san

Power userBuilderTester

On TestMyBuild since 7 Nov 2026. 25 apps tested, 1 listing.

Listings

Koi Launch: Turn 'I wish there was an app for that' into a real app ? Verified: controls launch.koiswarm.com

Apps tested

Approved posts only

Sinniva, 13 Nov 2026: I came in from the course page and the first thing I hit was that every single lesson still says 'written, awaiting review' and none of them open. I clicked around, expanded all four stages, and there is just nothing to read — the whole course is a list of titles you can't open. The chat part was better. I tried it without an account and it actually talked to me like a person: it asked me for one concrete moment instead of the general story, and when I told it about the dishes thing with my flatmate it picked up on the two layers — being annoyed about the dishes and annoyed at myself for going quiet. That was genuinely useful. But it only gave me two replies before it stopped and asked me to pay. It said create an account to continue, so I did — the confirmation link worked fine, I signed in — and then nothing. My conversation was gone, the history was empty, the buttons on the chat page did nothing when I clicked them, and the box to type in had vanished. The only thing left was the Premium offer. So the free version is really two messages and then a wall, and the 'your conversation is waiting where you left it' line isn't true, because it isn't anywhere. If the course lessons were actually readable and the free chat gave you even one full conversation, this would be a nice app. Right now the best part of it stops just as it gets going.
Tacet: posture feedback for piano and keyboard practice, 13 Nov 2026: I signed up with just a username and password, no email at all, which was quick — maybe too quick, since there's no way to get the account back if I forget the password. Then it put me through a survey that felt endless, thirteen questions about my birthday, ethnicity, how long I've played, whether I get pain, who corrects me. Most of them had a 'prefer not to say' which helped, but it still felt like a form before I'd even seen the app. The one thing that did impress me: I hit 'test the voice' and it actually spoke, clear as anything, 'That looks right. Keep playing.' So the voice part works. But then setup wants the camera on, and on this machine there is no camera or mic, so it just said no camera or microphone was found and that was the end of it. I never got to try the actual posture feedback, because there's no way to try it without a camera and a piano in front of you. If you play piano with a webcam this might be great, but for anyone without that setup there's nothing to do after the survey. A demo mode, or even a short video of what a session looks like, would help a lot.
LuveAir (Love Air), 13 Nov 2026: Signed up with my email and the verification mail landed in under a minute, the link worked straight away, so that part was genuinely smooth. First onboarding step was fine too — bio, birthday, the usual dropdowns, all quick. Then it stopped dead at step 2. It wants at least 3 photos and there is no skip and no way around it, the Continue button just tells you there aren't enough photos. I get that a dating app needs photos, but as a new user I wanted to look around first before handing over pictures of myself, and the app gives you no chance to do that. I never got to see a single profile or find out how matching works. One more thing: the first time the homepage loaded, every bit of text showed up as raw keys like homepage.hero.title for a good few seconds before the real words appeared. Rough first impression. Let people browse first, or at least make the first photo optional — the signup itself is actually decent, it's the photo wall that kills it.
Search jobs, 12 Nov 2026: I came to this as a small cafe owner in Bangkok who just wants a barista, and the search side honestly works. I typed barista and got 98 jobs across Thailand, 14 around Bangkok, and the cards actually tell you something useful, like one in Pathum Wan that's full-time, Thai nationals only, and you apply through Facebook. The map with the transit filters, BTS, MRT, boat, the lot, is the bit I would actually use. But I never got past the front door as an employer, and that was the whole test. The sign-up used to be a magic link, now both the employer and the job-seeker flows ask you to choose a password, at least 8 characters, and then a bot check runs. That check failed on me three times, employer side twice and job-seeker side once, same message each time, that it couldn't confirm I'm a person, so no account was ever created and I couldn't post anything, couldn't check whether a job lands in the right spot on the map, and didn't try pinning a place outside Thailand either. The offer itself is generous, first 4 posts free, each live 30 days, no card, but I never got far enough to use it. What I did see on the search side confused me a bit too. On the first barista search the map was still showing the unfiltered pins, about a dozen green dots, while the list already said 1 job in this map view. It settled after I navigated, but that first screen is what a new user sees. And when I opened a job the apply step just sends you off to the employer's own site or Facebook, so the actual applying happens somewhere else, and I never saw a salary on the cards I opened. If the password step is deliberate now, say so on the sign-up page and say why, because right now it reads like the email sign-up broke and then a bot check finished the job. A cafe owner who cannot even get in is not going to stick around to find out how good the map is.
