AigentLab®Field Guide

Storyboard
First.
Ship
Tomorrow.

Adapted from “Claude Can Do MOTION Now And Replaced Me ($880 in One Day)”
Board first.
Skill once.
Ship next day.
No. 34
10.20.2026
Executive Summary / Video Takeaways

One Day.
$880 Kept.

A freelancer sells 30-second SaaS launch videos built entirely in code by Claude Opus 5.5. No After Effects. No keyframes set by hand. One $1,000 job, one day of work, $880 kept after fees.
The core ideaClients pay for a launch-ready video by tomorrow. The speed comes from two assets working together: a storyboard written before any render, and a reusable motion skill holding every timing, motion, and spec rule. A Fiverr motion designer given the same brief, files, and budget delivered on day seven of a five-day deadline.
01

The paying lane

The creator names SaaS and tech launch motion as the best-paid style, at $1,500 to $10,000 for 30 seconds on the open market. He says the lane accepts new freelancers with an empty portfolio.

02

Search wide on Upwork

Run several searches per session: SaaS launch video, product launch video, UI animation, app promo video, SaaS explainer. Applying to everything burns connects and returns zero replies.

03

Four filters before you apply

Payer: payment verified, $10K or more spent, hire rate above 50 percent. Freshness: posts under 30 minutes old. Competition: under 15 proposals and nobody interviewed. Money: a recent funding round on the company website. Fail one, skip the post.

04

The exception

Posts at $5 to $10 an hour from clients with zero hires often come from first-time buyers unsure what to type in the budget field. Verify the company, apply anyway, and quote your real price.

05

The proposal

The client sees two lines in the preview. His message opens on the client, repeats the exact keywords from the post, and follows one structure: their task, his solution, a clear deadline. A concept replaces a portfolio. A small first milestone lowers the client's risk. Five proposals, three replies within hours.

06

The job

A startup building an AI video recording tool needed a launch video before going live. The UI already existed in Figma. Quote: one day, $1,000, 30 seconds.

07

The control test

On Fiverr he skipped every designer without UI work, since character animation and UI motion are separate skills. He hired Alex: 47 reviews, $1,000 for 30 seconds in five days. Same brief, same Figma files.

08

The motion skill

Story: attention, problem, fix, payoff. Timing: key items hold about 1.5 seconds. Motion: nothing linear, half-second entrances with slight overshoot, faster exits, a press on each click, curved cursor paths, nothing fully stops. Parent groups, offset starts, overlapping moves. Specs: 1920 by 1080, 60 fps, vertical safe zones.

09

Two strict rules

Storyboard first, then animation. If the work is motion, teach the motion: UI components, easing curves, a baseline. He tested setups for more than 10 hours before this project.

10

Inputs and one prompt

Client brief, scene-by-scene storyboard, brand colors and fonts, background images, and Figma UI exported as SVG. SVG stays sharp when the camera zooms. A screenshot turns to blur. One prompt builds the board scene by scene in code, rendered frame by frame.

11

Render one broke in six places

Overlapping text, a fast pan, a layout off the canvas, a Times New Roman fallback, a counter jumping to 200 in one frame, stacked cards. Each break had one cause and a one-sentence fix. Every fix now lives in the skill.

12

The no-storyboard test

Same assets, skill, and prompt without a board: random scene order, everything moving at once, no story. With the board: problem, pain, fix, payoff.

13

Revisions in minutes

Changing a counter from 200 to 999 in After Effects means opening the project, fixing keyframes, and re-rendering. In code, one line changes and a new version renders in a couple of minutes.

14

The money

$1,000 paid the same day as delivery. A $20 subscription. Under one day of work, mostly on the storyboard. $880 kept after Upwork fees. Optional: ask a happy client to cover the platform fee, about $100 more.

15

The real offer

A SaaS company needs a launch video, then a feature update, then an ad, then the next launch. The freelancer who ships fast and never misses a deadline becomes the go-to motion person and an anchor client.

Render one: symptom, cause, fix
Symptom
Cause
Fix sentence
Layout off canvas
Figma screens at 1440 px wide, video at 1920
Lock the canvas to 1920 by 1080 and position everything from the center, not the edges.
Wrong font
Font not loaded before frame one
Load the font before rendering anything.
Counter jumps
No count duration
Make the count up last about one second.
Cards stack
No stagger
Bring cards in one by one, a few frames apart, each landing before the check mark.
Camera too fast
Hard start and stop
Slow the pan and give the move an easy ease.
New Video Concept / For a new professional's channel

Storyboard First.
Ship Tomorrow.

