AI Applyd
The auto-apply that proves your application arrived
I was about to panic when PostHog said my SaaS was burning $6.46/week on embeddings. Turns out it was tracking paid pricing for usage that was fully under Cloudflare's free tier. Zero actual dollars. That kicked off a full audit across every platform I pay for.
## TL;DR
- Real monthly infra cost for aiapplyd: about $35 to $55 after Browserbase cancels next month. Not $104 like my dashboard suggested.
- Hidden stuff my tools weren't telling me: a phantom $6.46/7d tracking artifact, a 67% error rate in my enrichment pipeline, and a BYOK Gemini key routing $5/mo outside my "official" AI budget.
- If you're at 5 paying users on $100+ infra, do the audit. You'll find money and bugs.
## What is aiapplyd
AI-powered job application platform. ATS scoring, auto-apply, interview prep, resume builder, tracking. Solo founder, TypeScript, Cloudflare Workers, D1, Hono, TanStack Start. Five paying users right now. That number stings to type but it's the honest state of things.
## Why I audited
Two triggers, both uncomfortable.
First, my ChatGPT referral traffic collapsed this week. I'd been riding ChatGPT as my #2 channel for a month. Woke up to a red line on my chart. Made me want to know what I could cut if things got leaner.
Second, the math on 5 users at roughly $104/mo infra is $20.84 per user just to keep the lights on. That's not CAC, that's raw infra burn per existing customer. It felt wrong. I wanted to know if that number was real or inflated by dashboard lies.
## The method
Nothing fancy. Playwright MCP driving through every platform dashboard I could think of. Gmail receipt search for anything that renewed quietly. PostHog HogQL queries to cross-check what my own app was logging. Ten platforms total.
The goal was one row per platform: actual cash going out, not rate card estimates.
## The numbers
Claude Max personal sub $200. Resend Pro $20. Browserbase Developer $20 to $0 after May 12 downgrade. Cloudflare free tier $0 with $5 lifetime spend. OpenRouter prepaid credits about $3 to $5 on $35 lifetime. Anthropic API Haiku auto-topup $18 one-off for Stagehand browser. Firecrawl $19 one-off. PostHog free tier under 1M events. Stripe $0 outgoing platform fee. GCP Gemini BYOK about $5 routed via OpenRouter.
Project infra total about $50/mo now, $35 to $55/mo after Browserbase downgrade.
Claude Max is the single biggest line item on my personal card, but it's my coding IDE subscription, not infra. Stripped out, the project itself runs lean.
## The three hidden findings
### 1. Phantom $6.46/7d embedding cost
PostHog's LLM costs dashboard showed Workers AI embeddings as a paid line item. Added up to $6.46 over a rolling 7 days. Looked like it was going to be my fastest-growing cost.
I pulled the actual Workers AI billing page. $0.00. The embeddings model I'm using is inside the Cloudflare free tier bucket for my tier. PostHog was applying public rate card pricing to events my provider never billed me for. Tracking artifact. The dashboard lied and I almost made a product decision based on it.
Fix: adjusting my PostHog event properties so it doesn't apply dollar per token on free-tier paths. Filing with PostHog because I doubt I'm the only one.
### 2. BYOK Gemini leaking outside the official AI budget
I routed one subset of AI calls through a user-provided Google AI Studio key via OpenRouter BYOK. It worked. But it meant about $5/mo was sitting on a Google Cloud bill I don't open often, completely invisible to my OpenRouter total spend number.
Not a big number, but the lesson is: BYOK routes will silently bifurcate your AI accounting. Tag them, total them, or kill them.
### 3. 67% error rate in enrichment
This one I'm embarrassed by. I run an enrichment pipeline that scrapes and normalizes data for one internal job. HogQL showed 67% of invocations were failing and retrying. The user-facing flow was fine because of fallback logic. But I was paying for browser time, AI tokens, and retries on a loop that mostly died.
Root cause: a selector change on an upstream site three weeks ago plus an overly strict schema parse on my side. The fallback was so graceful I never noticed the underlying failure rate. Cost me an estimated 40 dollars of Browserbase and AI time over the last month, bundled into the totals above. Also a chunk of latency my users silently ate.
## What changed as a result
Kicked off 10 parallel agents to stream events into a single cost and reliability Telegram channel. Daily roll-ups so I don't have to log into 10 dashboards.
Set hard monthly caps on every prepaid provider. If I blow through, the service degrades instead of silently 10x-ing.
Weekly review of HogQL error-rate queries on anything that calls a paid API. Fallbacks are great, but fallbacks without alarms are just hidden bills.
Browserbase from $20/mo Developer down to Free tier. I was under the session limit anyway, I just hadn't looked.
## Lessons
1. Your observability dashboard is not your bill. They're two different systems with different assumptions. Reconcile them at least quarterly.
2. Free-tier usage gets priced on rate cards by default in most analytics tools. Either override the pricing field or ignore the total.
3. Graceful fallbacks are the enemy of error visibility. If a pipeline can silently retry, it will, and you will pay for every attempt.
4. BYOK is convenient but splits your cost truth across accounts. Consolidate or tag rigorously.
5. Audit the stack is not a one-time task. I'm adding it to my monthly cadence because a $200 personal sub can hide a lot.
## Honest ask
IH, anyone else running on roughly $50/mo infra and finding weird tracking artifacts in analytics or billing dashboards? I'd love to compare notes on where the reported number diverged from the bill. Specifically interested in PostHog LLM cost tracking, Cloudflare Workers AI free-tier accounting, and Browserbase session metering.
Also if you're a solo founder under 10 paying users and want to swap infra-cost spreadsheets, drop a comment. I'll share the Playwright MCP script I used to scrape all 10 dashboards.