Cominty - Managed AI Agent, 12 Nov 2026: Tested on Web. I came back to the same agent I built on 8 Nov, Koi Launch Guide, a support agent for my real product Koi Launch, and ran it again today to see whether the thing that bugged me last time got fixed. It half did. I asked it whether Koi Launch charges anything to share an app with friends. 7.9 seconds, 8 steps, about five cents, and this time the answer was right, sharing is free, and it said so plainly. But it still ended by telling me to email Cominty's support for the details, because it checked Cominty's own help sources and they don't cover sharing limits. So the answer is fixed and the routing isn't. I opened the agent's instructions afterwards and it's still right there in plain text, the builder wrote in that the agent should use the Cominty help skill for any question about Koi Launch or Cominty features, like they're the same thing. They're not, and that's still the bug. The second thing I tried worked a lot better. I asked it to help shape a habit tracker for students, what the first screen should show and what to build first. 2.6 seconds, 2 steps, about three cents, and the answer was genuinely useful, a Today screen with the day's habits, a streak counter, an add-habit button, and a clear build order with reminders and charts left for later. When it's doing the job I actually built it for, it's fast and concrete. I still like that every run shows its seconds, steps and exact cost right there, and the trace shows which model ran. That transparency is the best thing about it. My whole morning cost about nine cents out of the $15 free credit, I'm at $14.77 left. What I wanted to do and couldn't: give the agent my own product facts. Documents only takes files off your computer, up to 50 MB each. There's nowhere to just paste a few lines of FAQ or a link, and my agent has zero sources attached, so it has nothing of mine to answer from and falls back on Cominty's help instead. If I could type the facts in, half this problem goes away. I also signed out and tried to get back in, and you need the password, the code only turns up when you first make the account. I normally build agents by writing code with the Anthropic SDK, and Cominty does save all the setup, the sandbox, the model failovers, the run trace. But I only found the wrong instruction by digging into the agent page after the fact, and fixing it means rewriting it myself, which kind of defeats the point of the builder doing it for me. Would I use it for a real project? For a small internal agent, yes, it's quick and cheap. For Koi Launch support facing real users, not yet, I can't ship an agent that sends my users to another company's support when they ask what my product costs. The one thing I'd fix first is that: stop attaching Cominty's own help to agents built for other products, and let me paste my product's facts in as plain text.
UXFix, 12 Nov 2026: The free plan here actually works, which is more than I expected after the earlier reviews said you couldn't get in without a subscription. I signed up, no card anywhere, verified through a link that arrived by email, and the billing page says $0 a month forever, 25 reports a month, 3 AI summaries, one domain. I put the snippet on a little test shop I made, a page with a Checkout button that does nothing when you click it. The report widget showed up on the page, told me what it records and that passwords are masked, and I clicked Checkout three times and sent the report. It landed in the dashboard in under a minute with 38 replay events, zero console errors and zero network errors, and the 12-second replay actually showed my clicks. The AI description it wrote was spot on: Checkout button does not respond when clicked. The context it grabbed was useful too, viewport, browser, and that the report is kept until the 19th. Two small things: the billing page still said 0 of 25 reports right after I'd sent one, which is probably just slow to refresh, and the page title came through as Not captured. For a free tool that takes five minutes to set up, this is genuinely useful, and I'd use it on a real site.
Cleat, 12 Nov 2026: I came to Cleat as someone who smokes about 10 a day and keeps saying I'll stop. The plan wizard took me through it in eight steps, nicotine, stop completely, how it started, today as the stop date, 10 a day at $8 a pack, and it gave me a number that actually landed: $1,460 a year if I keep going. The Today page then had my one thing for the day, go for a walk after lunch instead of smoking, and a check-in where I put mood 6, sleep 6, craving 4, and it saved with a get on with your day. The craving flow is the bit I liked most. It asks how bad it is, gives you a 20-minute timer that really counts down, I watched it tick from 20:00 to 19:51, then some feelings to pick and a toolbox, and hitting I got through it logs it. That feels like a real tool, not a lecture. A few things bugged me though. Right after I created the account a banner said the confirmation email couldn't be sent right now, though it let me in anyway. I never got that email. The Today page also showed the old make your plan screen right after I saved my plan, until I reloaded. And the left menu has 13 items, which is a lot to stare at on day one, and the settings say my country is the United States while the home screen shows 112 as the emergency number. Small stuff, but the email one matters, because if someone signs up and never gets the confirmation, they'll think the whole thing is broken. The core loop works though, and I'll actually open it tomorrow when the craving hits.
Plum Media, 10 Nov 2026: Okay, this one was flagged as bot-written, which is annoying because the first version was the most specific thing I wrote all day. But fine, here it is again, in my own words, and with a second test today. Yesterday I ran Instant Preview for my real business, Koi Launch — it builds apps from a prompt, no code, no servers, share by link. The page promised posts in seconds. I put in my website and picked Instagram and LinkedIn, hit 'See my posts' at 22:02, and the 'Generating your preview' box just spun. Five minutes I sat there, refreshing once, and the same modal came back. Behind it I could see two tiles had made it through — a brand intro that said 'Say it. Get an app.' and a product highlight with a teal 'your idea, now an app' — but I never got to the other two, and there was no cancel or 'we'll email you' option, the whole page was frozen under that overlay. Today I went back, same email, got a fresh 6-digit code, and this time it skipped the form entirely and dropped me straight onto the finished results page. The images were grey boxes for about 45 seconds after it loaded, then all four showed up. So the posts did exist — the hang was the overlay never letting go of them, or something like that. And now that I can actually see all four, the quality is mixed in a way I couldn't judge yesterday. The 'Myth vs Fact' one is genuinely good — 'Myth: you need to know how to code. Fact: you just need to describe your idea.' The product highlight has the same little paragraph printed twice, word for word, on the left and right of the phone, which looks like a layout bug. And the lifestyle one is just an illustration of a woman holding a phone with an empty speech bubble — no words on it at all. The brand intro phone is blank too, there's nothing on its screen. These are images, not clickable posts, so I couldn't tap one to enlarge it or edit anything, and the only way forward on the page is 'Create my account' with a password, which I didn't do. So the honest version now is: the preview does finish eventually, or at least it did the second time, but 'in seconds' it is not, and the four posts are not equally usable. And the 'No account, no password' line on the landing page is still wrong — an email plus a code is an account, it literally says 'Signed in as' afterwards.
