I don’t mind a credit system. What I do mind: getting the incessant “Your workspace is out of AI credits” popup every time I open a document. Make it stop. It’s especially annoying on mobile, the popup takes up a large chunk of the screen.
2 Likes
How is that “technically false”?
I said that it was for premium users. Which both Notion and Taskade indeed have UNLIMITED AI for paid users.
Your reference to Notion only backs up my point. Of course they are going to have in their terms that they might “REDUCE” (not restrict) usage if a user is misusing AI.
And then your point about Taskade does the same in that it backs up exactly what I said. lol
Regardless, not sure what point you’re trying to make other than confirm that both Notion and Taskade indeed have unlimited AI plans for their paid users. Coda on the other hand has a highly restricted credit based system for all users.
1 Like
Coda over the past few weeks has gone bonkers with their changes. After being a hardcore Coda fan and bringing it into two organizations, I can’t recommend them anymore.
What was a snappy little amazing tool has turned into a mess of monetization. I’m fine with paying for AI usage but throwing a limit on it, with no warning from what I saw, no way to buy more, and even including the previous months usage having it hit against the new limit… purely poor decision making.
Seriously, why on earth would you have me exceed my AI usage before I even have the chance to know there was usage in the app?
2 Likes
The two 40-minute YouTube videos I requested for summarization drained all my credits. I can achieve this at no cost using Chrome extensions. Notion AI offer a better feature with much better size.
CODA must address this issue. For meaningful AI-driven work, these limitations are impractical. We either remain constrained to minor tasks within one or two columns of a single DOC or seek a more efficient system.
1 Like
How is that “technically false”?
“Technically” comes from technical. Marketing copy is not technical. According to the marketing you would conclude that they offer “unlimited” AI usage. Of course, marketing is known and valued for making sales, not for honesty and transparency.
And what I pointed out, is that in the case of Notion, they have undisclosed limits. Limits that they don’t tell you. So there are limits, therefore, it is not “technically” unlimited.
Taskade, on the other hand, has even less transparency than Notion on their page. If you read their pricing page, you would assume that AI is “unlimited” from the 8 USD Starter subscription, because they only mention credits for the Free account. But the reddit post that I shared earlier said that the “unlimited” starts at the 40 USD “Pro” subscription. They don’t mention limits anywhere but they do share that the underlying system is one of credits.
This is all marketing. And it clearly shows they’re not being transparent. There probably are limits, though no way to actually check.
Regardless, not sure what point you’re trying to make other than confirm that both Notion and Taskade indeed have unlimited AI plans for their paid users. Coda on the other hand has a highly restricted credit based system for all users.
I don’t know how you missed it (unless you didn’t read my post). My main point in all the posts, where I’ve even shown the math, is that a subscription model means you are overpaying for AI usage.
Maybe you should re-read the last part of the post where I compared the actual cost of the AI usage to the Notion subscription model, showing how much you need to generate to even break even.
As I repeatedly mentioned, I don’t care if you want to overpay for your usage through a subscription so you can feel that “unlimitedness” as long as there is the possibility for users to scale their AI usage cost according to the real value they get from such usage. It’s not really a big deal to set a cap according to your budget to avoid burning your credit card.
Of course, having a credit system, without the possibility of actually buying credits, makes no sense. I don’t know why Coda decided to deploy these changes without that in place.
I hope this clarifies to you what the previous post was about.
2 Likes
ALL of AI is run on credits! And yet, I use Notion AI constantly and have never hit any sort of “limit”. Why? Because I don’t abuse their unlimited usage plan.
This argument is like saying Verizon and ATT don’t provide unlimited text messages because they track how many you use. Now, why would they track something like that if it’s unlimited?
Perhaps because if they notice you are sending 1,000’s of texts a minute, you are likely abusing the system.
Since they track how many texts you send, it’s basically the same as another provider that limits you to only 1,000 text messages. So you might as well just use the other provide.
This is for some reason the hill you are wanting to die on for no reason.
1 Like
We hear you, and genuinely appreciate all the feedback. It’s especially helpful to have all the examples of how and where you are using the credits. We read every comment, and a sincere thank you for your candor and thoughts. We’ll have more news on the credit system soon, and if we can help in the meantime (whether that’s to understand where the credits are going or clarify what has happened with your account) please let us know at support@coda.io.
