4 months in. 3 Chrome extensions live. First mobile app done. Still £0. Here's my honest June update.
I'm not going to dress this up.
It's mid June. I've shipped three Chrome extensions and a mobile app in four months as a solo CS student with no funding and no team. Stripe is at £0.
That's the honest state of things. But here's what's actually happening underneath that number.
What's live right now:
Prompt Helix --> AI assistant for any webpage. Ask ChatGPT, Claude, or Gemini about the page you're reading without copy-pasting a single word. v1.0.3 shipped last week with PDF support, Gmail and Google Docs working, a free demo flow so new users see it work before setup, and result history. 179 total installs.
chromewebstore.google.com/detail/prompt-helix/ffjppocigpeamhokbpnknlplkbccjpin
FocusForge --> AI focus tool with time tracking, site blocking, grayscale mode, and a Nuclear Option that locks every distracting site for up to 8 hours with zero bypass. 76 total installs.
chromewebstore.google.com/detail/focusforge/hdkabchfflgnnonnhffkcmhgbenfoaci
CookieNuke --> Auto-rejects cookie banners, categorises trackers into advertising, analytics, social and fingerprinting, gives every site a privacy risk score, and shows a dynamic shield in your toolbar that goes green, amber, or red based on what the site is trying to do to you. Just launched this week.
https://chromewebstore.google.com/detail/cookienuke/ibfgjdjeckhcgdkpeiphocfioacgnpom
All three are BYOK. your API key stays in your browser, never touches my servers.
Helix Decide is the first mobile app, submitting to Google Play July 1st:
Take a screenshot, paste a link, or describe anything you're thinking about buying or committing to. Helix Decide gives you a structured decision card and verdict, confidence score, reasons, risks, missing information, what you'd regret.
Four modes: Buy?, Worth it?, What's the catch?, Risk Check.
Built in under two weeks. Auth working. Decision cards working. Camera input working. Locked modes working. Just waiting on the production build.
The honest activation problem I'm still solving:
290+ total installs across three products. 1 weekly active user. The people who install aren't experiencing the value moment before they forget the product exists.
For Prompt Helix the fix was a demo flow where one free proxied call so new users see it work on the page they're already on before being asked for anything. That went live three weeks ago. Still waiting to see if it moves the needle.
For FocusForge the fix needs to be making the first block feel significant — not a quiet banner but a real moment. That's the next update.
If you've installed any of these and it didn't click I'd genuinely love to know why. What happened in that first session? What was the page you tried it on? Did the demo flow work for you?
And if you've been thinking about upgrading Prompt Helix Essential is £10/month for unlimited queries across ChatGPT, Claude, Gemini.
FocusForge Essential is £7/month for AI coaching and Nuclear Option.
CookieNuke Essential is £7/month for AI tracker explanations.
The free tiers are genuinely useful. The paid tiers are where the full product lives.
helixlabs.studio
The 290 installs / 1 weekly active split is the honest version of a thing most people hide, so respect for posting it as-is.
One reframe that helped me on a free, no-signup tool: install and activation are two separate funnels with two separate fixes, and it's easy to pour energy into the wrong one. Sounds like you've correctly diagnosed yours as activation — the value moment — not acquisition, and the demo flow is the right instinct.
One cheap adjacent lever while those fixes bake: the store listing itself. I once had a landing page that ranked fine but almost nobody clicked, because the title described the feature ("free X tool") instead of the words people actually type when they have the problem. Rewriting it to match real intent moved click-through massively — same ranking, same product. Chrome store titles and first lines work the same way: someone scrolling results decides in about a second whether this is for them. Worth checking whether yours lead with the problem people search for or with the feature name.
When the FocusForge "first block feels significant" moment is built, I'd be curious whether activation moves — that's the more interesting number than installs anyway.
the two separate funnels framing is exactly right and I've been conflating them. acquisition is working.. 290 installs from organic Chrome Store search with minimal promotion. activation is broken — people aren't reaching the value moment. fixing acquisition when activation is the problem would be pouring water into a leaking bucket.
the store listing angle is one I haven't looked at properly. you're right that "Prompt Helix" and "FocusForge" describe the product not the problem. someone searching Chrome Store is typing "AI assistant for webpages" or "stop doomscrolling Chrome" not the product name. if the first line of my description leads with the feature rather than the words they typed to find it I'm losing clicks I already earned through ranking.
going to audit all three listings this week against real search intent. that's a free fix that could move click-through without touching the product at all.
and yes the FocusForge first block moment is the next build. will report back on whether activation moves when it's live.
The thing that jumps out: 290 installs split across three products is really ~97 each, and none of them has enough weekly users to learn from. With one weekly active user you can't tell a discovery problem from a value problem, and those need opposite fixes. I'd pick the single product with the least-bad retention, put the other two on pause for a month, and go deep. One product at 290 installs teaches you far more than three at 97.