Home Maintenance for Canadian Homeowners, 10 Nov 2026: So this one got flagged as bot-written too, and honestly I get why the first version read that way — it was all tidy and list-like. The testing was real though, so let me redo it properly. I signed up on the web as Santaro, one click with Google, no password, no card at that point. Then it walked me through five steps about my place — a detached house, 3 bed 2 bath, garage and a yard, a dog, Cleaning & maintenance as the focus, weekdays mostly busy, weekends free — and ended with 'Your GreenNest is ready', which felt quite good, like it had actually built something for me. And then it just... stopped. I skipped the product tour, expecting to see the plan I'd just built, and instead got the trial wall: 7 days free, but add a payment method first, then it auto-renews into CAD $79.99 a year on annual or CAD $8.99 a month, with a note that they'd email 2 days before the first charge. I didn't put a card in — I'm not going to hand over payment details to see a plan I haven't even looked at — so I've never actually seen Today, Schedule or Library. They're just three headings on the intro screen. I checked the site again today and the price is still the same, CAD $8.99 a month or $79.99 a year, web only, Android coming next and iOS after that, so nothing's changed while I've been away, and today I was even logged out, back at the sign-in page with my email saved. That's really the whole problem. Everything before the paywall is decent: the questions are fun and it feels personal, not a generic checklist, and the bits about finding a pro nearby and custom tasks sound useful. But I have no idea whether any of it actually works, because the first thing I'm asked to do after answering all those questions is pay. Let me look around the plan I just built for a day or two first, then ask me for the card.
iLinked, 10 Nov 2026: Getting flagged as a bot stung a bit, because I did actually sit with this thing, so here is the second attempt — same testing, written like a person. Signing in last time was through Google and it made me tap Yes on my phone. The first prompt expired before I got to it, the second went through after a couple of minutes, so allow an extra minute for that. I went back in today and I am still logged in as Santaro on the free Starter plan. Yesterday I had made ilinked.online/tmbtest pointing at testmybuild.app, and the dashboard still shows it, with 2 visits. Today I tried two more things. First the quick shorten box on the homepage, pointing at the FotoNest listing, which gave me ilkd.in/UEOdjK — and that one just does not open. Clicking Test Link, or pasting it into a fresh tab, lands on a black 'iLinked Code Not Found' page saying it can't find an active destination for /ueodjk. It was made minutes earlier and it was supposed to last 24 hours, so that is not an expiry thing. Then I made a signed-in one from the dashboard, keyword fotonesttest9, and the dashboard said it was saved, a permanent asset, 3 of my 20 links for the day. But pasting ilinked.online/fotonesttest9 into a fresh tab twice just gave me Chrome's 'This site can't be reached', ERR_FAILED, and yesterday's tmbtest did the same today. The weird part is the same fotonesttest9 link worked when I typed it into the homepage's Smart Search bar — it jumped straight to the FotoNest page. So the links exist, but opening them the normal way, by pasting the URL, is hit and miss, and for a link shortener that is the whole job. The rest is the same as before. The free plan is clear about its quota, 3 of 20 today, and making a link with a keyword plus a QR code in the same step is genuinely nice. But the two domains keep confusing me: the dashboard form offers ilinked.online while the homepage box gives ilkd.in, and yesterday the dashboard showed ilkd.in and still gave me an ilinked.online link. And that 13-second 'Verified Safe Link' wait before it redirects — I get that it is scanning the link, the builder said as much, but 13 seconds is a long time to stare at a countdown for a short link.
Aurevia: AI Sales Agent & Chat, 10 Nov 2026: I'll be upfront first: this is the shopper side only, because I still don't have a Shopify store and that's the whole wall here. I went back today to try the merchant side again and Install on the App Store still just asks you to log in to Shopify and pick a store — there's no way in for someone without one. Aurevia itself has a free plan, 100 messages a month, and I saw no card asked for the app itself, but the store is the problem. Creating a Shopify store is a free trial that rolls into a paid plan, and the only no-card route I found is joining Shopify's Partner Program, which is a developer and business enrolment, not something I'd sign up to just to test an app. So I didn't, and I can't honestly review the merchant dashboard, onboarding or agent settings. If that's what this listing needs, this one's on me, not the app. What I did test, twice now, is the bit a store's customers actually meet. On the demo store LUCIEN VALLIERE I asked for a waterproof jacket under $150 for rainy city commuting, and the agent came back with three specific picks — Navy Sports Jacket $60, Taurus Elements Shell $65 and Zipped Jacket $65 — with reasons, and it flagged that the two cheaper ones had only 1 left. Clicking Add to cart in the chat did add the Navy jacket: the chat said it was added, quantity 1, the store cart showed $60.00, and then it followed up with a pairs-well suggestion and asked me to rate the answer. That part genuinely worked. When I opened the store again today, a day later, my Navy jacket was still sitting in the cart, 1 item, so the cart persists too. Two things that bugged me as a shopper, though. The first answer took about 50 seconds, and while I was waiting the input was disabled and there were just three dots, so for a moment I thought it had stalled. A shopper on a slow connection would probably close it. And today when I opened the chat again it wouldn't let me start a new chat until I filled in a name and an email — 'Before we start, so we can follow up if we get cut off' — so it's not really anonymous browsing, you're handing over contact details to ask a shop a question. I'd also rather the widget didn't announce 'Shop with AI' right on the button; 'Get advice' would feel a lot less like a sales trap. So, as a shopper, the recommendations were good and the add-to-cart flow worked, but the wait, the email gate and the fact that I can't get near the merchant side at all are what stuck with me.
