The creator economy’s real bottleneck isn’t content — it’s the pitch
Every social media operator I know has, at some point in the last eighteen months, quietly become a job-seeker. Not because the work disappeared — if anything, the demand for short-form video editors, lifecycle marketers, and “fractional heads of content” has never been higher — but because the platforms keep reshuffling the deck, and the roles keep getting redefined underneath us. The people who survive that churn aren’t the ones with the best Canva templates. They’re the ones who can walk into a room, explain what they do in ninety seconds, and back it up with numbers. That’s the skill Toby is trying to train, and even though it’s pitched at job candidates, I think it’s the most interesting thing a creator can steal from this week’s Product Hunt slate.
Let me explain why an AI mock-interview tool should be on your radar when your job is ostensibly to make TikToks.
What Toby actually is, and what problem it’s solving
Toby is a browser-based interview simulator built by Rob Kealey, a product manager who describes it as “the first idea I’ve taken from a sketch to something people can actually use and pay for, on my own.” The workflow is three steps, and the maker lays them out plainly in his launch post:
- Paste the job description. Toby researches the company and maps every round, from recruiter screen to case study.
- Take the call in a Meet, Teams, or Zoom-style room. “Real mode” runs the interview straight through; “Coach mode” gives you a note after each answer and a retry.
- Get the debrief — a score against the role’s rubric, the quote behind each score, pace and filler-word analysis, and a stronger version of each answer built only from what you actually said.
Pricing is the part that made me sit up. There’s no subscription. You pay 50p per minute of call time; briefs and debriefs are included. That’s a fundamentally different business model from the SaaS seat-licensing that dominates the AI tooling space right now, and it’s worth unpacking.
Why the per-minute model matters more than the feature list
Most AI products targeting creators price on seats or credits. Jasper charges per seat. Descript meters transcription hours. Opus Clip meters upload minutes. The metered-minute model isn’t new — what’s new is applying it to a conversational product where the value is entirely in the interaction, not the artifact. If Toby’s only output were a PDF debrief, per-minute pricing would feel punitive. But because the product’s whole thesis is “practice the conversation,” paying by the minute aligns the incentive: the maker wants you to have shorter, sharper sessions, not to keep you subscribed indefinitely.
My take: this is the pricing model more creator tools should experiment with. If you’re building anything that involves a live AI conversation — coaching, brainstorming, script table-reads — per-minute beats per-seat almost every time, because it removes the “I’m paying for this even when I’m not using it” resentment that kills retention.
How it stacks up against the incumbents you’re already paying for
The obvious comparison is Interview Warmup by Google, which is free and has been sitting in the Google Career Certificates ecosystem for a couple of years. It transcribes your answers and flags talking points, but it doesn’t push back. It doesn’t research the company. It doesn’t score you against a rubric. It’s a mirror, not a sparring partner.
The second comparison is the generic LLM workflow: open ChatGPT or Claude, paste the job description, ask for likely questions, and type your answers. Kealey calls this out directly in his launch post — “a list doesn’t push back, and it can’t tell you that your best example had no numbers in it.” He’s right, and this is the exact failure mode I’ve watched play out with creators using LLMs to prep for brand pitches. You get a tidy list of “tell me about a time when…” prompts, you write answers that feel good, and then you get on the actual call and discover that the brand manager wants to talk about CAC payback, not your creative philosophy.
The third comparison — and the one that should make social media operators pay attention — is Yoodli, which has been the dominant player in AI speech coaching for a couple of years. Yoodli is broader: it coaches you on sales calls, presentations, and interviews, and it has a free tier that’s genuinely usable. Toby is narrower and, based on the maker’s description, more opinionated about the interview format specifically. Whether narrow-and-deep beats broad-and-shallow is an open question, and I’d want to run both side by side before picking a winner.
Why TikTok creators should care more than LinkedIn ones
Here’s the counterintuitive bit. The people who most need Toby are not the people LinkedIn thinks need Toby.
