Show HN: I made a site to quick identify any plant and learn how to care for it
frondly.appI'm still working on it, curious but what feature may be most valuable for you? What do you think about personalised care instructions, and an interactive chat feature for each plant.
This looks cool! I would be a causal user of this, but $7 puts it out of my price range, even with a free trial.
Though I think there is something interesting you are exploring here —- I imagine this is backed by an LLM API? If that’s the case, I would naively assume that I can get similar information using my chat gpt subscription directly — personally that’s where I find myself going for many of the random questions that come up in my life these days.
That brings up a couple of interesting questions that I would be curious to hear the results over time on (not that you have any obligation to share) 1) is there a wide audience that finds a value in this that don’t otherwise have access to ChatGPT/claude/whatever llm — and value this enough to pay just for this sort of ‘niche’ AI product? Or 2) alternatively — is the prompting/fine tuning/curation of the ai content you are providing better than what a naive LLM user could do on their own in a casual chat, that paying for this directly in addition to an LLM service would be worth it?
The de facto for Western plant species ID is Pl@ntNet/iNaturalist (free and citizen scientists will ID if there is uncertainty). Then you just look up care instructions? I would absolutely not trust ChatGPT.
I say Western as the training data is skewed by common species and usually they’re a bit geographically limited (for example BirdNET works best if you use a localised model).
Also if you use these free services, you can contribute natural training data which is valuable - even for well represented species.
I really wish per use pricing was normalized. I would gladly pay a few cents per use, and I'd use it a few times a month.
But a $7 subscription is far more than the utility I'd get from it.
The problem with this is also in the case of fraud refunds. I have a site where the subscription was only 2€/m but I had to increase it because when customers asked for a refund via their bank I had to pay 16€ in fees.
At what price would you pay for it? I'm trying to identify a price for an early consumer product in a different market
I probably wouldn’t pay a monthly subscription (which may very well just mean I’m simply not part of your target audience, and that’s fine) — though if I could use something super simple like Apple Pay to buy a few scans for $1, I would probably do that.
Interesting thank you. It looks like an additional resource-usage model could work in OP's case. Though with financial transaction fees starting at $0.33 of that $1, starting higher at 3-5 might be more appropriate.
A different product would be evaluated differently based on its usefulness and quality compared to free options.
As many others here stated, there are free trustworthy alternatives like PlantNet and iNaturalist. For now, even Google Lens is more reliable… until Google gets flooded with bad data and AI generated images of plants.
$3 seems like a better entry point for a product to test the market. Equivalent to a cup of coffee in most cities.
I think this idea is beautiful! I definitively would use it for some plants at home. But I'm not buying new plants each month, and even if I do, I don't know if I would use this for all of those. I have only 8 plants I would like to scan and forget, and the idea of subscribing throws me off, even if I can unsubscribe.
A better pricing schema for this, that also combat today's subscription fatigue, would be to sell X amount of plant scans. Like you can sell 10, 30 or 60 in different pricing scales. Pay once, the already scanned plants stay there in the users library. At least, I would find that pricing to be much more realistic and fair, and I suspect plenty of potential users are in the same boat as me. I will be able to personally scan the aforementioned 8 plants today, and 2 new plants in the long run, and it will feel great and fair.
Not to poopoo this, but there are multiple apps for this, some with master Gardeners behind them. Some that are straight up suggested by states for use, based on the input from their master Gardeners.
What makes yours different?
I have no idea what a master gardener is (a US thing I guess?) or why I would care if they endorse an app. But I have tried a couple of the top rated apps for this on Android and found them utterly terrible. I remember one result, I took a very clear photo of a bunch of bananas growing on a tree and got some kind of ferns as the answer. If this app can do better I expect there's a place in the world for it.
Try Flora Incognita.
https://floraincognita.de/
I assume they mean "a famous gardener". Here in the UK, there are at least two household names that present gardening shows, whose endorsement on such a product would be a huge pull.
That’s incorrect. See: https://en.wikipedia.org/wiki/Master_gardener_program
Ok, so it is a US (well, North American) thing? I stand corrected.
Thanks for your input! Yes, there are many apps out there. Originally, I made it for a friend who was paying $20/month for GPT. I really appreciate you taking the time to share your thoughts!
Such as?
PictureThis is the standard and it’s pretty damn good.
I wanted to see how it looks and what exactly it does without reading too much. Since you're offering a free trial I went for that. It was annoying that I had to create an account first, but I can see how that might be necessary. But now the page is asking me to pay even though I just want the trial you advertised? That's just sketchy. Other obstacles: no PayPal. Maybe this app really is just a wrapper around an LLM chat and that's why you're not showing it?
Edit: I can't even delete my account? This app seems just super sketchy now. My impression is that it's either a scam or build by someone lacking the necessary experience and skills.
Thank you for your feedback. I’ve deleted your account, so there’s no need to worry about that. This is the first launch of the app, and I’m working on resolving all the issues quickly. I appreciate your patience and understanding!