6 Likes
Thank you @DavidK. For those reading, the support team have been awesome on this issue. Can certainly recommend reaching out to them if your levels of frustration are red lining.
3 Likes
Since they track how many texts you send, it’s basically the same as another provider that limits you to only 1,000 text messages. So you might as well just use the other provide.
This is for some reason the hill you are wanting to die on for no reason.
Nice strawman argument. Apart from the fact that rate limiting requests and token limits are not the same thing, and we just don’t know whether Notion is implementing both…
You conveniently completely ignored in all your replies all that I wrote about my main point, that I’ve summarized for you in this last reply:
My main point in all the posts, where I’ve even shown the math , is that a subscription model means you are overpaying for AI usage.
Maybe you should re-read the last part of the post where I compared the actual cost of the AI usage to the Notion subscription model, showing how much you need to generate to even break even.
It always was about price per usage. I don’t care about arguing about definitions of words.
“Unlimited” as a marketing term is valid (as I’ve said previously). Whether it is technically true will vary company to company and how they implement their service.
As of today, 10 USD buys 5.000.000 chatGPT3.5 turbo tokens. Which is roughly the equivalent of Harry Potter’s whole 7 book series multiplied by 4. Most users are not likely reaching that level of usage. So you wouldn’t hit a limit even if you exceeded that usage, because you’d be subsidized by the host of all the users that don’t use their cost.
Paying extra for a subscription is a perfectly valid position! It would be great to have both, in the sense that casual users that get more stressed with math go for the subscription, and small businesses and teams that are mindful of costs can still use AI and scale it according to their application.
4 Likes
Especially when you compare the “unlimited” aspect, even if it’s just marketing as you say 
.
But let’s also compare the price. When it was released, it should have had a feature for buying more tokens (other than buying other DocMakers), it’s safe.
Since you like to compare to notion here are some prices: the AI subscription is 10 euros on top of the pro subscription and per user (10ePro + 10eAI per user!). if you have just Pro plan.
Emi say : “As of today, 10 USD buys 5.000.000 chatGPT3.5 turbo tokens per user. …”
That’s right. And the person who pays 10e, only he can use it.
Those who are in the same space and want to use it have to pay an extra 10 (per user)!
Sorry, but we can’t compare it to Coda.
Right now, we’re paying the same as before, with a little AI credit, so we can’t complain. ^^
Indeed, they should have thought of a token-add system. They’ll fix it, they’re very open and listen to the commu.
I think 2 possibilities :
Pay an additional subscription with an “unlimited”.
- per user ? why not ? notion works like this. but coda is not notion
- Risk of overpaying
Pay jeton, use your credit card and set a budget like GPT, which works very well.
- you really pay for what you consume. I think the best and most transparent way
A pleasure
2 Likes
I think people are ready to hand you money for AI credits. I know I am. Don’t get me wrong, It’s great that you have it built into Coda. However, this was NOT thought out. Please release an update for independent AI credit purchases… something. In most cases, your model would work, however for the months of using AI and even personally testing AI models, the problem is that using AI is not a constant thing. It has it’s uses and sometimes that means a LOT, and sometimes just a few things. Stand alone AI credits takes care of it.
3 Likes
2 days since the reset on AI credits and again they’re all used. So, now I wait another month for some more. Or I pay for another Doc Maker, which at this rate it would be a waste of time as it most certainly wouldn’t last for another week let alone a month.
Until there’s a better solution for AI credits, I’m getting off this treadmill and turning off AI; I’ll settle for my own intelligence.
5 Likes
Yeah I really enjoyed using Coda AI when it was in beta. It was pushing me more and more to using Coda over Notion, which is what I have been using for years. However, as soon as 4.0 went live, all credits were used day 1 and I haven’t been able to use the AI features since.
Coda AI was lengths and bounds better than other solutions. But as they’ve done many times before, they squandered the opportunity to gain market share.
And this is coming from someone that has been a Coda user since you had to be personally invited to even use Coda.
3 Likes
Hi Coda
Struggling as we go into November with the launch of our current AI payment framework, and as I shut down my client facing AI enabled systems. I’m working to rebuild these systems and set them up with a manual copy and paste workflow for moving data out to a 3rd party AI solution, and then back to Coda for storage. On my main project this amounts to 16 copy and paste operations per row on a system made for hundreds of rows per working day.
