Skip to content
MakerGain
← Back to cases

Study what makes ads work, then generate your own: Starpop puts four steps in one conversation

Edited by MakerGainPublished: Article updated:

Starpop handles the stretch of performance advertising between understanding an effective ad and turning it into material ready to produce. It researches ads, then develops strategy, scripts, and storyboards; approved storyboards can lead into image, video, voiceover, and music generation in the same conversation, with final decisions still made by people. In the interview, Alexander Van Le described his previous VC-backed FinTech as too focused on terms and strategy, and too little on marketing and understanding customers. He shifted to marketing-first and later bootstrapped AI Flow Chat and Starpop. The business puts research before generation and lets other agents call its generation tools through MCP: connection and chat are free, while media generation costs credits.

Performance advertisers often face this stretch of work: browse competitors’ creative, find an ad that has run for a long time, and work out why it works. Is it the action in the first three seconds, the voiceover line, or the closing discount? After figuring that out, they still need to turn it into a script, storyboard, voiceover, and soundtrack, hand it to an editor, and go through several rounds. For a 15-second talking-head ad, for example, the distance between “I like its opening” and “we can make one with a good opening too” includes finding people, revising copy, adding music, and exporting.

That is the stretch Starpop wants to take over. Its homepage says its AI ad agent researches effective ads, turns insights into angles and scripts, then generates video, images, speech, and music for a brand (Starpop[1]).

Four stages, and an “approved”

The AI Creative CMO page divides the workflow into Research→Strategy→Production→Iteration, with each stage generating structured context for the next. Within the same conversation, an approved storyboard can lead into image, video, speech, and music generation while retaining the brief and brand context (Your AI Creative CMO | Starpop[2]).

Image: From the product website[1], obtained on 2026-11-18; original image retained. An official demonstration is not an independent hands-on test.

“Approved” is worth a pause. In this storyboard example, it means a person has reviewed and signed off before the machine moves into production. The FAQ on the same page says Starpop does not replace human creative judgment: people retain control over the brief, approvals, and final decisions. That is more restrained than promising fully automatic finished videos, and it locates the product clearly. It does not decide for you; it organizes and connects the work around those decisions.

Understand customers before discussing generation

In the interview, Alexander Van Le looked back on a VC-backed FinTech he co-founded: four co-founders, zero revenue at pre-seed, and poor GTM execution—go-to-market, or taking the product to market. The team spent too much time on terms, strategy, storytelling, and slides, and too little on marketing and understanding customers (Indie Hackers[3]). His conclusion was that building products now should be “marketing-first.”

Starpop’s choice to put research before generation points in the same direction as that “marketing-first” stance. What can be confirmed is that he later bootstrapped AI Flow Chat and Starpop. The AI Creative CMO page’s line, “Most tools start after the important decisions have already been made,” is consistent with that direction, too.

The homepage excerpt also includes “Message from founder David,” whereas the interview lists Alexander Van Le as a founder. It does not detail the full founding team or division of work, but this is evidently not a one-person project.

Connection is free; generation is billed

Another interesting choice is that Starpop has made itself a tool for other people’s agents. Through MCP, ChatGPT, Claude, Gemini, Grok, terminal-connected Codex and Claude Code, and Hermes running on a schedule can call Starpop to generate images, video, voiceovers, and music, with outputs saved in the Starpop asset library (Starpop Updates[4]).

Connection and chat are free on every plan, including the free plan, and do not consume Starpop agent credits. Media generation is charged at the usual credit rates. The limits are specific: at the time, Claude Free allowed only one custom connector; Gemini custom apps required a personal Google account with Apps Activity enabled, with work/school accounts not yet supported. Users control permissions: choose the organizations an agent may use at login, and revoke access under Settings → Connected agents whenever needed, with immediate effect.

The bill arrives before the finished video

The pricing page lists Solo at $50/month, including 500 credits a month and 1 user, and Teams at $100/month, including 1,000 credits and 2 included users. Both include all AI models and a handpicked ad library, with top-ups for extra usage. A separate done-for-you finished-video service targets brands or agencies spending more than $50,000 USD a month on ads and requires booking a call. The page does not say that $50,000 is paid to Starpop (Starpop[1]).

