Kajabi's first monetized AI-generation product — turning existing creator content into marketing assets to address one of the company's largest churn drivers.
AI input intake, generation states & editable outputs — AI as a tool creators tweak, not the final author
Co-authored product strategy with PM/PMM; co-led roadmap, pricing & the pre-beta → beta pivot; ran formal research with Research, PMM & PM; ran my own user testing that shaped scope (e.g. YouTube import)
Business outcome
First monetized AI-generation NRR line — GA Mar 2024; 131 paid add-on users in ~2 weeks (109% of March goal); 17.2k marketing assets; 43% of beta creators replaced outside tools.
Design outcome
AI as a tool, not the final author — every AI highlight and generated asset stayed editable; patterns reused across Kajabi's AI portfolio.
My role & constraints
Lead Product Designer — end-to-end on a net-new product line, from problem framing through pre-beta, beta, and GA. Co-authored the Creator Studio strategy with PM and PMM. Cross-functional team of 11: 1 designer, 1 PM, 1 PMM, 2 researchers, 5 eng, 1 QA.
What I owned individually:
End-to-end design across pre-beta, beta, and GA
Assistive AI — not autonomous AI. I drove the principle that creators must be able to edit what AI created or highlighted (including transcript highlights), so the model stayed a tool they could work with and tweak — never the ultimate content producer.
Solo user testing — I ran my own sessions with creators (on top of formal studies with Research). Findings directly changed the roadmap — including YouTube import, because creators' best footage already lived there, not only on their desktop.
The AI input intake — accepts any media file (video, audio, blog, PDF, web page) in one flow, without asking creators to think like prompt engineers
Generation states and feedback patterns (loading, success, error, retry) — now reused across Kajabi's AI surfaces
Editability patterns end-to-end: AI transcript highlights could be added or removed; every generated output stayed editable through light editing and, at GA, Adobe Express
Co-ran 4 research studies and creator interviews with Research, PMM, and PM; co-wrote interview scripts and synthesized findings into design direction
What I co-led with PM:
Product strategy (co-authored) and roadmap prioritization — including the pre-beta → beta scope pivot
Formal user research with Research, PMM, and PM — prioritized questions and translated findings into roadmap direction
Pricing model — 1 free project per month plus paid usage, Kajabi's first monetized AI-generation NRR line
Constraints:
Net-new product line — no codebase precedent, no DS components, no patterns for AI-rich-input flows
Creator skill range from total novice to seasoned marketer — the flow had to serve both
Pre-beta → beta pivot meant re-scoping mid-flight without losing team momentum
Third-party video tooling — transcription, clipping, and light edits ran through Pictory, so depth was capped by what the vendor could do
"Creators shouldn't have to accept whatever the AI spit out. If it wrote it or highlighted it, they had to be able to change it."
Design principle I held from pre-beta through GA
Business context
One of Kajabi's largest churn drivers is creators who never earn. 70% of creators who leave never earned on the platform at all. They're experts in their domain — but ~60% aren't highly confident in marketing, and even with seven funnel templates, they still spend 8+ hours building one from scratch. 3.8% monthly churn — and 85% of it comes from the unprofitable segment. Earners rarely leave (8% of churn earned $1k+ in six months). Earn = stay.
70%
of churned creators never earned on the platform
60%
of creators aren't highly confident in their marketing
65%
lower churn when creators gain 1 contact every 30 days
Creators lacked the time, tools, and confidence to market — not ambition.
Prior Kajabi AI features already moved activation metrics (course success, landing-page engagement, content assistant). The pattern: creators want help where the learning curve is steepest — marketing — but each feature only took a 3–10 word prompt, capping personalization.
D7 course success
7% → 12%
D7 landing-page success
16% → 25%
"Trial and do nothing"
43% → 30%
Prior AI features weren't just working — the trend was compounding. Creator Studio's bet was on richer inputs to steepen the same curve.
