... profile completion meter and a nagging banner that followed users until they filled it out. Completion rate after 90 days: 26%. We removed the meter

... talking. GBrain bundles $100 of usage into a $199 seat, which is fat margin or underwater depending on which model the user happened to pick that week. Those aren't three pricing mistakes. They're one. The meter measures something the vendor can see and the user can't feel. After building two products on usage pricing, the rules I'd defend: never charge for a retry you caused, show the cost of an action before it runs, and pick ... ... question is one I answer daily. I write copy for a free browser tool with zero credits, and even at $0 people still search for the fee they think is hiding. Have you tried a flat cap instead of metering

... study nobody is selling. There is exactly one large dataset here from an organisation with no product in the category: Pew Research Center's analysis, published July 2025, of 900 US adults who agreed to have their browsing metered for the month of March 2025. It covers 68,879 Google searches, of which 12,593 produced an AI summary. Three numbers from it get quoted. Users who saw an AI summary clicked a traditional result in 8% of visits; users ... ... didn't, 15%. Users ended their browsing session 26% of the time with an AI summary present versus

... testing via interviews and real users, and we made sure to account for users in different markets (ie., someone in the US was willing to pay more than someone in Australia). We also compared to competitors and other tools that our audience uses. That was all pretty successful. More recently though, we identified new use cases for our platform agency and enterprise buyers and the pricing model needs to reflect that. We've introduced an enterprise plan, which is negotiable ... ... ones for each premium feature) and then if the use veiwed, started, bought, cancelled or dismissed the paywall and what was the trigger. Comment from Varun Torka(@thevarun): Most of this thread is on the value side what to meter

... meter used to treat every change the same. Ask for a new image, one credit. Change the text on an image you already made, one credit. It reads as fair on a pricing page and it quietly taught people to accept whatever came out first, because a fix cost the same as another roll of the dice. We split the meter

... rather say them than imply something bigger. Things it deliberately doesn't do: no bulk or marketing sending, no calendar or drive, no lifetime deals. And if your problem is one 200 GB archive, storage is metered, so this won't be your cheapest option. Better you hear that from me now than at renewal. The launch is up if you want to have a look. Happy to take questions here too, including the awkward ones ...

... announced first. An agent can write the session down; Flare checks the story against the writes it watched and says what was left out or never happened. UI & UX One sidebar: explorer, a permanent Inspector (metrics, risk/reuse meters, symbols, importers), Changed, Structure the right-hand panel is gone and the canvas keeps its full width. Bare canvas: Lens/View/Re-layout in top-right menus, no floating folders drawer; Districts open fully visible, Wheel arcs tethered to the ring. Terminal: app-palette ... ... every design decision for your review so you can agree on a second time (or stop your agent beforehand if you prefer a more hands-on approach). This already goes in the direction of flagging the decision point between alternative

... monthly rate, requests pile up in a queue, one task active at a time, pause or cancel whenever. The bet is that scope negotiation is the thing people actually hate about agencies, not the price. So there is no meter. No estimates, no change orders, no proposal for every new request. You keep adding to the queue and we ship things one at a time. The part I'm not sure about: "unlimited requests" has been worn out by cheap ...

... through a short "audit" of a match, before or after a date, and gives you a Red Flag Score (0 100) , a breakdown of what's concerning, and a straight-talking counselor's analysis. It even has a Ghost Meter that estimates how likely someone is to ghost you based on how the date actually went. A few things I care about: Privacy first ! It never stores or shows the person's real name. Everything stays on your device ...

... worst-case provider cost before execution, settle the actual returned usage, refund the unused reserve, and refuse the call before upstream spend when the balance is insufficient. The difficult parts are not the checkout button. They are retry-safe meter events, signed and idempotent webhooks, reconciliation, failed-request releases, explainable ledgers and preventing a pricing rule from turning heavy usage into a negative margin. I have packaged that work as fixed-scope Stripe and AI billing audits and implementations performed ...

... sense. If usage is the expensive part, the annual discount is basically a bet on future behavior. If costs are mostly fixed and the value compounds over time, annual feels much easier to justify. Curious, if the metered part of your service became predictable enough, would you ever offer an annual option? Comment from Ben Cohen(@bencohen): @erez_aharonovitch1  Exactly. With normal software I can more or less guess what I m paying for. With AI tools it could ... ... contract. Mine is a mix, and the same test applies to me so I will describe it plainly: the infrastructure is fixed, roughly $140 a month whether one customer sits on it or ten, and the work itself is metered

... retry path, which previously had a proud total of zero: one pins that the second request really goes out carrying the same tools, the other that an endpoint refusing even under none gives up instead of hammering a metered API in a loop. 266 tests green, cargo fmt appeased, Scoop manifest caught up with 1.0.25. Big thanks to @ondrahracek for the fix and for verifying it against the live API rather than vibes. Get it here: https://github.com/nikserg/skim/... ...