I’m reflecting on building and populating the core of this system within my first 24 hrs of using the AI Beta, delivering the first phase of this project and its data analysis to the client last week, and now walking the whole system back instead of delivering phase 2 in Coda due to the payment structure disabling AI for client services at any meaningful scale. It’s an unfortunate and painful experience.
I would like to ask how as makers we are understood by monetization. In practice: are our values and needs aligned?
1 Like
Hi @Nate_Gerber2 we will share some updates soon on what we’ve been working on behind the scenes, but in the meantime want to do everything we can to help. Can you contact our support team at support@coda.io? This will get a ticket logged so we can further dig in. Thank you!
1 Like
Hi everyone,
Thank you again for all the candid feedback on AI credits that you’ve shared with us though the community and other channels. Just to reiterate—we’ve read and discussed every comment. We’ve been working diligently on some updates behind the scenes, but I thought it might be helpful to offer some early transparency.
Taking a step back, our goals were for AI to feel included for the vast majority of makers, but we needed some limits to manage costs in extreme scenarios. We wanted to design a system that scales, and can cater to teams that have power users, while still giving flexibility for other team members to test out AI.
That said, we understand the credit system has created some challenges, particularly for soloists and small teams that use AI a lot, since they don’t have additional teammates to add as Doc Makers. We also heard your feedback about wanting better visibility into where credits are going, so they can be aligned with the use cases that provide the highest value. Given this, we quickly fast-tracked some updates, which we’re hard at work on building:
- We’re going to introduce AI credit packages so any team, regardless of size, can get the credits they need. For those who’d rather not think about credits at all, we will include an option to purchase “unlimited AI” for your workspace, billed as a monthly charge for each Doc Maker in your workspace.
- We’re adding self-serve dashboards to help you understand where your credits are going. While this has been on our roadmap, it’s been fast-tracked. If you have any usage questions in the meantime, our support team is happy to help answer questions about your account.
- We’re creating a richer preview window for AI column prompts. Currently, if you add AI to a table column, you can see a small preview, which does not deduct any credits. We’ve been working on a larger and more detailed interface, so you can feel confident in the output before using credits to fill every row.
- We’re making updates throughout the product to improve credit efficiency. We’re changing some of the default settings and doing further research on how we can help drive credit usage where makers are getting the commensurate value.
We have additional product updates in the works that will help makers get even more value from Coda AI, and we’ll share further details soon. Thank you again to everyone who has been engaging with us. We’ll share some more news with you all as soon as we can.
Thanks,
David
Coda AI Product Lead
15 Likes
Re: Coda AI, I was hoping to be able to use for the following kind of usecase. It currently doesn’t work for this at all, and it is clear that it would cost a boatload of credits even if it did work.
Usecase:
I have 2 tables that have similar columns and similar row values, but they are not the same.
(Specifically, my usecase is comparing the high school math standards from different states. They’re very similar but not identical in text, but the IDs are different, the orders are different)
I want to use an AI tool to do a relation from one table to the other and link the standards that are similar.
This should be something that a vector embeddings database system should be able to do, and LLMs should be able to do (even if you have to chunk up the context window)
1 Like
Thanks for the update @DavidK .
I don’t get the sense from this message or any previous messages that Codans understand this.
Maybe you can clarify the nuances surrounding generative AI and how Coda views the subtle persona differences between Makers who are power users vs Makers who also seek unknown and uncharted pathways to innovate with Coda AI.
2 Likes
Oh, no-- I understand it. I was considering cancelling my plan until I saw this. In teams where there are many people, there are larger numbers of credits available and it sounds like teams have “power users” who consume the majority of the credits. When you have a small team, there is no such possibility and adding new doc makers for a few more credits becoming incredibly expensive.
I tend to use it for tasks related to datasets requiring text-based analyses, rather than a finished product, if that helps.
Hi everyone,
I’m an Engineer on the Coda AI team. As David mentioned, we’ve been working on a richer preview window for AI column, making it easier to view AI-generated results based on your current prompt.
These previews do not consume any credits, so feel free to iterate on the prompt as much as you’d like, and to view the corresponding results for each row. When you’re happy with the output, you can select Fill column to apply to all rows.
Our team will continue to share broader updates on Coda AI functionality and the credit packages. Let us know if you have any questions about the new preview window.
3 Likes