Creator Studio's answer was richer inputs: start from content creators already have, so the output can sound like them — not a generic 3–10 word prompt. An hour of course video is roughly 12,000 words already in their brand voice; that depth is what prior AI features couldn't reach. That became Kajabi's first monetized AI-generation surface: a new NRR line pointed at the contacts and GMV levers behind earn = stay. Demand showed up early — 10.5k creators joined the waitlist by late 2022, clearing the 10k target before Creator Studio charged for anything.
The problem
User problems
Can’t convert audience into contacts or revenue — expertise doesn’t translate to marketing; exit interviews cite weak marketing help as a top-4 reason.
Not confident marketers — need professional-quality output without a pro’s skill set.
Too slow, too many tools — 8+ hours even with templates, often across Canva, ChatGPT, and other paid apps.
Business problems
Earn = stay — help creators market and earn so the unprofitable segment shrinks.
Contacts as a retention lever — social posts and landing pages built to capture leads (1 contact / 30 days → 65% lower churn).
New NRR line + less tool sprawl — first monetized AI-generation add-on; keep marketing spend inside Kajabi.
Risks we designed against
Unit cost at scale — drove the GA vendor switch (Pictory → Picsart) and capped free + paid beyond.
Add-on pricing resistance — little a-la-carte precedent; free monthly project before the ask; value framed vs. tools creators already paid for.
Hard-to-see ROI — performance analytics stayed on the post-GA roadmap to close generation → contacts → GMV.
Research we ran4 studies · interviews · usability tests
Before we built Creator Studio, three studies framed the opportunity. Once it was in creators' hands, we validated flows through interviews and usability tests, and a Dec 2023 beta survey confirmed impact and pointed at what still needed work before GA.
01 · Marketing Experience Survey · Feb 2022 · n=578
Marketing is the hard part
Only about 2 in 5 creators felt highly confident marketing their business. Content publishing and social media drove awareness across the funnel; most marketed on a tight budget and wanted help that felt authentic — without taking over their day.
Finding Remove the marketing skill barrier; don't teach another marketing course.
02 · AI Content Creation Survey · Nov 2022 · n=356
Help from content they already have
Creators felt less comfortable marketing than creating products; 52% struggled with marketing best practices or effective copy. About half were interested in assisted copywriting inside Kajabi — especially lead magnets, headlines, and landing pages — and preferred offerings built from their existing content over blank-page AI. Separately, 61% wanted AI support designing funnels, automations, and pricing — demand that later justified expanding beyond clips. Price was the top concern; quality and ease of use ranked as must-haves.
Finding Start from assets they already filmed or wrote — then earn the right to generate the full funnel.
03 · Third Party Tool Usage · Aug 2023 · n=534
They're already paying for this stack
Creators used about 9 tools outside Kajabi on average. Video editing (83%) and transcription (62%) led usage; Builders and Scalers who loved Kajabi spent roughly $1,300/mo on extras. Many expected transcription from Kajabi, while deep video editing was often something they'd rather integrate than replace.
Finding Win the repurpose + transcript job inside Kajabi; partner for heavy editing later.
04 · Creator Studio Beta Survey · Dec 2023 · n=68
The thesis held — the stack needed work
57% found Creator Studio easy or very easy; 51% said their business would be somewhat or very negatively affected without it; 43% had fully or partly replaced tools like Descript, Canva, and CastMagic. Top inputs were course video, Zoom, and marketing video; top destinations YouTube, Instagram, and their site. Ongoing beta feedback also surfaced shallow editing, weak AI highlights, slow processing, and asks to upload their own captions.
Finding Ship GA without delaying for roadmap asks — but rebuild the editing stack and vendors underneath first.
01Pre-beta
Start narrow: prove repurposing on video before building a funnel
The opening bet Grounded in research with Research, PMM & PM
The decision: validate the core thesis — creators will repurpose content they already have — with the smallest possible bet before committing to a full product. I framed and ran this research with the research team (including Marika Massoglia), PMM, and PM. It pointed at one format to start: video, where creators already had hours of untapped course and coaching footage.