Clients buy tomorrow. Not code. Not keyframes.
Core Message

You do not need After Effects, a motion reel, or five years of experience to sell a launch video. You need a storyboard written before the first render, a motion skill holding every rule you learn, and a pitch built around the client's product instead of your portfolio. Write the board, teach the motion once, fix every break by cause, and you become the freelancer who delivers by tomorrow, every time.

Opening Premise

You know Figma, or Canva, or how a product screen should feel. You have never set a keyframe. Startups post launch-video jobs every hour, and each post fills with proposals from people holding five-year reels. The standard advice says build a portfolio first, which takes months you lack.

In “Claude Can Do MOTION Now And Replaced Me ($880 in One Day),” the creator landed a $1,000 job with an empty portfolio and a concept for the client's own product, then built the video in code in under a day. His first render broke in six places. He fixed each break with one sentence, saved every fix as a rule, and the rules became a skill. The tool did not carry the job. The board and the skill did.

Opening question: If a startup hired you tonight for a 30-second launch video due tomorrow, what would you build first?

LaneOne paying niche, four filters per post.
PitchTheir task, your concept, a deadline, a small milestone.
BoardFour beats, every scene timed and framed.
SkillTiming, motion, specs, and every fix saved.
RepeatOne-line revisions and the next job from the same client.
01
Teaching Point 01 / Lane

Pick the lane. Filter the post.

Volume burns connects. One niche with a known price band, plus four checks on every post, puts your proposal in front of a client who pays and still reads. The creator sent five proposals and heard back from three within hours.
Search wide
Run five searches in one sitting: SaaS launch video, product launch video, UI animation, app promo video, SaaS explainer. Same time every weekday, 8:30 a.m. ET.
Check the payer
Payment verified, $10K or more spent, hire rate above 50 percent. All three, or skip.
Check freshness
Sort by newest. Apply to posts under 30 minutes old, while the client still sits in the inbox.
Check competition
Open activity on the job. Under 15 proposals and zero interviews means nobody has won yet.
Check the money
Open the company website. A recent funding round or a dated launch signals budget plus deadline.
Log every post
One sheet: link, four checks, applied yes or no, reply yes or no. Ten rows a week shows your real reply rate.
Use the exception
A real company with zero hires and a $5 to $10 hourly placeholder gets a proposal at your full price.
Rule
Four checks pass, or you skip the post.
Common mistakeApplying to every post in the feed, then deciding freelancing does not work after twenty silent proposals.
Week one checkTen posts logged, five proposals sent, reply rate written down.
02
Teaching Point 02 / Pitch

Pitch the concept, not the portfolio.

The client sees two lines before deciding to open your proposal. Spend both on their product, in their words. A concept for their launch shows the work before the work exists, and a small first milestone takes the risk off their side.
Line one
Copy two or three exact keywords from the post (product launch video, Figma, UI animation) and name their product.
Line two
The concept in one sentence: the problem their user feels, and the moment the product fixes the problem.
The beats
Four beats, one line each, with timestamps: problem by 5 seconds, fix by 15, payoff by 25, logo at 30.
The deadline
Storyboard today for approval. First cut within 24 hours of approval. Revisions the same day.
The milestone
Storyboard approval as milestone one, a small share of the fee. Low risk for them, a paid start for you.
The close
One question about their files: are the screens in Figma, and will you share them as SVG?
Proposal template (copy and edit)

Your [product] launch video, built around [keyword 1] and [keyword 2]. Concept: [user problem] in the first 5 seconds, [product moment] fixes the problem by second 15, [payoff] lands by 25, logo at 30. Plan: storyboard today for your approval, first cut within 24 hours of approval, revisions the same day. Milestone one: the storyboard, $[amount]. Are your screens in Figma, and will you share them as SVG?

Rule
No sentence about you comes before a sentence about them.
Common mistakeOpening with “Hi, I'm a motion designer with X years of experience.” Every rival opens the same way.
Week one checkThree proposals sent inside 30 minutes of the post going live, each with a concept for the client's own product.
03
Teaching Point 03 / Board

Storyboard before you animate.