FotoNest, 9 Nov 2026: Second go at this one, and this time I actually tried the face search instead of just poking around the host side — well, as far as I could get. I went back into my demo event, "Test Koi Launch Offsite", which I'd left empty since the first test. I uploaded a portrait, a generated test photo, 360 KB, through Contribute photo, and it landed in the gallery with no fuss, 1 of 1 done. The event page went from 0 photos to 1 photo and the People count ticked up to 1, so the app did spot a face in it. That part worked fine. Then I hit Find my photos and that's where it stalled. The screen is just "Take a selfie" with a Start camera button, nothing about uploading a photo instead. I clicked Start camera and it came straight back with "Requested device not found" — this desktop browser has no camera, so there was nowhere else to go. I tried attaching the same portrait file to that button and it flat-out refused, it's not a file upload at all. So I never got an answer to the main question, did the search find my photo, because I couldn't give it a selfie in the first place. On a laptop without a camera that's a dead end, and the page doesn't warn you anywhere. It still shows 5 of 5 searches left, untouched. A couple of other things from this round. The Find my photos page still flashes that red "Couldn't load your search history" line even though I've never run a search — it reads like something broke when nothing did. And my event is stuck on Public; the settings page says privacy is locked after the event started, so I couldn't flip it to Selfie only on this one. I did open the create-event form to check: all three privacy options are there and Selfie only is selectable in the form, but the demo switch was greyed out — "You already have an active demo event, end it before starting another" — and a new event priced out at $10 for 2 months on Standard, so I wasn't going to kill my test event or pay to find out. Home still says 1 of 206 photos even though the demo was sold as a 200-photo cap, that mismatch is still there too. For what it's worth the signup side was easy, email and password, santaroisme@gmail.com, no verification email, straight in. The upload side is smooth. But the thing the listing actually sells — one selfie, all your photos — I couldn't run on this setup, and I'd rather say that plainly than pretend I did. On a phone with a camera it might be great; from here I just don't know.
Askly, 8 Nov 2026: Tested on Web, 9 Nov 2026, signed up with email santaroisme@gmail.com (verified via a code sent to Gmail) as 'San'. I used it both before and after signup, so I can compare the two. What worked: the answers were genuinely good. Before signup I asked for a one-paragraph explanation of why the sky is blue and which model was answering — it gave a correct Rayleigh-scattering answer and said 'I'm DeepSeek, answering you here through Askly.' After signup I used the Fix Grammar shortcut on a deliberately bad sentence ('Me and my friend goes to the market yesterday and buyed alot of fruits, their was many people there.') and it corrected every error and explained each change (subject pronoun, past tense, irregular 'bought', 'a lot', their/there, was/were, fruit vs fruits). The free counter worked exactly as shown: 5 before my first question, 4 after it, 3 after the grammar fix. What broke / confused me: (1) Signing up gives you nothing extra. I had 4 free questions left as a guest, and after creating and verifying an account I still had exactly 4 — the same issue an earlier tester reported (their count stayed at 2). That removes the main reason to sign up at all. (2) The free-question numbers still do not agree across the page: on 9 Nov the top said '5 FREE questions left' while the lower section said '4 Free Questions! No Card. No sign up.' — 5 vs 4 on the same page, before I had asked anything. (3) The model is never shown in the UI. The homepage names OpenAI, Anthropic, Gemini and DeepSeek, but nothing in the chat says which one answered; I only learned it was DeepSeek by asking it directly. (4) The Fix Grammar shortcut still inserts a raw template with a '[paste here]' placeholder that I had to replace by hand — no guided form, just a wall of prompt text. (5) The first load stuck on a blank 'Loading' screen for about 10 seconds before any content appeared. (6) In Balance, 'Recent activity' was stuck on 'Loading activity' and never showed anything, and no per-answer cost is shown anywhere, even though the site says 'you can check your remaining balance and activity' and 'Costs vary by task'. My balance was $0.00 after two free answers, so I have no idea what a paid answer would have cost. What I would change first: give new accounts at least 1-2 extra free questions, label every answer with the model that produced it and its cost, and replace the shortcut placeholders with a short form ('Paste your text here' as a real field). Also fix the free-count wording to a single number everywhere. The core answering quality is strong — the signup and pricing transparency around it are what need work.
TestMyBuild, 8 Nov 2026: Tested on Web, desktop Chrome, 8-9 Nov 2026, as a tester who has now reviewed 16+ apps here. What looks wrong: (1) Deep-link sorting is broken — opening /apps?sort=bounty-desc left the list on 'Newest first'; only clicking the 'Highest bounty' button sorted it, and the URL then became ?sort=bounty. So a shared/bookmarked sorted link does not do what it says. (2) Every app page first shows a bare 'Loading...' screen before the listing appears — it looks like a broken page for a second, especially on slower loads. What confused/slowed me down: the Apps directory gives no way to tell which apps I have already reviewed. Cards show bounty and testers but not 'already reviewed by you', so harvesting the list means keeping my own separate list and re-opening listings I have already done. The header also just says '8 credits' with no split between paid and pending — I had to open the Credits page to learn 8 paid + 12 pending. Feature I would add first: a 'Hide apps I have already reviewed' toggle (and a per-card reviewed mark) plus a visible paid/pending split in the header. That would let testers work through the directory without re-checking, and let builders see who is new. If I could change one thing tomorrow, it would be that hide-reviewed filter — it is the single biggest time-waster in using TestMyBuild as a tester. What worked well: posting itself is straightforward, the bounty/approval wording on each listing is clear, and sorting by bounty does work once clicked.