... back to is that a run which fails your own quality check shouldn't bill at all. It moves the judgement onto you instead of making the customer fight to get it back. Not clever, just cheaper than the alternative. So if you're pricing on credits, what does your refund path actually look like? Most founders I ask haven't decided. They find out during their first bad week. Comment from Serdar D(@serdarships): Founder here, and I sell ... ... action-metered

... layer, managed: Tools give your agent capabilities without writing the harness RAG attach a knowledge base, retrieval handled Connectors plug into the systems your agent needs to read and act on Prompt orchestration chain, branch, and route across steps Metering every run is measured, so billing and cost visibility are built in, per agent And the part we care most about: you're not locked in. Export any agent as an SDK and run it inside your own product. Build ... ... once, deploy where you want. Metered

... automation. Doesn't scale, but might be correct at my size. Where I'm stuck is that I can't measure the thing I'd need to decide. I don't know what share of retries are genuine misses versus taste. I could ask on the retry "what was wrong with it?" but I suspect that mostly annoys the people I most want to keep. For anyone running something generative and metered

... above it. Our heaviest users are people going through something. The person journaling every day is in a hard month, and she is simultaneously the most expensive user to serve and the one I least want to meter. I don't have a clean answer to that. We made an unrelated decision last year to have the AI forget everything after thirty days, for trust reasons, and it happens to cap how much context every call carries ... ... generation path: empty response, refund the credit; model delivers way under the requested count, refund. That forced me to actually price a single call, which flat-rate never made me do. The trade I'd flag: credits move the metering

... RunEvr for the whole team. Works with Claude, Cursor, and ChatGPT. The best part: you don't pay us a penny extra. It runs on your own AI account - your existing free or paid Claude/ChatGPT plan. We don't meter it, don't upsell it, don't touch your bill. And one line we won't cross: your AI can never see more than you can. Same permissions as you - superpowers, not a backstage pass. Which AI do you live ... ... context without breaking your flow. Comment from Arash Rahimi(@irahimiam): Claude, most of my day too. First thing I'd want out of this isn't a task list, it's which of those tasks actually need my decision versus

... minutes of digging through benchmark tables and dev threads. So I built Reality Filter. You paste in a launch and get one verdict: worth trying now, watch later, too early, or mostly fog machine - plus a fog meter for how much of the claim is still unproven. Every verdict names its sources: real benchmark numbers, named reviewers, actual user reports. Never "reports suggest". Two things I did deliberately: Verdicts are dated, public, and I don't quietly edit them ...

... through benchmarks and threads before I trusted a claim enough to act on it. You paste in a launch and get one of four verdicts - worth trying now, watch later, too early, or mostly fog machine - plus a fog meter for how much of the claim is still unproven. Every verdict names its sources: real benchmark numbers, named reviewers, actual user reports. Not "reports suggest". Two deliberate choices: Verdicts are dated and public and I don't quietly edit them ...

... both settings and in the website features section. I am trying to make it work out properly as much as I can, so please do bear with me :) Comment from Rabnoor Singh(@rabnoor_s): @m1tul5  Dropping the meter for ambiguous sessions is the better call than any default. A safety net value is indistinguishable from a real reading at a glance, so it does not degrade, it lies quietly. Blank is honest and people learn what blank means ... ... m1tul5): @galdayan   Haha, yeah that was a close one :) Comment from Mitul Sheth(@m1tul5): @rabnoor_s Yes I have fixed Fable fallback in 2.1.1, it will report 1 M correctly now. I am planning to not show the meter

... built it because I kept rebuilding the same thing for different projects: extraction, storage, and some way to track what changed over time. Wanted that as the product instead of rebuilding it each time. Pricing is flat, not usage-metered. Free tier included, no card required: https://quorel.vercel.app Would love feedback, especially from anyone who's been burned by a scraper that silently broke, or wished they'd kept last week's data to compare against ...

... every rendering change we tried. App size was third. We were shipping every CPU architecture in one binary. Dropping to ARM only brought it to about 38 MB. That reads like a vanity metric until you picture someone on metered mobile data deciding whether your app is worth the download. The pattern across all three: the fix was never in the code that runs. It was in the code that waits, the code that repeats, and the code that ships ...

... feature needs an API. Our product speaks questions out loud and that runs on the browser's built in speech synthesis. Costs zero, no round trip. Worth auditing which pieces the browser already ships for free before you meter them. And gate your AI endpoints. We found ours could be hit by anything with the URL, a scraper could have quietly run the meter

... build: a desktop voice AI studio for people making real audio, not just testing one prompt in a browser tab. What changed: 100+ studio-quality voices Local generation on your desktop Scripts stay on your machine No per-character meter on paid plans Voice cloning from a consented sample on paid plans Voice Design and Omni in Pro for broader voice creation and 600+ languages Script workflow, retakes, mastering, and export in one app I built this because cloud ... ... pricing punishes iteration. Retakes, pacing fixes, pronunciation edits, and alternate

... guys! I'm Ameya, a recent grad who spent the last few months obsessing over a simple question: what if Wordle, but facts with number guessing? Built Estimate. A daily number-guessing game with a heat meter instead of color tiles. Launching tomorrow at 9:00 AM CST. Would love any feedback from this community before we go live: [ https://estimate-phi-beige.verce... ] (Will be getting a domain soon!) Comment from Rian Robertson(@rianbrob): Sounds like a clever spin on Wordle with ... ... that heat meter