The terms make customer billing clearer. Subscriptions are prepaid monthly or annually and renew automatically. With automatic top-ups enabled, falling below the configured balance threshold triggers a charge for the selected credit pack. Credits are consumed when generation starts, whether or not the user keeps, downloads, or uses the result. Failed generation caused by an error on Starpop’s side is credited back. Unused plan credits generally do not roll over and are removed when the subscription ends; separately purchased credit packs remain. Cancellation stops future renewals and takes effect at the end of the current billing period; the plan and remaining credits can still be used during that period. Subscription fees, renewals, credit packs, and automatic top-ups are nonrefundable unless the law requires otherwise (Starpop terms[5]).

The homepage says “No hidden lock-in” and “top up instead of upgrading.” Both can coexist with those terms, but predictable spending comes with conditions: automatic top-ups charge when the balance drops below a threshold, unused plan credits do not carry into the next billing period, and refunds generally are not available. Put those rules into a budget, and total spending is not as predictable as a flat monthly fee. The terms also say the service may use third-party AI/ML and media-processing providers, does not guarantee outputs, and makes users responsible for reviewing and verifying generated content before use. Alongside “charged when generation starts,” that makes the self-service subscription’s economics for the customer clear: it does not charge for delivery of a finished video; it spends credits when generation begins. What users buy is attempts.

Price comparison can only go so far. Creatify’s pricing page lists Pro at $38/month, the equivalent monthly price with annual billing selected by default; switching to monthly billing gives $55/month. It includes 280 credits a month, 1,500 realistic AI actors, and 500+ ad templates. Its FAQ says unused credits can roll over for one additional month (Creatify[6]). Both offer monthly billing, but their credits are not the same unit: different models and resolutions consume different amounts, so placing 280 beside 500 does not establish which is cheaper. Rollover differs, too: Creatify keeps credits for an extra month, while Starpop removes expired plan credits. Both have MCP. Starpop explicitly says connection and chat are free on every plan, with normal credit charges for media generation. Creatify’s comparison table says “Every plan includes ... MCP,” but the visible text on that page does not explain how MCP connection is billed.

Revenue is only a combined figure

There is just one combined revenue figure: AI Flow Chat and Starpop together at MRR of $20k/month. There is no standalone Starpop figure, paying-customer count, refund or churn data, or revenue period, so ARPU cannot be calculated. MRR is not profit; without COGS, third-party model and media-processing fees, or credit unit costs, profit remains unclear. The website’s “Loved by $10M+ brands” is promotional language. “100+ fast-growing DTC brands” and “Track 118 brands running AI ads” describe the scope of the ad library it tracks, not its customer count.

Starpop has turned “first see what works” into a capability rented by the month. If an ad team really puts research before generation, that money buys an earlier view of which angle is worth filming. If it generates only a few videos occasionally, the unused portion of Solo’s 500 credits will not carry into the next billing period. This billing structure rewards repeated attempts, not occasional ones.

Continue reading

AI video
Break down the script before opening AI: how Layer8 sells films

Layer8 is an AI film studio co-founded by Ali, Kevin, and James. It breaks down scripts before opening AI tools, with directors, writers, artists, and AI experts working together while filmmakers retain visual judgment. After the team sent early samples to contacts, California Highway Patrol became its first paying client. An earlier attempt at templated ads failed to take off, and a trademark dispute prompted the Layer8 name. Agencies, production companies, and brands buy finished films.

AI video
Putting pictures on the beat: neural frames turns synchronization into a business

neural frames handles the synchronization involved in setting music to pictures: it separates elements such as drums and vocals so generated visuals land on the beat. Founder Nicolai Klemke began with general text-to-video, then followed advice to narrow the product to music videos because the product lacked a well-defined use case. The small team buys models and APIs externally, concentrating its work on separating songs and passing parameters to the models.

AI video
Free video ideas, finished videos by the minute — how Subscribr runs a YouTube workflow

Subscribr handles the long-form YouTube workflow from ideas and scripts to thumbnails and finished videos. Video ideas and scheduling are free, and finished video is priced by the minute. Founder Gil Hildebrand left a funded startup and bootstrapped Subscribr, which launched in 2024. He sold lifetime licenses before the product was built, with delivery promised within 60 days and refunds available through two weeks after delivery. The tool later grew from ideation and scripting to the full production workflow.

AI video
Zero monthly fees, with pricing by the second: VOSU unbundles AI video, then asks who will deliver it

VOSU aims to take a campaign from concept to an editable workflow: users describe the requirement, and agents plan concepts, build prompts, select models and generate variants in bulk. Its three full-time creator founders chose pay-per-use billing and non-expiring balances, while betting growth on a marketplace connecting brands with AI creative talent.