To reach validation in weeks rather than quarters, I made a deliberate build-vs-buy call: rather than build a video pipeline, we white-labeled Pictory and used the ChatGPT API for auto text-selection to cut clips to each platform's size — Instagram, Facebook, TikTok, LinkedIn, and YouTube. The trade-off was less control over the underlying engine; the payoff was testing the repurposing thesis fast, with scope held deliberately narrow — clips only, no funnel. Four design calls shaped that pre-beta experience:
Reuse existing lessons & recordings
Why: the fastest path to a first clip wasn't new filming — it was the back catalog creators already had. So the source picker let them upload or select existing course lessons or coaching recordings, meeting them where their content already lived.
Auto-highlight the transcript
Why: a blank timeline is where non-editors freeze. Auto-highlighting the transcript gave creators a starting draft instead of a blank canvas — while keeping them in control to add or remove highlights. Only the highlighted passages get cut and assembled into the clip.
Light editing after the cut
Why: a raw cut rarely fits every platform's frame. Rather than ship a full editor, I scoped the minimum that made output publishable without video skills — light edits and per-platform cropping — so creators could finish a clip, not learn a tool.
Syndication
Why: every export-and-re-upload step is a drop-off point. Publishing straight from Creator Studio to Instagram and Facebook kept creators in the flow from source video to posted clip — closing the loop the whole bet depended on.
02Beta
Expand from one clip to the entire funnel
In response to Feedback 01 Widened the bet from social video to full-funnel assets
The decision: stop treating Creator Studio as a video tool. Being able to make a clip didn't solve the real problem — creators struggled to market, and marketing is a funnel, not a format. So Beta let creators turn one source video into 37+ additional media types, meet their footage where it already lived (YouTube), make course content easy to find, and — for the first time — carry a pricing and monetization model. The alternative was to keep deepening the video editor; I rejected it because a better clip still left creators staring at an empty funnel. The trade-off was scope and complexity; the payoff was a tool that finally addressed the business problem, not just the editing one.
Paths considered
Rejected
Deepen the clip tool
Richer video editor
More social formats & crops
Stay in “repurpose for social”
Why not: Feedback 01 asked for a funnel and YouTube — a better clip still left marketing unsolved.
Shipped
Become the marketing product
37+ funnel media types
YouTube import + course search
Pricing / add-on model
Why: matched what creators needed to earn — and what the business needed for NRR.
Same source content either way — the fork was whether we stayed a video utility or became a path to contacts and GMV.
37+ media types from one video
Why: making a video didn't fix creators' core struggle — marketing. Beta expanded output to 37+ additional media types generated from a single source video, so creators could assemble the entire funnel — landing pages, emails, social, and more — instead of just a clip.
Import straight from YouTube
Why: my own user testing after pre-beta made this impossible to ignore — creators' best footage already lived on YouTube. Letting them paste a YouTube link — not just upload a file — removed a download-and-re-upload step and met their content where it was.
Search your course videos
Why: scrolling a long content library kills momentum. Adding search when selecting a source video from their course inside Creator Studio let creators find the right footage in seconds.
Pricing & subscription model
Why: vendor cost scales with usage, and creators already pay for editing tools off-platform. We tested a model with capped access on plans plus an optional add-on — so creators could unlock more usage without upgrading their whole Kajabi plan.
03GA
Rebuild the stack underneath: new vendors, a real editor
In response to Feedback 02 Fixed editing, status, and vendor economics
The decision: before scaling to GA, fix the layer creators and the business were straining against. That meant swapping the vendors powering generation and giving creators a real editor rather than a light one. I moved the video vendor from Pictory to Picsart to bring costs down and widen capability, and moved editing to Adobe, giving creators access to Adobe Express. The alternative — patching the existing integrations — was cheaper short-term but capped how far we could take editing and left the cost problem in place. The trade-off was a heavier migration; the payoff was an editing experience and cost structure GA could actually stand on.
Switch to Picsart
Why: Pictory's economics were hard to scale — and its AI highlights couldn't carry the core job of turning long content into short clips. We moved the video vendor to Picsart and leaned on GPT-4 for highlights instead.