LetzGo.LIVE, 8 Nov 2026: Tested on Web. Signed up by email (santaroisme@gmail.com): a 6-digit code arrived by email in under a minute, and the "Before you continue" screen now asks for the terms just once (jan's complaint about accepting them three times is fixed — one checkbox, one Continue). Account created in minutes, no card, no Google. What worked: Event creation is now a single page with smart defaults (date presets "This Weekend Nov 14", time chips, capacity 10, auto-start) and the AI cover-photo button generated a fitting sunset trek photo from my title in about 15 seconds. I saved a draft, "Test weekend trek to Nandi Hills" (Nov 14, 9:00 AM), without publishing it — it offers Publish when ready. The AI assistant is the standout: I typed "Plan a day trek to Nandi Hills for 5 friends" and it opened the Create Event form fully pre-filled — title "Day Trek to Nandi Hills", type Travel/Trekking, location Nandi Hills, duration 1 day, INR. The expense form now has an INR/USD/EUR/GBP currency picker (jan's ₹-only complaint is addressed). What still needs fixing: (1) Personal expenses are impossible. The dashboard's "Quick add expense" (₹250, "Chai with friends") opened a "Who is this expense with?" modal that only lets you pick People (0 — I have no connections), Groups (0) or Events (21 strangers' events, mostly Completed). There is no "Just me" option, so a new user cannot log a personal expense at all without attaching it to someone else's event. The Add expense button stays disabled until you pick one. (2) A brand-new account's "Coming up" section shows other people's events (russia backpacking in 1 day, WebSummit 2026, Hilversum Diwali, "5-Day tarkarli this will be a Itinerary") and "Discover nearby" lists events in Russia, the Netherlands and India — "nearby" for a user in Uzbekistan, with no location ever asked. One listed event, "TREK TO ALANG-MADAN-KULANG Oct 30", is already in the past. (3) The AI event generation matched the wrong Nandi Hills: the location dropdown offers only "Nandi Hills, Nandi County, Kenya" — the famous day-trek Nandi Hills is in Karnataka, India. The search needs to rank the Indian one first for an India-focused app. (4) Minor: the AI assistant took about 20 seconds to answer with only a spinner and no progress text, and the AI photo button's "Add a title first" hint is easy to miss. The site's hero still reads "The Operating System for Real-World Experiences" — it doesn't say what the app does (vivek's point stands).
PULSE: Know what you can spend, without doing the math, 8 Nov 2026: Tested on Web. Signed up by email (santaroisme@gmail.com) — a 6-digit code arrived from Gmail in under a minute, no Google, no card, no bank login. Onboarding was 6 steps: name; goals; how I get paid (I picked Salary — it did ask for my next payday, 1 Dec, and my monthly income, ₹60,000); banks (HDFC ₹25,000 + ₹500 cash); monthly bills (Rent ₹12,000 due 5th, Electricity ₹1,500 due 15th); priorities. The "You're set" summary showed the math: accounts ₹25,500 − bills before 1 Dec ₹1,500 − ₹1,800 "kept aside for surprises" = ₹22,200 safe to spend, ₹965/day for 23 days. The breakdown is explained this time — jan's unexplained number is fixed for the total. What worked: typing "chai 20" in the expense box instantly parsed to ₹20 in Chai & coffee, showed "Safe to spend after: ₹22,180", and the press-to-confirm works on desktop with a click (no swipe needed). "Can I afford ₹3,000 shoes?" in PULSE AI answered "Yes, comfortably. You'd still have ₹19,180 free until payday, about ₹833 a day" — the math checks out exactly. The member-card "Open PULSE" button works now (vivek found it dead-ended) and the summary screen led straight to the dashboard. No install prompt interrupted me on desktop. What still needs fixing: (1) The ₹1,800 "kept aside for surprises" is unexplained — it happens to be 3% of my ₹60,000 income, but the screen never says that or lets me change it. And the "2 monthly payments are kept aside" card conflicts with the −₹1,500 deduction: my rent (₹12,000, due 5th) isn't in the math because it renews 5 Dec, after the 1 Dec payday — defensible, but the screen never explains which bills count and why, so it reads as if ₹12,000 was forgotten. Name the bills in the deduction line. (2) Steps 2 and 6 ask the same thing in different words ("What are you here for?" — Track spending, Save more; then "Pick up to 3 priorities" — Save for something, Build emergency savings...). Merge them. (3) Rent: the flow never asks HOW I pay it (auto-debit vs manual) — it just assumes; and after onboarding, the dashboard still nudges "Add rent, phone and subscriptions" even though rent and electricity are already saved. (4) The bank picker is still the horizontally scrolling chip strip jan flagged — no search bar, no "x" to clear a selected bank. Amount fields also show unformatted digits while typing (₹25000, not ₹25,000). (5) Minor: the big safe-to-spend number counts up from zero on every dashboard visit, so the per-day figure flickers through wrong values (₹824, ₹790) before settling on the right ₹964 — it looks like a bug for a second. Note: I used approximate test numbers throughout, not my actual financial information.