Full build log and the phantom-cost deep dive going up on https://aiapplyd.com/blog this week.
Last year I was job hunting. Applied to 127 positions manually. Tailored resumes, wrote cover letters, filled out the same forms over and over. Got 3 callbacks. Three.
Then I found out my resume was scoring 34% on ATS systems. Nobody told me — the system was filtering me out before a human ever saw my resume.
So I spent a year building the tool I wish existed.
AI Applyd runs a 6-step pipeline while you sleep: finds jobs across LinkedIn, Indeed, and Glassdoor, scores your resume against ATS, tailors it per role, writes cover letters, and submits up to 170 applications a week on the top plan.
Two modes: Copilot (review each application before it goes out) or YOLO (full automation — wake up to applied jobs).
Built the entire thing solo. TanStack Start + Hono on Cloudflare Workers. Bootstrapped. No investors. Just frustration turned into code.
Free to start, no credit card. Pro was $12/month when I posted this. It is $19.49/week or $38.99/month now.
Would love feedback from fellow indie hackers — especially on pricing and the auto-apply angle. Is "YOLO mode" too aggressive, or exactly what job seekers want?
-
1
this hits hard, especially the ATS part — most people don’t even realise they’re getting filtered before a human sees their resume
the copilot vs YOLO mode is actually a smart approach, gives control + automation both
since you're already getting traction and thinking about pricing, you could also try putting this into a competition
good way to validate positioning + get more visibilityalso, prize pool just opened at $0, so your odds are the best right now
-
1
the invisible-filter part is what got me too. i spent months assuming humans were reading and passing on me, when most of the time no human ever saw it. copilot vs full-auto exists because trust is personal, some people want to watch the first ten go out, some never want to think about it again. thanks for reading.
-
-
1
Getting filtered out before a human sees anything is the real problem, so fixing that alone is valuable.
YOLO mode will appeal to people who just want volume without the grind, but it comes down to how accurate the tailoring is. If it’s off, people won’t trust it.
I’m seeing a similar shift with AI Home Health Software integrating AI into documentation and billing. It’s great when it removes repetitive work and catches things, but small mistakes create bigger problems fast. Same here, accuracy is what decides whether this becomes something people rely on or just try once.
-
1
accuracy is the whole ballgame, agreed. the failure i actually watch for isn't a bad application, it's a confident bad one, where the tool tailors something wrong and reports it as done. volume without accuracy just automates the rejection faster. that's the reason i'd rather show someone exactly what got sent than hand them a reassuring number.
-
-
1
This is exactly why generic job applications are broken.
You didn’t fail — the system did.
Building your own AI to bypass it is actually the logical move.
-
1
appreciate that, though i'd only half-take the 'you didn't fail' part. the system is genuinely broken, but i also fired off a lot of lazy applications early and blamed the funnel for all of it. building the tool forced me to separate which rejections were the system and which were just me not trying. both were true, and that was the uncomfortable part.
-
-
1
The gap wasn't the number of applications it was the 34% ATS score you didn't know about. Most job seekers optimize for volume because they can't see the filter that's silently rejecting them. You built the visibility first, then the volume. That's the right order.
-
1
that ordering was almost an accident, honestly. i built the score first because i wanted to know why i kept getting rejected, and only later realized the volume was worthless without it. seeing the filter is the part nobody sells, because it's less exciting than 'apply to 500 jobs while you sleep.' thanks, that framing is cleaner than mine.
-
1
Yes! Exactly.
-
-
1
This hits hard. That “127 applications → 3 callbacks” experience is way more common than people admit.
What I like most is you didn’t just complain about ATS — you actually built something that solves it end-to-end. The 6-step pipeline + sleep automation is a strong angle.
On “YOLO mode” — I think people want it, but also fear it. Maybe the sweet spot is control + transparency (like showing exactly what’s being sent).
$12/month feels very reasonable for the value, especially if it actually improves response rates.
Curious to see how users behave long-term — do they trust full automation or stick with copilot?
Really solid build 👏
-
1
late correction on the one number you quoted back to me. the $12/month was real when i wrote this post, but it is not the price any more. it is $19.49/week or $38.99/month now, and the top tier is $33.99/week. i have edited the post rather than leave your comment anchored to a number that stopped being true. the free tier still exists and still excludes auto-apply, so the honest version is: free to try the scoring, paid to have it actually apply.
-
1
early answer: almost everyone says they want full automation, then reaches for copilot the first time it touches something that matters, like a cover letter or a salary field. trust isn't a setting, it gets earned per action. the people who go full-auto are usually the ones who watched it handle the boring 80 percent first. 'control plus transparency' is exactly it, showing what's actually being sent is what converts a skeptic. thanks for the thoughtful read.
-
-
0
This comment was deleted a month ago
-
1
the churn point is the real one, and you're right that the use case is spiky rather than steady. someone lands a job and the reason to pay evaporates overnight. i've gone back and forth on sprint pricing for exactly that reason. the wrinkle is that a search rarely ends cleanly, people land, then get laid off, then start again, so the 'sprint' is really a recurring event with long quiet gaps between. i don't have this solved and i won't pretend the pricing is finished. the honest version is: price it to the search, not the calendar.
-
2 Comments
Hey, your SaaS audit post was incredibly detailed.
I’m also building a SaaS (AI image/video generation platform) and currently going through similar scaling + infra optimization challenges.
The part about hidden costs and dashboard mismatches really hit home.
Would love to stay connected — your insights are exactly what early-stage builders need.
thanks, glad the audit was useful. hidden costs and dashboard mismatches are weirdly universal, every stack i've looked at has at least one line item nobody can fully explain. good luck with the image and video build, that's a brutal infra-cost category to keep honest.