Adobe Express editing
Why: creators asked for far more editing control than pre-beta's light editor offered. Moving editing to Adobe gave creators Adobe Express inside Creator Studio — professional-grade editing without leaving the flow.
Generation status messaging
Why: with 37+ media types generating in parallel, silence read as failure. Toast notifications surface progress on whatever page the creator is on — with a CTA to jump back to the project when files are ready. The same update is also sent by email, which creators can turn on or off in settings.
What's next — roadmap at GA6 roadmap bets
At GA I left a clear next arc — one asset in, a full funnel out — into the phase after generation: distribution and measurement. These bets were designed to answer the question GA raised: once creators can build the funnel, how do we make it work for them automatically?
Auto-syndication & scheduling
Publish and schedule generated posts to a creator's social channels from Creator Studio, using the Adobe SDK for scheduling.
→ Strategic value: Activation + NRR
Performance analytics
Show which repurposed assets actually convert — closing the loop from generation to GMV.
→ Strategic value: Retention + GMV
Multi-language repurposing
Turn one asset into funnels in multiple languages without new filming.
→ Strategic value: NRR
Remove filler words
Auto-strip ums, uhs, and dead air so clips feel polished without manual cleanup.
→ Strategic value: NRR
Bulk create projects
Select multiple source videos — e.g. 10 lessons — and create 10 projects at once.
→ Strategic value: NRR
Video to audio — podcast episodes
Turn course or coaching video into podcast-ready audio from the same source.
→ Strategic value: NRR
Outcome
Creator Studio went GA on Mar 13, 2024 — Kajabi's first monetized AI-generation product, and the first in-product path for creators to turn one piece of content into marketing built to win contacts.
131 paid add-on users
by Mar 24 — 109% of March goal (120). 79 Basic ($9) · 52 Premium ($29)
17.2k marketing assets
from 3.3k users · 3.8k projects in the first ~2 weeks
43%
of beta creators fully or partly replaced tools like Descript, Canva, and CastMagic
Beta survey · Dec 2023 (n=68)
57% found Creator Studio easy or very easy
51% said their business would be somewhat or very negatively affected without it
43% fully or partly replaced 1+ external tools
User problems → impact
Creators can't convert audience into contacts or revenue.
Early use centered on funnel assets built to capture leads — clips, captions, posts, and landing-page-adjacent copy — at the volume above.
Creators aren't confident marketers.
Survey above — non-marketers could ship without a pro skill set.
Marketing takes too long and too many tools.
Tool replacement and asset volume above compressed the 8+ hour, multi-tool funnel build into generation plus light editing.
Business problems → impact
Add a new NRR line and reduce tool sprawl.
First AI-generation add-on — paid uptake and tool replacement above, aimed at a $650k annualized NRR target.
Reduce churn by helping creators earn.
Shipped on the earn = stay path so creators could market and earn without leaving Kajabi.
Reduce churn by helping creators gain contacts.
Social syndication and landing-page assets shipped for lead capture. Contact-gain and profitable-creator rates were the longer measures we still wanted tighter instrumentation for after GA.
Company context
Company retention and NPS metrics also moved in this era alongside other growth and retention work — not claimed as Creator Studio-only wins.
Reflection
What worked: Research conviction on rich inputs; solo user testing after pre-beta (YouTube import, faster funnel pivot); pricing and vendor economics treated as design problems so GA could scale.
What I'd do differently:
Validate the funnel bet earlier — would have saved a release cycle.
Push DS contributions onto a parallel track sooner.
Stand up leading-indicator dashboards in week 1, not at launch.
Beyond design: co-authored strategy and pricing/NRR with PM/PMM; co-ran formal research; shaped reusable AI patterns — intake, generation, editable outputs, source transparency.
On AI products: anything the model wrote or highlighted had to stay editable — AI as a tool creators tweak, never the ultimate content producer. Source transparency and unit cost are part of that trust equation.