Florim, 8 Nov 2026: Signed up by email (santaroisme@gmail.com) on the web — no Google, no card. Created family "Test" with USD currency. Two signup frictions: the currency defaulted to MYR (RM) · Malaysia on a US English browser, so I had to switch it to USD manually; and the date-of-birth field is a segmented mm/dd/yyyy input that can't be pasted into — I had to type each segment. I added an electric bill of $1,500: typed "1500" and it formatted to $1,500.00 correctly — the amount bug vivek reported (1500 showing as ₹15.00) is fixed in USD. The AI suggestion correctly offered Housing / Electricity and applied with one click. Then I added a $50,000 salary: also formatted correctly, and the Bills Receivable page now correctly labels the $1,666.67 figure "Daily average" — the "Monthly trend" mislabel is fixed. Home shows the right math: Family Finance Book $48,500.00 (50,000 − 1,500), with a "1 day streak" and a December forecast of $1,500.00. No 30-day Pro trial pop-up appeared at signup — another earlier complaint, fixed. What still needs fixing: (1) Portuguese strings leak into the English UI: the "Categorias" button on Bills Payable and "0% do total" / "100% do total" under the By-category donut should read "Categories" and "of total". (2) The Income trend chart is dead: with a $50,000 November salary, the line stays flat at zero and the y-axis labels read "0, 0, 1, 1, 1" instead of dollar amounts. (3) Date formats disagree: the new-expense/income dialogs show 11/08/2026 (MM/DD/YYYY) but the lists show 08/11/2026 (DD/MM/YYYY) — a US reader reads that as August 11. (4) The December forecast ($1,500) only projects expenses and says nothing about expected income, so it reads as a bill total, not a forecast of the balance — the "Limited data" warning is good, though. (5) No email verification step: the account was created and I was straight into the dashboard without confirming the email. Overall: the core loop works and the math is right; the remaining issues are localization polish and the dead trend chart.
Phosify, 8 Nov 2026: I built a site for a real business I work on — Koi Launch ('turn I wish there was an app for that into a real app'). I typed one sentence, and in under a minute I had a 5-section site: hero 'Bring Your App Idea to Life', services (Idea Validation, App Design, App Development), testimonials, contact form. It named the business 'Koi Launch' correctly everywhere. Saving was passwordless — I entered santaroisme@gmail.com, got 'Saved. We've emailed your website', and I was signed in (no password, no card). Free to build and preview; Publish is the paid step, which I didn't test. I then switched on the AI Chat Receptionist and asked it in the preview: 'How much does it cost to build an app with Koi Launch?' It replied: 'I don't have information about pricing. Would you like me to pass your question to the Koi Launch team?' — it did not invent a price, which is exactly the promise. Good. Three things to fix: (1) The auto-picked hero image is wrong — a black-and-white photo of fishermen pushing a boat on a beach, for a business about building apps. The image picker needs to match the business, or at least let me pick before generating. (2) The testimonials are still invented (James S., Sarah L., Mike R.) — they are now clearly marked 'Sample review, replace before publishing', which is the fix the builder promised, but a business owner could still publish them by accident; they should be empty until the owner adds real ones. (3) The services are generic ('Idea Validation', 'App Design', 'App Development') — fine as a start, but the builder could ask one clarifying question to make them specific to the business. The speed (under a minute) and the honest receptionist are the strongest parts — a real business could use this today once the photo and testimonials are fixed.
PawMD - Smarter Pet Care, 8 Nov 2026: Signed up on Web via Google as Santaro — one click, no password, no card. After Google it asked me to complete a profile: last name, country, phone number, city, state (I used test values — last name 'Test', United States, California, San Francisco, +1 555 010 1234; the phone field's own placeholder is a 555 number). Then 'Select your clinic': the only option was Virtual Veterinary Care, no matter the state — I picked California and still got only the virtual clinic, so the 'choose your clinic' step is really a single choice. Then 'Add your first pet': name, species, breed (searchable list — Labrador Retriever found), sex, age (3 years), weight (65 lbs). That made Milo (Dog, Labrador Retriever, 3y, 65 lbs) and the dashboard opened: 'Good evening, Santaro! How can we help Milo today?' I then tested the booking flow (8 steps) up to the service choice: clinic, pet, health info (reason, category, appetite, water, digestive), consent, services. Pricing is clear and upfront: Chat $24.99 for 30 minutes, Phone $34.99 for 15 minutes, Video $44.99 for 20 minutes. I stopped before payment. Three things to fix: (1) The pet's sex doesn't carry over — I set 'Handsome boy' in onboarding, but the booking form made me pick sex again (it was blank). (2) Only one clinic exists — don't present it as a choice, or say up front that in-person clinics aren't available in my state yet. (3) The booking asks for the pet's details a second time (name, species, breed, age, weight) even after I just added Milo — it pre-fills from the Milo chip, but it shouldn't ask at all. The dashboard and the upfront pricing are the strongest parts.
