Why I built Phillipe.

Updated September 24, 2026 · Product details checked against Phillipe’s current workflow.

Editing takes longer than filming

Short-form video looks simple from the outside: one person, one phone and less than a minute of talking. The filming usually is simple. You put the phone at eye level, say the idea a few times and stop recording.

Then the edit begins. You scrub through the clip to find the take where you did not stumble. You cut every pause and every “um.” You remove the sentence you said three times. You fix the captions word by word so they land with your voice. You look for supporting clips and place them where they actually show what you are saying. You choose music that does not fight your voice. Then you export a vertical file and check that nothing drifted out of sync.

None of those steps is hard on its own. Together, they take longer than the filming did. For anyone trying to post consistently, that is where the week goes. Phillipe exists to solve that one problem.

I found the bottleneck the way I find it in businesses

My consulting work starts with a simple sequence: observe how the work moves, find where time and quality are lost, define the result and only then choose the technology. I explain that sequence in more detail in what an AI operator actually does.

When I applied the same sequence to making videos, the answer was clear. Ideas were not the bottleneck. Most people who make content have more ideas than they can publish. Filming was not the bottleneck either, because a phone is enough. The bottleneck was the mechanical work between raw footage and a finished post.

That diagnosis also told me what not to build. I did not want another tool that generates ideas or scripts. That part should stay with the person whose face is on the video. I wanted something that removes the mechanical part of the edit without taking away the decisions.

Great edits. No prompting.

I kept running into two kinds of editing tools. The first kind gives you a faster timeline. That is useful if you are an editor, but it still assumes you will do the edit yourself. The second kind asks you to describe the edit in a prompt. That sounds easier until you notice that writing a good prompt for a video is its own skill, and a vague prompt gets you a vague edit.

I wanted neither. The line on the Phillipe home page, “Great edits. No prompting.”, is a product decision, not a slogan. You should be able to upload your footage, pick a look and get a real edit back without writing anything. If you have something specific in mind, you can say it. Instructions are optional, not the price of admission.

What Phillipe does

Phillipe is an AI video editor for short-form video: vertical for Reels, TikToks and YouTube Shorts, or horizontal for YouTube. It is built for creators and personal brands who film themselves talking. Starting an edit takes four steps.

• Upload your footage. Your talking-head clip is the A-roll. Supporting clips are optional B-roll, and you can add photos and audio too. You label each video as A-roll or B-roll, and Phillipe trusts that label.
• Pick a style. Kinetic is the signature style: sculpted typography, flowing script and expressive graphic moments that give every word a little motion. Cinematic is built for storytelling. A simpler Clean style is coming, but for now it is preview-only.
• Add a little direction, if you want. On a paid plan you can leave a short note, such as “keep it under 30 seconds and use my original audio.”
• Start with one click. You see a credit estimate before the edit runs.

From there, Phillipe works in two passes. The first is the rough cut. It transcribes your speech with word-level timing, cuts pauses, filler and repeated takes, keeps the complete story and places B-roll on specific spoken beats. You can review that cut yourself, or enable auto-approval to keep the edit moving. The second pass adds the style: word-timed captions, pacing, supporting footage and music under your voice. After the first cut, you can ask for changes in chat and make the final call. The result is a finished video ready to post, in 9:16 or 16:9.

If you want to see each step in order, there is a step-by-step tutorial for editing a talking-head video with Phillipe. For the wider picture of how AI editing works, start with the complete guide to editing videos with AI.

Choose a style, keep your voice

A useful style should make the small decisions for you without changing what you mean. Kinetic starts with more expressive type and graphic moments; Cinematic gives the story a more restrained visual rhythm. Neither choice needs a technical prompt.

For example, “keep the beach shots, make the opening calmer, and leave the last sentence untouched” describes an outcome. It tells the editor what to protect as well as what to change. For more examples, see Phillipe’s guide to giving editing instructions.

You can also enable auto-approval when you want the edit to advance without waiting at each checkpoint. That is permission to continue processing, not permission to publish a video on your behalf. Review the finished result before sharing it.

What building it taught me about AI products

Ask for the one thing people know instantly

One of the most useful decisions in Phillipe is also one of the smallest. It asks you to label each upload as A-roll or B-roll. You know the answer in half a second. A model has to guess, and a wrong guess ruins the edit: your main take treated as filler, or a random clip treated as the story. Asking for one label is cheaper and more reliable than any amount of clever guessing. Good AI products ask for the context only a person has, then do the rest.

Put the checkpoint where judgment matters

The rough cut is its own step because the story is the part you should approve. If a cut removed a sentence you needed, you want to know before captions and music are built around it. My operating principle has always been “use AI to build, not to think.” In a product, that means the system does the building and the person keeps the decisions. You review the cut. You make the final call.

A blank prompt box is a hidden job

Every empty prompt field quietly hands work back to the user. Most people do not know exactly what they want until they see a first version. So the default path in Phillipe produces an edit from a style choice alone, and revisions come afterward, when you have something concrete to react to. The style is where the craft lives: each word animates in as it is spoken, a hero word carries the line, text can sit behind you in the frame and B-roll lands just before the phrase it illustrates, so the result looks designed rather than captioned. “Remove the second example” is much easier to write than a description of an entire video.

Show the cost before the work

AI work has real costs that change with every job. Transcription, model calls and rendering all grow with the amount of footage. Hiding that behind vague promises ends badly for someone. Phillipe shows a credit estimate before every edit, and corrections use credits based on the processing performed, so what you see reflects the work being done. I would rather explain an honest number than defend a surprising one.

Trust is part of the product

Creators hand over their face and their voice. Phillipe does not use your footage to train its own AI models. The site has optional first-party analytics and anonymous public-page request totals, rather than advertising pixels or cross-site advertising tracking. Payments go through Stripe. Read the current privacy notice for data processing and retention details. None of that is exciting, but each one is a reason someone can say yes.

What Phillipe is not

It is not meant to replace an editor on a documentary or a brand film. It is built for one job: turning footage of a person talking into a finished short-form video. Being specific about that job is what lets it do the job well.

It is also not unlimited. The free plan includes one finished video a month using Phillipe’s own pass. Custom instructions and corrections in chat start on the paid plans, and every edit uses credits that depend on the length of your footage.

Why this problem is worth solving

The people Phillipe is for are rarely short of ideas. They are short of hours. If it gives some of those hours back, and leaves the creative decisions with the person on camera, it is doing its job.

If you want to see how it fits into a real routine, I wrote about how I produce a week of short-form video with AI. If you are thinking about automating editing for a client, read the AI operator’s guide to automating video editing.

And if you have a clip on your phone right now, you can try it with your own footage at phillipe.ai. The free plan includes one finished video a month.