Same assets, same skill, same prompt, no storyboard: scenes in random order, everything moving at once, no story. The board is the part of the brief Claude has no way to guess. The creator spent most of his one day on the board, not the render.
Four beats
Attention, Problem, Fix, Payoff. Every scene serves one beat. A scene serving none gets cut.
Split the clock
For 30 seconds: Attention 0 to 5, Problem 5 to 14, Fix 14 to 25, Payoff 25 to 30.
Five fields
Each scene gets SCENE, ON SCREEN, HOLD (seconds), CAMERA, and CUT TO.
Hold the key line
Every line a viewer must read holds about 1.5 seconds. Long enough to read, short enough to keep pace.
Name the camera
Push in, pan, or hold. Name the start and end framing, never “something dynamic.”
Get approval
Send the board to the client as milestone one. Approval comes before render one.
Fig. 01 / Sample board rows
Scene
On screen
Hold / Camera / Cut to
01 Attention
Counter of open tabs climbs to 37
1.5 s / Hold, slow push / Tab chaos
02 Problem
Cursor copies between tabs, hesitates
3.0 s / Pan left to right / Product UI
03 Fix
Product dashboard, cards land one by one
4.0 s / Push into button / Result
Rule
No render before an approved board.
Common mistakeRendering first to “see what happens,” then rebuilding the story inside the code one revision at a time.
Week one checkOne board approved with zero scene changes after render one.
04
Teaching Point 04 / Skill

Teach the motion once.

The skill holds what a senior motion designer carries in their head. Write the skill once, and every project starts where the creator's second hour started, not at zero. Feed the inputs right and half the bugs never appear.
Story
Attention, problem, fix, payoff. Cut any scene serving none.
Timing
Key lines hold about 1.5 seconds. Counters count up over about one second.
Motion
No linear moves. Entrances about 0.5 seconds with a slight overshoot. Exits faster than entrances. A small press on every click. Cursor paths curve like a hand.
Flow
Group related elements under one parent and move the group. Offset each element a couple of frames. Start the next move before the last one settles. Nothing fully stops.
Specs
1920 by 1080, 60 frames per second, safe zones so a 9:16 cut keeps every key element.
Inputs
UI exported from Figma as SVG, never screenshots. Brand colors and fonts in one file. Fonts loaded before frame one. Canvas locked, positions measured from center.
Never
No Times New Roman fallback. No element entering without easing. No two key lines on screen at once.
LINEAR: STARTS HARD, STOPS HARD EASE + OVERSHOOT, OFFSET 2 TO 3 FRAMES 0 S0.5 S 0 S0.5 S
Fig. 02. One element on a straight line versus three grouped elements easing in, overshooting, and settling a few frames apart.
Rule
A fix you give twice is a rule you forgot to save.
Common mistakeTyping the same correction into every new project because the fix lived in last week's chat.
Week one checkOne skill file with seven sections and every fix from render one saved as a one-line rule.
05
Teaching Point 05 / Repeat

Debug by cause. Sell the turnaround.

Most people quit at render one and decide AI motion does not work. Each break has one cause and one sentence of fix. Once the bugs clear, revisions take minutes, and minutes are the product: a launch video, then a feature update, then an ad, then the next launch.
Watch with a pad
Watch render one start to finish. Write a timestamp, what you see, and what to change for every break.
Bugs first
Fix layout, fonts, timing, and stacking before any taste note. One sentence per fix.
Taste second
Then the details: a glow on a climbing number, a cursor pausing between tabs. Still one sentence each.
Save both
Every bug fix and every taste note goes into the skill as one line with the example timestamp.
One turnaround
First cut within 24 hours of board approval. Revisions the same day. Put both in every proposal.
Name the next video
With the final files, send one line naming the next asset their launch calendar needs: feature update, ad, or a vertical cut.
Fig. 03 / Why the turnaround holds
Client ask
Timeline tool
Code plus skill
Counter 200 to 999
Open project, find layer, fix keyframes, re-render
One line changed, new render in minutes
New brand color
Recolor every layer by hand
One value in brand.md
Vertical 9:16 cut
Rebuild the comp
Safe zones already in the skill
Rule
Bugs first, taste second, then save both.
Common mistakeRewriting the whole prompt after render one instead of fixing six causes one sentence at a time.
Week one checkRender two fixes every break from render one, and the skill holds at least six new rules.
Translation / One new professional, one month

Kai's First
Launch Video.

Kai, a first-year freelance product designer. Strong in Figma, zero motion experience, no motion reel. Goal: two launch videos a month at $750 to $1,000 each.
Lane
SaaS launch videos on Upwork. Five searches every weekday at 8:30 a.m. ET. One sheet with the four checks per post.
Pitch
The proposal template with a concept written for each product. Milestone one: storyboard approval at $150.
Board
Four beats, nine scenes, five fields per scene. Client approval before render one.
Skill
launch-motion.md seeded with Story, Timing, Motion, Flow, Specs, Inputs, Never. Six rules added after render one, three taste rules after render two.
Repeat
First cut within 24 hours of approval. Revisions the same day. Delivery note names a 15-second vertical cut for the client's social launch.
Sample month one (planning figures)
Measure
Weeks 1 to 2
Weeks 3 to 4
Posts logged
14
8
Passed four checks
6
3
Replies / jobs won
2 / 1
2 / 1
Hours per video
12 (spec), then 9
6
Rules in skill
9 to 14
14 to 17

Planning estimate: two jobs at $900 average, $1,800 gross. About $180 in platform fees and a $20 plan. Hours per video fall from about 12 to about 6 as the skill grows. Log four jobs and replace every number with yours.