On the demo flow: a single free proxied call proves the thing works, but it doesn't build the habit. For an "ask about any page" tool the value moment isn't the first answer, it's the second or third time someone reaches for you instead of opening a new tab to ChatGPT. So the number to watch isn't "did they see the demo," it's "did they come back within 48 hours." Instrument first-query-then-second-query, not installs.
And the reason extensions get forgotten is they sit in the toolbar competing with muscle memory. People already have a reflex for "I have a question": Cmd+T, chatgpt.com. You only win if you're faster than that at the exact moment it fires. A keyboard shortcut triggered on a text selection, or auto-surfacing on the page types where people actually get stuck (docs, dashboards, long articles), beats a button they have to remember to click.
I'd hold off on the paid tiers entirely until one product has 20-30 weekly users. At 1 WAU, pricing isn't what's stopping you, and that tier work is time you're not spending on activation.
this reframe stings a little because it's correct. 290 across three products is 97 each and 97 installs with 1 WAU doesn't tell you anything actionable. you're right that I can't distinguish discovery from value problem at this sample size.
the keyboard shortcut point is the one that hits hardest. the reflex is already set — problem appears, Cmd+T, chatgpt.com. I'm not competing with ChatGPT's product, I'm competing with a muscle memory pattern that fires before anyone consciously decides to use a tool. a keyboard shortcut on text selection that surfaces Prompt Helix at the exact moment the question forms that's competing at the right layer. building into the next update.
on holding off paid tiers until 20-30 WAU is honest pushback. the tiers are already built and live, pulling them now creates more confusion than leaving them. but the point about where to spend time is taken. activation over conversion work at this stage.
the 48 hour return metric is the right number to watch. not did they see the demo, did they come back. going to instrument that specifically.
one question....when you went deep on one product over others, what was the signal that told you which one to focus on?
I'd separate two things you're lumping together: installs and activation. 290 installs is a distribution win; 1 WAU is an activation problem, and the two need totally different fixes. The number that actually predicts retention is time-to-first-value — seconds from open to the user feeling the product do something for them. On my little memo app the whole value is one tap (capture, gone), so there's almost no gap to lose people in; the one time I buried value behind a settings screen, nobody reached it and I never knew why. For FocusForge, could the very first block fire in session one before any setup — so the Nuclear Option is the first thing they feel, not the last thing they configure?
time-to-first-value is the exact metric I've been ignoring in favour of installs. the gap between open and feeling the product do something is where everyone is disappearing and I haven't measured it at all.the FocusForge question is the right one. right now the flow is install, open, configure sites, set limits, then wait to get blocked. that's four steps before any value fires. the product is passive until the user does setup work first.your memo app example makes the contrast obvious. one tap, value, done. FocusForge is the opposite. setup first, value maybe later if they remember to use it.the Nuclear Option being the first thing they feel rather than the last thing they configure is a genuinely interesting inversion. what if first open showed a single prompt "block everything distracting for the next hour?" one tap yes, Nuclear Option fires immediately, they feel the product working before they've configured anything. that's session one value without any setup friction.going to prototype that. one tap to immediate value on first open is probably worth more than any retention email I could send.
Yeah, you've got it — and the prototype framing is right. One thing I'd flag from the one-tap side, though: instant first-run value is necessary but it quietly creates a second-session problem. A chunk of my early users did the magic tap once, got the hit, and never came back, because the value was complete and left no trace. What fixed it wasn't onboarding — it was making the action leave something visible they'd want to return to. So after the Nuclear hour expires, I'd show them what they reclaimed ("you just bought back 47 minutes from X") so the win is legible and pulls them into session two. Time-to-first-value gets the install to fire; time-to-second-value is the thing that actually predicted retention for me. Are you instrumenting what happens the moment the block lifts?
Your website only described what it is, it's not showing me anything. Why no screenshots, gifs/video's of it all in action? It really raises the bar for me. I have to connect all the dots myself and take the install plunge to find out what it really does. Show, don't tell.
fair and completely valid. the site describes the products but doesn't show them working. someone landing on helixlabs.studio has to imagine what it looks like inside the extension before deciding to install . that's too much work to ask of a cold visitor.screenshots and a short GIF of Prompt Helix analysing a real page, FocusForge blocking a site with the time saved visible, CookieNuke shield turning red on a tracker-heavy site those three visuals would do more than any amount of copy. going to fix this week. highly appreciate the direct feedback. show don't tell is the right standard.