LinkedIn’s culture already trains you to write the STAR-format answer. The platform’s whole grammar — “I’m humbled to announce,” “excited to share” — is interview prep in disguise. If you’ve spent three years posting on LinkedIn, you’ve been rehearsing for the recruiter screen whether you meant to or not.
TikTok creators are the opposite. Their skill is compression: hook in 1.5 seconds, payoff in 12, loop back to the top. That’s a performance skill, not a narrative skill. When a TikTok creator walks into a brand partnership call or a full-time role at a media company, they often can’t translate “I made a video that got 2M views” into “here’s the strategic insight behind why it worked, and here’s the metric that proves it.” Toby’s debrief — which quotes your own words back at you and scores them against a rubric — is essentially a translation layer between creator-speak and hiring-manager-speak.
I’d bet the highest-leverage use of this tool for a creator isn’t job interviews at all. It’s brand pitch rehearsals. Paste the RFP. Let Toby generate the likely questions the brand’s marketing lead will ask. Run Coach mode. The rubric is different, but the muscle is the same.
What creators and social teams can borrow from this
Strip away the interview framing and Toby is a template for something much bigger: structured AI rehearsal with a debrief loop. That pattern is applicable to almost every high-stakes conversation a social media operator has.
The three-part loop: brief → simulate → debrief
Most AI content tools skip the middle step. They generate a brief (a content calendar, a script outline, a caption pack) and then they generate a debrief (analytics, performance scores). What’s missing is the simulation — the part where you actually have to perform the thing and get feedback in real time.
Think about how you’d apply this to a creator workflow:
- Podcast prep. Brief: paste your guest’s last three episodes. Simulate: have the AI play the guest and answer your questions in their voice. Debrief: which of your questions produced the most interesting answer, and which fell flat?
- Live shopping rehearsals. Brief: paste the product spec. Simulate: run the pitch, with the AI interrupting mid-sentence to ask about shipping times. Debrief: pace, filler words, and whether you actually mentioned the price.
- Client QBR prep. Brief: paste last quarter’s analytics. Simulate: the client asks why reach dropped 30% in March. Debrief: did you answer the question or deflect?
The interruption feature is the detail that makes this credible. Kareem Ben, a commenter on the Toby launch thread, specifically praised the fact that “Toby runs a realistic mock interview with an AI hiring manager you can actually interrupt mid question, like a real conversation,” and Kealey responded that he “put quite a bit of focus on interruptions/random question handling.” That’s the hard engineering problem. Turn-taking, interruption recovery, and topic drift are what separate a toy from a tool — and they’re exactly the capabilities that would make an AI rehearsal useful for podcasters and live sellers.
The debrief is the product
Here’s the thing most AI tooling gets wrong: it treats generation as the value and feedback as an afterthought. Toby inverts that. The interview itself is table stakes — any LLM can play a hiring manager. The debrief, with its rubric scoring and its “stronger version of each answer built only from what you actually said,” is where the actual learning happens.
For social media operators, this maps onto a workflow you probably already do badly: post-mortems. Most teams run a post-mortem that’s either too shallow (“the Reel did well, let’s do more Reels”) or too deep (“here’s a 40-tab Looker dashboard”). The Toby pattern — quote the specific moment, score it against a predefined rubric, suggest a revised version that stays faithful to what actually happened — is a much better template. You could run it manually this week with nothing more than a transcript and a spreadsheet.
Where my judgment says it falls short
I want to be careful here, because Kealey explicitly asked for honesty in his launch post: “the most useful thing you can give me is honesty: run one round for a role you actually want (or are hiring for), then tell me where the debrief was wrong.”
So here’s where I’d push back.
The latency and pronunciation problem is not cosmetic
When Ilyas asked how Toby handles latency, Kealey’s answer was refreshingly candid: “That part is still quite tricky to be honest, and one thing that makes it feel artificial more than the latency is the inability to pronounce certain terms/company names properly; for example I heard it say ‘How much experience do you have with A slash B testing’ which was a little weird haha.”