--- Jfkgjfkckfu Eurufjc cnsudjf si deleted. --
I will set 7 days trial without payment, from what i got from feedback
File a GDPR complaint to test if that is working
PlantNet has a ID feature that is based on the part of the plant, I.e leaf, flower, fruit etc that makes it much better at ID than just what you would get with Copilot or Gemini.
Companion planting, pests, common diseases and treatments tend to be other questions that we get ( as master gardeners )
Im kind of the target audience since I am trying to identify certain species of native trees in my new area for a forest garden project and am not very good at it yet. That said I’ve tried these ML apps and it didn’t outperform me googling so I kind of gave up.
Especially with saplings they may not show the characteristics of the mature plant well but you can use context clues like if the parent tree is next to it.
If this is just a wrapper around an LLM, what’s the game plan when a poisonous mushroom gets misidentified as something edible?
I have warning they should not ingest or apply to skin... Mushroom is not plant is it?
Mushrooms are not plants but berries are, misidentified berries can be poisonous.
Can your LLM always detect that? You realize that foraging is a huge reason people do this, and bad directions have very real consequences. From a culinary perspective, does it matter? Is it a growing, potential edible (technically not a plant the way tomatoes are technically a fruit) thing on the ground?
FWIW, even ignoring the fungi aspect, you probably have a related issue around small red fruits.
User error because mushrooms are not plants
just hope they ate enough to not be able to report it
I can't speak for anyone else, but the offer of a free trial made me bounce right off, despite my curiosity and the value such a product could potentially offer me. I want to know up front whether this is going to be useful to me before I sign up for anything. To that end, offering even a single no-strings-attached identification, even if the details are redacted, would go a long way towards conversion.
I get it but if you’re a solo founder without VC backing, completely free accounts are really difficult to support.
Free users can create a lot of support tickets. This could be good or bad.
Free users may never convert to paying, which isn’t ideal for bootstrapped ML business with expensive cloud costs.
Try Community Support (Forum/Discord/Slack, etc) for the free tier. Support Tickets/Email for Premium.
Creating features and fixing bugs based on Support feedback (Community Support or otherwise) may result in an amazing experience for the free tier, but not-as-good-as-it could be for your paid subscribers.
Wow, this really opened my eyes! I hadn’t thought about it from that perspective. hank you so much for helping me see that! I will completely remove subscription, change pricing model to credits, pay as you go.
I think it's better to continue give some free credits for new sign-ups.
I have to redo logic for credits system, so might take a time, how many credits you suggest would be helpful to test is properly?
Some startup business models and development processes don't work well if they collect massive amounts of free ephemeral users instead of a dedicated base when starting out.
Additionally, this would be a great way to gather tons of plant data that you can then use to increase recognition accuracy.
that's not free
Let me scan a plant before I sign up. This kind of specialty app requires proof before anything else or users skip it. I am.
That's valuable insight, noted! However, I need users to sign up before using the scanning feature. Without registration, it would be too easy for users to bypass restrictions and misuse the app.
I just use Google lens in the Google mobile app for this because the price is right
I thought Hacker News was legit place. But so many upvotes for this clearly shows it is not.
No it's not scam. I'm curious to understand your perspective in more detail.
From the FAQ: "Our database includes thousands of plant species, and we continuously update it. However, in rare cases or with very similar species, there might be slight inaccuracies. We always provide a confidence level with each identification"
As a casual user of LLMs (Gemini, ChatGPT) with multimodal capabilities, I've snapped a few pictures of random insects/plants and gotten pretty good identification out of the box.
The first thing you should do is point out how deep your moat is, and what makes it different. Your site says its uses AI and an internal database. I would give some clear examples then of how your product has better accuracy then any of the widely accessible LLMs already in use.
I used to use PictureThis a lot, as it was the most accurate I tried.
Any idea how this compares? For me personally, the other stuff is useless, I can Google care and such later. I'm after the most accurate, point - shoot - ID app out there.
I used PictureThis almost daily as a gardening tool, but as soon as ChatGPT accepted photo inputs I cancelled it and now just send pictures to ChatGPT with a prepared prompt asking it to ID the plant and give me details on its care and health etc.
It’s been better in every single way.
That's interesting. It's probably still more work than I'd want to do, especially when out and about. Is there an app that essentially does this? Does this one?
Is it an LLM wrapper or not? As far as I can tell you haven't answered that.
yes LLM, we use vector databases, adding now databases for non-western plants and system to learn if output is bad.
Based on your feedback, I’ll be switching to a credit-based pricing model. I’m also looking to build a community around this and add a gamified experience. Thanks so much for all your input—I really appreciate it!
Just a note, you have couple of places with different copy
> 7 days free trial.
> New users can enjoy a 1-day free trial of Frondly Premium
There's also a banner offering 3 day free trial. Are you testing which is the most popular time period for the trial?
My mistake, nothing is behind that
It’s a nice idea but the site says under support so if I were you I’d have wait until you had something to show that remains available before posting for feedback.