Respect for shipping this much and posting the zero honestly. Most people hide that number. I am going to be blunt because it will help more than a compliment: 1 weekly active user across 290 installs is not an activation-copy problem, it is a focus problem. You have three products and a fourth in the oven, solo. That is not three shots on goal, it is one person's attention split four ways, which is exactly why none of them have caught. The market is quietly telling you nothing has hit resonance yet. What I would do: pick the one with the best signal (Prompt Helix at 179) and freeze the other two for 60 days. No new features, no fourth product. Then go interview the people who installed Prompt Helix and never came back. Ten of those calls will teach you more than another launch. You are clearly good at building. The reason you are at zero is that building is the comfortable part, and the uncomfortable part (one product, real users, hard conversations) is where the revenue actually lives. If you could keep only one of these, which would you bet on?
this is the comment I needed to read even though it's not the one I wanted to.the focus problem diagnosis is probably right. three products plus a mobile app is four directions for one person's attention and none of them have hit resonance yet. the market hasn't been quiet about this 290 installs and 1 WAU is a clear signal that something fundamental isn't working, not three separate activation problems waiting to be fixed.the interview suggestion is the one I've been avoiding. I know how to build. I don't know how to sit with someone who installed Prompt Helix and never came back and ask them what happened. that conversation is uncomfortable because the answer might be that the product isn't what I think it is.if I could keep one Prompt Helix. it's the oldest, has the most installs, solves the most universal problem, and the v1.0.3 improvements are the most significant changes since launch. if anything has a shot at resonance it's that one.but I'm not going to pretend I'm freezing the others. FocusForge and CookieNuke are live and generating organic installs daily. I'm not pulling them down.what I will commit to — no new features on FocusForge or CookieNuke for 60 days. Prompt Helix gets the active improvement work. and I'll find ten people who installed and ask what happened.how did you find those conversations when you ran them?
290 installs and 1 weekly active user is actually really useful data — it tells you the problem is activation, not discovery. People found you, they just didn't reach the moment where your product clicked for them.
The free demo call idea for Prompt Helix is smart. Sometimes people need someone to show them the value before they can feel it themselves. What does the first 60 seconds look like when someone opens the extension for the first time?
the first 60 seconds is honestly still too passive. someone installs, the popup opens, they see the interface with intent buttons and a provider dropdown, and then nothing happens until they do something. there's no guidance on what to do first or what page to try it on.the demo flow in v1.0.3 was supposed to fix this — one free proxied call so they see it work before setting up an API key. but even that requires them to click a button and be on a page with readable content. if they open it on a new tab or the Chrome Web Store page it still doesn't work.the honest answer is the first 60 seconds probably goes — open popup, look at buttons, not sure what to do, close it, forget it exists. the value moment never fires because there's no guidance toward a page where it actually works.what the first 60 seconds should look like — open popup, see a single clear prompt "you're on [page title], want to summarise it?", one click, answer appears. zero configuration, zero friction, immediate value on whatever page they're already on. that's the version I haven't built yet.
The gap between 290 installs and 1 WAU is an activation problem, not a product problem — and you already know that, which puts you ahead of most people at this stage.
The demo flow for Prompt Helix is exactly the right move. The install-to-value window is probably under 3 minutes for browser extensions: if someone doesn't feel the thing work in that window, the extension icon joins the forgotten toolbar graveyard permanently. One proxied call on the page they're already reading is a good answer to that.
The harder version of this problem is FocusForge. Productivity tools have a cold-start problem where the value is delayed — blocking only feels useful when you're about to lose focus, not when you're calm and setting it up. The "significant first block" instinct is right, but I'd think about it as: what does the moment feel like when they've just stopped themselves doing something they were about to regret? That's the emotional hook. Lean into that.
290 installs in 4 months solo as a CS student is the foundation. The activation rate is fixable. Most people who hit $0 for this long are stuck on the product, not on the experience of the product. You're already asking the right question.
"the moment they've just stopped themselves doing something they were about to regret" — that's the emotional hook I've been trying to articulate and couldn't. the block itself isn't the value. the feeling of having caught yourself before the spiral is the value. that's what needs to be visible and celebrated in that moment not just a banner saying "site blocked."
the cold start problem for productivity tools is real and it's why the Nuclear Option on first open idea from another comment in this thread is interesting. don't wait for the user to almost lose focus — give them the feeling of control immediately on first session before they've even set anything up. one tap, everything blocked for an hour, they feel in control before they've configured a single thing.
the install to value window being under 3 minutes is the number I'm designing against now. every step between open and feeling the product work is a step where someone leaves permanently.
appreciate the framing — most people stuck at £0 are stuck on the product not the experience of the product. that distinction is exactly where my attention needs to go next.