Copy and paste / Edit the brackets

The Prompts.

01 / Skill setup

Create a skill named launch-motion from the rules below. Sections: Story, Timing, Motion, Flow, Specs, Inputs, Never. Write each rule as one line. Confirm the total rule count when you finish. Rules: [paste the starter below].

02 / Build

Using the launch-motion skill, build this storyboard as a 30-second launch video, scene by scene. Canvas 1920 by 1080, 60 frames per second. Load every font before the first frame. Use the SVG files in /ui, never screenshots. Position everything from the center of the canvas. Report each scene's start and end time. Storyboard: [paste board].

03 / Fix and save

Apply these fixes: [timestamp, what I see, what to change]. Then add each fix to the launch-motion skill as one line with the example timestamp. Re-render and list the rules you added.

04 / Revision

Change [element] from [old value] to [new value]. Touch nothing else. Re-render and confirm the total duration stays at 30 seconds.

Skill starter / launch-motion.md
Story
Four beats: attention, problem, fix, payoff. Cut any scene serving none.
Timing
Key lines hold 1.5 s. Counters count up over 1 s. Logo holds 2 s at the end.
Motion
No linear easing. Entrances 0.5 s with slight overshoot. Exits faster. Press on every click. Cursor moves on arcs.
Flow
Parent groups move together. Children offset 2 to 3 frames. Next move starts before the last settles. Nothing fully stops.
Specs
1920 x 1080, 60 fps. Keep key elements inside a centered 9:16 safe zone.
Inputs
SVG from Figma only. Brand colors and fonts from brand.md. Fonts load before frame one.
Never
No fallback fonts. No two key lines at once. No element positioned from the canvas edge.
Counterpoint / Read before you pitch

Counterpoint.

One project, one test, one seller's view. Use the method and keep your own receipts.

One job, one rival.

One hire against one Fiverr designer proves little about the market. Judge your own numbers across five jobs.

The $880 hides the hours.

More than 10 hours of setup tests came before the one-day build. Count your first job's hours honestly, spec work included.

“Same quality” is the seller's verdict.

The creator judged both videos and asked viewers to vote. Ask your client what they would change, and log the answer.

“Clients have no idea.”

The hook frames a hidden method as a selling point. Sell the outcome, answer tool questions honestly, and read the platform terms on AI work.

Connects and fees cost money.

Upwork charges for connects and takes a fee on every payment. Track cost per proposal next to reply rate.

Fee shifting has rules.

Asking a client to cover the platform fee changes your quoted price. Keep payment on the platform and state the full price up front.

The render stack goes unnamed.

The video never names the code library or renderer behind the frames. Budget setup time to choose and test one.

The skill is a lead magnet.

His free skill carries his look and his fixes. Start from his rules if you like, then replace them with rules from your own renders.

The low-budget exception cuts both ways.

A first-time buyer at $10 an hour often has a $10 budget. Verify the company and quote once. Do not negotiate down.

Model names age fast.

Opus 5.5 is today's tool. The lane, the pitch, the board, the skill, and fix by cause carry over to any model.

Closing Challenge / Seven days, one launch video

Board First.
Skill Once.
Ship Tomorrow.

One spec video, one skill, three proposals. By day seven you hold a reel of one, a rule file, and your first reply rate.
D1

Log the lane.

Run the five searches. Log ten posts in the sheet and mark the four checks on each.
D2

Write the skill.

Create launch-motion.md with the seven starter sections. Run the skill setup prompt.
D3

Board a spec video.

Pick a real product you admire. Write a nine-scene storyboard with all five fields per scene.
D4

Render one.

Export the UI as SVG and run the build prompt. Watch with a pad and timestamp every break.
D5

Fix and save.

Bugs first, taste second. Save every fix to the skill. Render two.
D6

Pitch three.

Send three proposals with a concept for each product, inside 30 minutes of posting. Link the spec video.
D7

Count and cut.

Count hours, rules, and replies. Delete one rule you never used. Decide: pitch more, or train the skill another week.
Clients buy the next day. Build the system to deliver one.
AigentLab®No. 34

Join The
AigentLab.

Field guides, prompts, and skill files for solo builders. Every issue turns one video into one system you run this week.

Join the AigentLab ↗
Build alone.
Not from scratch.
Storyboard First.
Ship Tomorrow.