That “A slash B testing” moment is a bigger deal than the maker lets on. The whole premise of the product is realistic rehearsal. If the AI stumbles on the vocabulary of the role you’re interviewing for — A/B testing, CAC, ROAS, CTR, UTM parameters, DTC, MQL — then it breaks the illusion precisely when it matters most. For a creator prepping for a performance marketing role, an AI that can’t say “ROAS” is a non-starter. My take: this is fixable (a pronunciation dictionary per role vertical would do it), but it’s the kind of fix that requires domain knowledge the maker may not have yet.
Per-minute pricing has a ceiling
50p per minute sounds cheap until you do the math on a realistic prep session. A 45-minute mock interview plus a 15-minute debrief is an hour of call time. That’s £30 for a single session. If you’re prepping for a competitive role and you want to run three rounds, you’re at £90 — more than a month of Yoodli’s paid tier, and more than most creators spend on their entire tool stack outside of Adobe.
Kealey frames the no-subscription model as a feature (“Pricing is simple. No subscription.”), and for a first-time builder, it probably is — it removes the pressure to build retention mechanics before you’ve validated the core loop. But I’d bet the model breaks down for power users. The people who’d benefit most from daily practice are the ones who can least afford per-minute billing.
The rubric is a black box
The debrief scores you “against the role’s rubric,” but the source doesn’t say where that rubric comes from. Is it generated from the job description? Is it a generic competency framework? Is it tuned per company? Not disclosed. That matters enormously, because a rubric is only as good as its provenance. If Toby is scoring you against a hallucinated rubric, the debrief is worse than useless — it’s actively misleading, because it gives you false confidence in the wrong direction.
I’d want to see the rubric before I trusted the score. That’s true of any AI feedback tool, and it’s the question I’d ask any founder building in this space.
Who this is NOT for
If you’re a senior operator with a decade of hiring experience, Toby is probably redundant. You already know how to structure a STAR answer; you already know to put numbers in your examples. The tool’s value is highest for people who are early in their career, career-switching, or coming from a non-traditional background (which, let’s be honest, describes a huge chunk of the creator economy).
If you’re hiring rather than job-seeking, the use case is less obvious. Kealey suggests running a round “for a role you actually want (or are hiring for),” but the product is clearly optimized for the candidate side. A hiring-manager mode — where you upload your own interview questions and calibrate against a candidate’s answers — would be a different product.
What I’d watch / test next
Three concrete things I’d do this week if I were a social media operator reading this:
1. Steal the brief → simulate → debrief loop for your next client pitch. You don’t need Toby to do this. Write the brief by hand, run the simulation with ChatGPT or Claude in voice mode, and do the debrief yourself with a transcript and a rubric you define in advance. The rubric is the hard part — define it before you run the simulation, or you’ll rationalize whatever happened.
2. If you’re job-hunting, run one round on Toby. The maker is explicitly asking for feedback, and at 50p a minute, a single 20-minute session is £10. Run it for a role you actually want, then write down every place the debrief was wrong. That’s the most useful thing you can give a first-time builder, and it’s also the fastest way to find out whether the tool is worth a second session.
3. Watch the pricing experiment. Per-minute billing for conversational AI is underexplored, and Toby is a useful data point. If it works, expect to see it copied by coaching tools, language-learning apps, and — most interestingly for this audience — AI podcast co-hosts. If it doesn’t, the lesson will be that creators prefer flat-rate subscriptions even when they’re paying for unused capacity. Either way, it’s a signal worth tracking.
The bigger takeaway: the creator economy’s next wave of tooling isn’t about making content faster. It’s about making the conversations around content — pitches, negotiations, performance reviews, partnership calls — sharper. Toby is a narrow product in a niche most creators don’t think they’re in. But the pattern underneath it is the one I’d bet on for the next eighteen months.