Four products in four months is real output, not just aspirational. The value-moment gap you spotted with Prompt Helix is the same thing I kept hitting early with DictaFlow. Users would install, poke around settings, and close it without ever actually dictating anything. The fix was forcing the first dictation to happen in under ten seconds: hold a key, speak, release, text appears. No setup, no config tour, just the core mechanic immediately. Once someone sees their words appear at the cursor, they get it. Without that forced value moment, the product is just another icon they installed. Is Helix Decide doing anything to shorten the gap between install and first real decision output?
the DictaFlow comparison is exactly the right parallel. hold key, speak, release, text appears — that's the entire product demonstrated in under ten seconds with zero explanation needed. the mechanic proves itself before the user has formed an opinion about whether it's useful.
Helix Decide is closer to getting this right than the extensions are. the home screen has four quick mode buttons — Buy?, Worth it?, What's the catch?, Risk Check — so the user immediately understands the product range. but there's still a gap between seeing those buttons and getting a real decision card output. they have to type or paste something first.
the forced first decision idea is interesting. what if the onboarding asked "what's one thing you're deciding right now?" before anything else — even something small like "whether to buy these headphones" — and immediately ran it through the decision engine. first thing they see is a real verdict card on their actual decision not a demo. that's the mechanic proving itself in under 30 seconds.
going to add that to the Helix Decide onboarding before July 1st submission. LifePilot suggested something similar earlier in this thread — the comments are converging on the same fix from different angles.
290 installs and 1 WAU is actually useful signal — it means your distribution is working but the first session isn't. You've already diagnosed it correctly with the demo flow for Prompt Helix, which is the right instinct.
For FocusForge specifically: "making the first block feel significant" is exactly the right direction. The way to do that is to make the outcome tangible immediately. Something like showing "You were about to spend ~20 min on [site] — blocked." The number makes it feel real. Right now most blockers just say "this site is blocked" and users feel annoyed, not grateful. The moment they see time saved, the value proposition clicks.
One thing worth watching for Helix Decide: the aha moment for a decision tool depends on the user having an actual pending decision when they open it for the first time. If they open it out of curiosity with nothing to decide, they'll close it and forget it exists. The onboarding flow probably needs to ask "what's one thing you're deciding right now?" before anything else — even if they type "whether to buy X" it creates the context that makes the whole app make sense.
the "You were about to spend ~20 min on [site]. blocked" framing is exactly what I've been missing. right now FocusForge just blocks silently. the value is happening but the user never sees it articulated. making the time saved visible in that block moment turns an interruption into a win. going to build that into the next FocusForge update.
on Helix Decide the "what are you deciding right now?" onboarding question is genuinely the missing piece. the whole app only makes sense when there's a real decision in the frame. asking that upfront before anything else creates the context that makes the first decision card feel relevant rather than just a demo. adding that to the onboarding flow before July 1st submission.
i appreciate both of these thanks. they're the kind of specific actionable feedback that actually changes what gets built.
Appreciate the honest update. Most people only share wins. The consistency of shipping 3 extensions despite £0 revenue shows real commitment — that's the hardest part.
thanks.. honestly the £0 is the hardest part to keep posting about but it feels more useful to share the real picture than wait until there's something to celebrate. the shipping has kept me going more than the metrics have. each product makes the next one easier to build and the ecosystem more complete. revenue will follow its just taking longer than I'd like.
I'd be careful treating this as an activation problem too quickly.
The interesting question may not be whether users are reaching the value moment.
It may be whether the people installing the products expect the same value you're actually delivering.
Those sound similar, but they can lead to very different conclusions about the buyer, the product, and which signals deserve attention.
I wouldn't make that call casually from the current data.
you're drawing a distinction I've been glossing over. activation problem assumes the right people are installing and just not reaching the value. expectation mismatch means the wrong people might be installing entirely people who expect something different from what the product actually delivers.
the honest answer is I don't have enough data to know which it is. Chrome Store doesn't tell me what search query brought someone in or what they expected to find. the post-install "what brought you here today" question hilgard suggested a few weeks ago is probably the only way to actually answer your question are the people installing expecting what I'm delivering or something else entirely.
which problem is it for you? genuinely curious whether you see something in the positioning that suggests expectation mismatch specifically.
Possibly.
The reason I stopped short is that I don't think the useful part is whether I see an expectation mismatch or not.
I think it's the decision that follows if that turns out to be true.
That's the part I'd be careful with.
Not because it changes activation.
Because it can quietly change who the product is really for, which feedback gets trusted, and what future signals end up looking like validation.
I wouldn't try to unpack that properly in a thread.
If you're curious, drop your email and I'll send over the tighter version.
you're right that the decision that follows is the harder part. identifying expectation mismatch is diagnosis. what you do with it — whether you chase the audience you have or the audience you wanted — is the part that quietly shapes everything else.
Exactly.
The reason I keep separating the diagnosis from the decision is that the diagnosis can be correct while the decision that follows is still wrong.
That's the part I'd be careful with.
I wouldn't try to unpack it properly in a thread.
If you're curious, drop your email and I'll send over the tighter version.