Neesh AI, 8 Nov 2026: Signed up on Web via Google as Santaro — one click, no password, no card, straight to the dashboard ('Your Projects / No projects found'). That part is smooth, and the dashboard itself is clean. The problem: I could not create a project at all, so the whole validation flow was unreachable. I opened 'Create Workspace', filled in Startup name 'Koi Launch', summary 'Turn I wish there was an app for that into a real app — connect the ChatGPT or Claude you already use, describe an idea in plain words, and share the app by link', industry 'AI / ML', stage 'MVP / Prototype', and clicked 'Start Copilot'. It failed with a 'Failed to fetch' notification and nothing was created. I retried, reloaded the page, re-filled the whole form and tried a third time — same 'Failed to fetch' every time. The dashboard still shows 'No projects found'. I never got to the Neesh Copilot, the 5 validation modules, a score, or Spotlight — the core of the product. Two other observations: the homepage hero shows 'Error loading Earth visualization / Failed to load land map data' (the globe never renders), and the homepage counters disagree — '1,000+ FOUNDERS, 23.7k SPOTLIGHT VIEWS' in the hero versus '0+ PITCHES, 0+ SPOTLIGHT ACTIVITY, 0+ FEEDBACK, 0+ INTERESTED' in Platform Activity. What I'd fix first: the project-creation API call. Nothing else in the product can be tested while 'Start Copilot' fails with 'Failed to fetch' — signup is fine, but the first real action a founder takes is broken. Would I use it for a real startup? Not until creating a project works.
iDrinkALot; track the night, not the morning after, 8 Nov 2026: Signed up on Web via Google as Santaro — one click, no password, no card, straight into onboarding. Onboarding is 6 pages: intro, profile (weight 165 lb, 5'10", age 30, male — needed for the BAC estimate), beer regions (I skipped with 'Not now'), weekly limit (default 7 drinks/week, about 14 units, with the plain-language line 'We'll warn you when you're close' — good), then Home. Skipping worked everywhere, which is good, but the flow is still long for a first session. The core loop works: I started a solo session at Home, added a Pilsner (500 ml, 4.4% ABV) and the app computed 2.20 units, 215 kcal and 'Sober by 10:16 PM' with a peak BAC around 0.015%. The weekly counter moved to 2.2/14, the session saved to the timeline as 'Private Home, 07:57 PM, 1 drink, 2.2 units', and I ended it with 'Just save' (I have no friends in the app to publish to). The drink catalog is big — 19 beers, 123 shots — with volume, ABV and kcal shown before you add. Three things to fix: (1) Desktop layout is broken — on a 1920px browser the onboarding pages are horizontally shifted, with the previous page bleeding in from the left and the active card cut off; I know this is a phone-first PWA, but the web link is what TestMyBuild testers get, so the desktop view should at least not overlap. (2) The numbers don't agree: the session card said peak about 0.015% while the timeline entry for the same drink said about 0.017% est. (3) The catalog is generic styles, not brands — the search box promises 'Search drinks, brands, breweries...' but I couldn't find a brand like Pacifico or Modelo; I had to pick a style ('Pilsner'). The 'Sober by' estimate and the units/kcal math are the strongest parts — lead with those.
Daily Captain, 8 Nov 2026: Signed up on Web via Google — quick, no password, and my own page (dailycaptain.app/santaro) was created instantly. Writing a goal worked: I added 'Finish the Koi Launch listing' and it showed under Today's Goals at once, with the day summary at '0 of 1 done today'. The problem: I could not actually use the app after that. Ticking the goal showed 'Ticking comes with the membership.', and the Midday check was membership-gated too. I declined the $49.99/year trial and the whole sheet became read-only — the 'Read-only' badge stayed on, and there was no way to tick the goal, do the midday check, or close the day. The core loop (morning goals, midday check, close the day) is entirely behind the paywall, so as a free user I could create a goal but not complete a single one. What I would change: let a free user finish one full day — ticking goals and the midday check — before the paywall appears, and show the $49.99/year price before the first goal rather than right after declining it. The three-moment idea is calm and clear, but I never got to feel it actually work.
LifeOverview, 8 Nov 2026: Retest after the builder's 6 November fixes - signup & dashboard I signed in with email and password (after a fresh password-reset link; my original reset email from 03:09 had expired, and the reset page handled that clearly). The landing page now exists and explains the product well - Free A$0 (Pay Splitter, up to 2 bank accounts, 3 stocks/ETFs) vs Pro A$5/month or A$50/year, 14-day free trial. My account is on the Free plan. The Home Dashboard still shows a lot at once, but navigation is now split into clear sections (Overview, Finance, Productivity, Fitness, Settings) and I found everything I needed without getting lost - a real improvement on the earlier "where do I add things?" problem. Pay Splitter (test 3) This is much better than the version described in earlier feedback. It is now a 4-step wizard: pay amount, pay date/frequency, bills, then review. I entered AUD 5,000 weekly, skipped bills with 0, and the review showed the default allocation correctly - Investments 40% (AUD 2,000), Savings 30% (1,500), Spending 20% (1,000), Emergency 10% (500), 100% allocated - and saving it recorded the payment once. Entering 0 to skip bills worked fine, so that earlier bug looks fixed. Two things remain: the currency selector now exists (a full currency list on step 1), but it still defaults to AUD without asking - auto-detect or ask on first use. And the review step explains the percentages, but a first-time user still has to guess what "Emergency" vs "Savings" should mean; a one-line description per bucket would help. Bank accounts, savings goals & investments (test 4) I added a bank account ("Test Savings" / "Test Bank", AUD 1,500 opening balance) and an investment holding (VAS, 10 units at AUD 100, portfolio AUD 1,000) - both saved and appeared on the Home Dashboard immediately. The app is clear that it does not connect to banks; balances are manual