLE by some objective score. Resolve has a better color page. Final Cut is faster on the same hardware. I know both.
Premiere stays because of multi-cam and because of handoff. When we cut a multi-cam editing job with three or four angles, the multi-cam source sequence workflow is predictable in a way that matters more than raw speed. I can hand a project to another editor and they will find things where they expect them. Predictable beats clever when more than one person touches the file.
The other reason is unglamorous: our project template. Bins in a fixed order, tracks labeled the same way on every show, adjustment layers already stacked. A new episode opens looking like the last episode. That template took a weekend to build and has saved hundreds of hours since. If you are curious how far we take that idea, it is the same instinct behind our file naming conventions, which sound like nothing and quietly protect the entire schedule.
Would I rebuild the studio on Resolve tomorrow if I were starting fresh? Maybe. But switching a working pipeline costs more than the switch appears to cost, and I have watched people lose a quarter to it.
Audio: iZotope RX, used with restraint
RX does one job in our stack and does it better than anything else: fixing what should not be there. Air conditioner hum. A chair creak. The click of a ring against a desk. Spectral repair on a single bad breath is the closest thing to magic in this business.
What RX is not, for us, is a tone shaper. I see editors run everything through six modules and hand back a voice that sounds like it was recorded inside a plastic bag. Clean is not the same as good. A slightly roomy voice that sounds like a person beats a surgically dead voice every single time.
Our order of operations is fixed:
- Repair only the specific problems, one at a time, not a blanket pass
- Level with a gentle compressor, not a limiter doing compression's job
- Match loudness across speakers before touching EQ
- Leave the room tone in
That last one is deliberate. Silence in a podcast is not empty, it is texture, and stripping it makes the whole thing feel synthetic. I have written more about that in the case for leaving pauses in, because it is the note I give junior editors more than any other.
Transcripts, as a map and nothing more
We pull a transcript on every episode. Not to edit from it, to navigate with it.
The difference matters. Text-based editing tools sell you on cutting the transcript and letting the video follow. It works, and it produces a show that reads well and sounds wrong. Speech is not prose. A sentence that looks redundant on the page is often the sentence where the guest finally relaxes.
So the transcript is a map. An editor uses it to find the moment at 34 minutes where the good story starts, then goes to the timeline and cuts by ear. We also use it to pull vertical clips later, because scanning text for a quotable thirty seconds is much faster than scrubbing an hour of footage twice.
Add-on clips are $25 each, and the honest reason they are cheap is that the transcript already did the searching.
The graphics side, and why templates win more often than they should
I used to build custom lower thirds per show. Real design work, per client, from scratch. I stopped for most shows, and I think I was wrong for about two years.
Here is what changed my mind. A custom animated title sequence gets watched once by the host, complimented, and then skipped by roughly everyone forever. Meanwhile a clean, readable name card that appears at the right moment on every guest, every week, does actual work. Consistency reads as professional. Novelty reads as a studio that has time on its hands.
We still do bespoke intro and branding work when a show has a real reason for it, usually because the host is building a brand beyond the podcast. But my default recommendation now is a strong template system, and I laid out the whole argument in templates versus custom graphics if you want the long version.
The exception is thumbnails. Thumbnails are not decoration, they are the click. That is a different discipline with different rules, and it is why our sister studio ClickTheory handles them separately at 8 for $250 rather than treating them as an afterthought on the edit invoice.
What I do not use, and why
Some of these will be unpopular.
AI auto-editors. Not because the technology is bad. Because the mistakes they make are the expensive kind. They cut on silence, and silence is where the meaning lives. They also fail invisibly, which means someone has to watch the whole thing anyway to catch it, which was the work in the first place.
Cloud collaborative NLEs. Tempting for a remote studio. I am in Dubai and every client I have is somewhere else, so on paper this is my exact use case. In practice, upload times and proxy management ate more hours than the collaboration saved. We use plain file transfer and a strict folder structure instead, and it is faster.
Project management software for a two-week pipeline. We tried three. Every one of them turned into a place where tasks went to be beautifully organized and never done. A shared board with four columns does the job.
Loudness plugins that promise broadcast compliance in one click. They hit the number and flatten the performance. Hit the number by mixing properly.
The stack only matters because the schedule is fixed
Here is the part people miss when they ask what software we use. The software is downstream of the promise.
We deliver a first cut in 48 hours and revisions the next working day. That deadline is what forces every tool decision. A tool that is 20 percent better but adds an unpredictable render step does not survive contact with a fixed turnaround. Predictability beats capability when you have made a promise in writing. I broke down exactly how the timeline holds together in how we hit 48 hours every time, and almost none of it is about software.
It also shapes how we price. Starter is $199 an episode for a single camera at 1080p with one revision. Studio is $299 for two to four cameras, 4K, and unlimited revisions. Both cover up to 60 minutes, 30 minute shows cost less, 90 minute shows cost more. Packs of four save 15 percent and monthly saves 25 percent.
Flat per episode, not hourly, and that is a deliberate choice tied to the stack. If I billed hourly I would have a financial incentive to keep slow tools around. I would rather eat the cost of my own inefficiency and go find faster ones. There is a longer argument about that in hourly versus flat rate editing if you are weighing the two models for your own show.
How to build your own version of this
If you are assembling a stack, do not start with the tool list. Start with the constraint.
Write down the one promise your show makes. Weekly on Tuesdays. Always under 40 minutes. Always three angles. Then buy only what protects that promise, and be genuinely rude about everything else.
A short test I use before adding anything:
- Does it remove a step, or does it add a step that feels productive?
- If it disappeared tomorrow, could I still ship on time?
- Does it make the show sound more like the host, or less?
- Would I still pay for it at three times the price?
Most tools fail the fourth question instantly, which tells you how much they were really doing.
Nineteen apps installed. Four that matter. I do not think that ratio is unusual, I think most of us just never audit it. Open your applications folder this week and ask which ones touched your last episode. The answer is usually uncomfortable and always useful.
If you want to see how this runs on a real show rather than in theory, our how it works page walks through the whole handoff. And if you just want to argue with me about Resolve, my inbox is samy@pixelsamy.com and I do read it.