I know but if you cant see a lot of happen for last 12 hours :) I just want to have best for community, so trying my best to solve every needs, thanks for your time to write, if you mind write us email at support@frondly.app and i will notfiy you via mail with free credits to try it
My wife is very much into gardening, but also habitat. She'll look up endangered butterflies, find out what plants they rely on for laying eggs, food sources as larvae etc. She'll plant them in a corner of the garden and like magic, we'll see those butterflies within a year or two. In some cases there will be an insect that relies entirely on a tiny number of plants, but those plants may also rely entirely on it, or on something else. My point being that except for cultivated and commercially available plants, ecosystems are critical. Much of our garden's flowers are wild. Commercial flowers usually provide pollen, but not habitat, which is why she focuses on wild native flowers. I'll ask her to try the app, but I'm gonna predict that it won't be much use for these.
Usually when she's looking up how to care for a particular plant she uses Kagi to exclude websites with a large userbase in the United States. She prefers websites that are UK/Ireland based, or to a lesser degree, Northern European (using Google translate). Why? Because of subtle differences in plant naming, species etc and wildly different climate. She finds advise from U.S. based sites to be extremely questionable here, and you might waste a whole year trying to grow something before you realise that the advice doesn't work here.
But hey, every British band in the 60s and 70s wanted to break into America because they're are hundreds of millions of English speaking people living in a single TV nation. It makes sense for you to focus your efforts there. And whatever AI is behind your app is likely to have been trained on commercial plants in the US.
Thanks for sharing this thoughtful perspective! You’re absolutely right. What I’m thinking is to train the app with user input and gradually build a more region-specific database on top of existing models.
Thanks for highlighting this, it really helps guide where I can take the app!
You slapped together a wrapper around an LLM and are expecting to be able to charge $7 a month for it? Why?
"Personalized care instructions" - An LLM responding in a seemingly personalized way but with generic instructions is the average LLM chat experience. How is this different?
Congratulations on getting your MVP working! Ignore the whingers who say it’s too expensive. It’s not, and they are by definition not your target market.
Thanks a lot! But it surely open my eyes, with subscription model + need to provide more "real" value.
Wow! I am building something similar around this. Glad you built it. Too happy, but pricing is out of my range.
Can you share it, i would also love to hear your journey building it. Pricing models is currently changing
Why would I use this over what Picture this offers for free? The free trail sign up is really odd putting
So the site is inaccessible due to being down for maintenance only 16 hours after this Show HN was posted?
I really appreciate you take time to write this, but keep in mind i'm indie developer, i don't know if you ever build and launch anything, but this is my first launch :)
I’m not able to signup :(
How does your project compare with the completely free-to-use citizen science project and mobile app, PlantNet?
https://plantnet.org/en/about/
The pricing seems... optimistic? It's not immediately clear to me what this offers that Google Lens/Gemini, Chat Gippity, or Claudius doesn't already do for free or zero marginal cost. Also specialised apps with basically fine-enough free tiers like PictureThis. Look for and focus on developing a unique angle -- and offer more for the money (or ask for a lot less from users). It's not facetious to say someone could replicate the current core features (a few prompted API calls and a wrapper) in the time it takes to boil a kettle. What's your moat?
For anyone in Apple's ecosystem: photos on iPhone (I believe on Mac too but haven't checked) does this out of the box; details of how to use it are here: https://support.apple.com/en-au/guide/iphone/iph21c29a1cf/io...
As with their live text (ocr on an image), subject selection (remove background) etc, there are probably slightly better implementations in dedicated apps, but for most casual uses the built in is likely more than sufficient.
Just tried comparing pictures of two trees (inc leaf closeups) between PictureThis and native Photos app. PictureThis got it right, was consistent re species when comparing closeups to wide tree shot, and gave lots of extra info. Photos app got the genus right but different species for the closeup and wide shots; not very helpful. The lack of organised extra info is also a pity. One of the nicest features of PictureThis is how it shows your photo vs other matching photos from its database, letting you visually confirm if the result is accurate.
Thats why I said "for casual uses". Most people probably don't lookup plants often enough, with that level of specificity required, to justify $100 a year for an app that does just that - the built in one also works for animals and food.
When I've used the built in lookup it also shows me other examples of the species it identifies the plant as.
Good catch!
Sorry, but I have about 60€ in total for subscriptions per month.
I am not going over that budget (despite not being poor, it's more of a principle) and 7$/month is too much.
I totally understand that subscriptions are nice for the seller side. But for me as a consumer, they are not in the majority of cases.
It has a lifetime option for 30 dollars.
(Honestly) Cool, the problem with such options is: it's the lifetime of the app, not mine.
Love the idea!
Thanks!
https://www.picturethisai.com/ does something similar.
This one looks really good, thanks. It's looks like i want my app to go the same way as picturethisai.com
Pl@ntNet already exists and it's free
Is this a joke site? I can just ask ChatGPT.
We need more of this - identification and clear instructions rather than any clickbaity "10 best plants to have in your home" nonsense.
Would second the opinions in the comments that I'd need to know if this is going to be useful to me before I signed up to anything.
Noted! I will redo everything.
[dead]