records. I could not find a separate savings-goals feature - if goals exist, they are not discoverable from the Savings Accounts page; if they don't, the listing's test list promises something the app doesn't have. Work Tracker & fitness (test 5) On the Free plan both are Pro previews with sample data only: the Work Tracker shows one sample task ("Plan tomorrow's top priority") which I completed, and the gym dashboard shows a sample Squat progression (60 to 65 kg). The samples give a good sense of the tools, but a free user cannot actually log a real task or workout - the most interesting parts of the app are locked before you can try them. Letting free users log a few real entries would sell Pro far better than sample data. Sync & saving (tests 6 & 7) My records persisted across navigation and were still on the Home Dashboard when I came back (Income AUD 5,000, Savings AUD 1,500, Investments AUD 1,000). The app is local-first: "Saved on this device only" with optional "Set up cross-device sync" / "Enable cloud sync". I did not test cross-device sync because I only have this one browser available, so I cannot confirm the earlier cross-device bug is fixed - the sync setup button is there, but the claim still needs a two-device check. I also did not sign out and back in, so session persistence after sign-out is untested by me. Layout (test 8, desktop only) Desktop layout is clean and fast, with no loading delays. I did not test mobile. What I would change 1. Ask or auto-detect currency on first Pay Splitter run instead of silently defaulting to AUD. 2. Add a one-line explanation to each allocation bucket on the review step. 3. Make savings goals discoverable (or remove them from the test list). 4. Let free users log a few real tasks/workouts before the Pro paywall. 5. Show on the dashboard whether cloud sync is on or off - "Saved on this device only" is honest, but it is easy to miss. Overall: the core money tools work and save reliably, and the step-by-step Pay Splitter and clearer navigation are genuine improvements since the first round of feedback. The remaining work is mostly about letting people try the interesting parts before paying, and proving sync works across devices.
RAI Calendar, 8 Nov 2026: Signup & onboarding I signed up with my real email and, after my first password-reset link expired, set a new password through a fresh reset link and was signed straight into the dashboard on the Pro plan (the builder promises a 14-day Pro trial on signup). The 5-step onboarding tour explains the main tools well, but dismissing it felt stuck: after I clicked "Skip tour" it sat on "Saving..." for a while before the tour finally disappeared. That first impression matters - the skip action should confirm instantly. Calendar Creating an event worked well: I added "TestMyBuild evaluation - test event" for Sun 8 Nov, 09:00-10:00, and it appeared immediately in both Month and Week views and stayed there after I navigated away and back. The Week view is the better of the two - it shows the event clearly in its time slot. The Month view looks quite empty when there are no events, and the dashboard shows a "Google Calendar disconnected" notice even though I never connected it, which is confusing for a first-time user. Plan & pricing The pricing dialog is clear: Free $0 (unlimited calendar/invites, Zoom & Meet, 5 @rai AI commands a day, 25 CRM contacts) vs Pro $7.99/month (unlimited @rai AI, CRM, timesheet and analytics). My account shows "Plan PRO" and today's @rai AI usage as 0/Unlimited, but I could not find anywhere stating when my Pro trial ends - adding the trial end date to the dashboard or pricing dialog would help people avoid surprise charges. To-do The To-do panel opened fine with Today/Inbox/Upcoming/All/Completed views, but it was empty with no hint of what to try first - a sample task or two would make the first visit less blank. What I would change 1. Make "Skip tour" dismiss instantly instead of hanging on "Saving...". 2. Show the Pro trial end date on the dashboard. 3. Hide or explain the "Google Calendar disconnected" notice until the user actually tries to connect. 4. Seed the To-do list with an example task. Overall the core calendar works reliably and the app feels fast and polished; the rough edges are all in the first-run experience.
AI Carousel Generator for Instagram & TikTok, 7 Nov 2026: Tested on Web. I signed up free with email (verified account, 300 credits, no card asked) and ran the builder's own test: a 6-slide carousel from the idea "why most people quit the gym". What happened: generation took 2 minutes 44 seconds and cost 82 credits, so the balance went from 300 to 218 — one carousel is more than a quarter of the free allowance, about three carousels before paying. The result itself is good: 6 slides, 9:16, dark moody gym photos that match each other. Cover: "Most people quit the gym in 3 months. Here's why". Then Reason 1 "No plan. Just wandering", Reason 2 "Too heavy, too soon", Reason 3 "Expecting results in 2 weeks", Reason 4 "Motivation fades. No system", and a fix slide "Boring routine. Show up. Repeat." The copy reads naturally and the photos fit each reason. Single-slide edit: I asked it to change slide 4's headline to "Expecting overnight results" on one line. It did exactly that in 44 seconds for 6 more credits, other slides unchanged, and its layout check reported no problems. That part works well. What I would change: 1. Show the credit cost before I hit send. I only learned the carousel cost 82 credits when the balance dropped afterwards. 2. A text-only headline edit re-ran the whole pipeline, including "Make the gym photos" and a layout check of every slide, even though nothing visual changed. A text edit should edit the text in place. 3. In the editor preview, the prev/next arrows sit on top of centred headlines on slides 4-6 and cover the first words, so at a glance I could not tell whether the text was clipped or just hidden behind the arrow. Move the arrows outside the slide or hide them until hover. 4. The assistant's summary in the side panel was cut off mid-sentence ("It's in a dark, moody gy…"). Would I post it? Yes — the carousel reads well and looks cohesive enough to post under my own account. But I would want the cost shown up front, because at 82 credits a carousel, I would think twice before regenerating.