Hey @Arbor_Advice!
Your hunch is spot on! The MCP server opens up many more possibilities for integrations between Coda and a variety of agents, including those deployed on Go.
At the current time, the Go assistant does not have these abilities, but we are working on it 
@Emma_Wyatt thank you for reaching out. The team has been fantastic. The way I signed up was an edge case that nobody anticipated. This merge of Coda, Superhuman and Grammarly is like a 3 Body Problem - totally unpredictable. The team was responsive and really keen to understand exactly what I did to reveal this bug in the process.
thanks! i’ll check it
Hi @Stefan_Huber!
We’re working towards sub-doc sharing and know this is a top request. Many requests to share pages are for pages with data on them, where it’s important that users see only the data shown in views on that page.
We’re focused on those use cases first, and one of the key goals in our work to separate data from docs is to make it possible to share parts of a table without sharing the full doc or table.
We’ll enable ways to share views of data directly first, then over time add the ability to share pages of text content or mixed text and data as well.
Hi @AJM, we understand the frustration. Our investments to support massive tables and databases will allow us to lift the current cross-doc and API limits.
Can you share anything regarding the timeline for the data scaling improvements Shishir mentioned in this post?
When? I was told that more than a year ago. My doc can no longer do simple calculations of formulas despite being extremely simple.
Stefan, I share many of your sentiments regarding the slow progress toward these essential features and basic needs. And I think your prediction concerning possible vibe replacements on the horizon is more than a definite maybe. It will almost certainly happen, and sooner than we think.
However, when it does, a tiny segment of today’s customers will find safe harbor in custom AI-crafted systems built to mimic many aspects of Coda and overcome several of the limitations that we have been waiting for years to be addressed. This is not to suggest that you or any of us are too insignificant for Superhuman to care. It is simply an attribute of a growth market whose technical foundation is presently unable to accommodate the many avenues of use cases that, while known and understood, fall into an increasingly narrow band of users.
If you look back to 2019 and compare the application requirements that Coda addressed at that time with the apps we are planning for 2026, there are significant differences. This was a harsh transitory period when data exploded and app complexity soared.
It’s not surprising that Coda was caught off guard regarding table size capabilities. Most companies in the low/no-code space struggled with this.
Data size is one of several pressure points, but it is a big one. AI was another big one. We didn’t know what we didn’t know, and to a large extent, this remains true in late 2025.
The small segment that finds a better way [ironically] using AI to displace/replace Coda and other tools was precisely the part of the market I wrote about a year ago. Coda is rapidly becoming less relevant to a slice of its original customers, and they will likely move on. But Coda’s customer base will also grow massively for different customer segments.
@Bill_French , as always it’s nice to have your thoughts. My main point is that the reasons why anything under the current organisation will likely fail is not the technology; it’s the culture. Hiding / not addressing concerns, ego-plays, secrecy, mind-games… that’s ill culture - far more obvious than Shishir might think.
My worry is that SuperhumanGo doesn’t stand a chance to Comet or Atlas or ChromeAI (whatever Google call it) in agents architecture. If nobody in this community cannot understand the goal and target segment of Superhuman, then we can’t recommend it to anyone. And any product with that “clarity” is pretty much doomed.
On the other hand, Coda could be relevant until my generation and the one after me still exist, but only if it delivers.
This is a hyper-accurate assertion if – and only if – you will indulge me by adding a few predicates. One that comes to mind is the customer persona (or personas) who will reject Superhuman and embrace Comet, Atlas, or Chrome browser tools. My perception is that the vast universe of Grammarly’s paid users has no interest in the bubble that surrounds agentic browser tools. Most Coda users also fall outside this bubble. These agentic tools are not yet packaged in a way that will appeal to users who want to make their work easier.
So what persona(s) remain that we could embrace to support your assertion? My feeling is that the overlap in the Venn diagram exists. Still, it is so small that Superhuman (going forward) cannot justify a technical posture that positions it in that bubble, which is currently occupied primarily by developers and power users. This is my way of saying that the agentic approach is not Superhuman’s jam. 
I can’t speak to anyone’s thinking. However, I believe that any company’s seemingly aloof attitude could be a misinterpretation of an unfamiliar reality. It is the data I use to understand the company’s future trajectory. They’re sending signals, many of which can be interpreted as uncertainty, and that’s why I continue to hedge bets with local-first decisions. However, one company’s signals can be perceived by certain users as a lack of care or a deliberate manipulation.
My experience with @shishir and the Coda team suggests a genuine commitment to achieving customer success with high-value tools. However, I’m under no illusion that the roll-up of Superhuman, Grammarly, and Coda would alter the definition of ‘customer’. I think some of us have, or will soon, fall out of that collection of targeted personas.
Some random thoughts around this:
Background: Many years ago I did strategic planning for Toyota South Africa. One of the things about ‘forecasting” that I learned during that time was: “Futurists always overestimate the amount of societal change, while underestimating the amount of technological change that will take place”.
All progress is non-linear, and hopefully it is exponential with a exponent larger than one. AI is progressing with a very large exponent. And the faster it grows, the more people it leaves behind. Working to include as many people as possible is going to be a very necessary endeavour - and one that I think quite a few makers in the Coda space is working very hard to achieve.
But as with all of major changes, it will have to move beyond being a fad, and into something that is really useful. And the scope of AI influence is vast, and it will continue at an uneven pace. In this forum we are debating the Superhuman/ Grammarly/ Coda environment. Elsewhere different activities are taking place.
One example that I am very familiar with: My son studied undergraduate biology and Data Science master at Kennesaw University. As part of that study, he and his team trained an AI to recognise the very dangerous, and very difficult to identify Candida Auris fungus. All you need to use this, is a lab level microscope with a camera, and an email to be able to send the image to the AI. The cost to identify the fungus was a very small fraction of traditional methods, and simultaneously a small fraction of the time. One of the professors was dismissive because it did not require expensive technology…. The results languished for six months while my son was trying to get Kennesaw management to progress funding - both to roll out the C. Auris identifier, as well as to train the AI to recognise more cell types. (They had six at the end, including a control cell.) He is now a full stack developer at a bullion trading company in Denver, lost to medicine, and the research is sitting somewhere, gathering dust.
At my current workplace, I have developed some (I think) really good and productive Coda tools. But the interface is not Excel, is not Word, and people just are not interested. Yes, it is not a company heavily into tech, and maybe my communication skills are not what it could be. But if I cannot move them the small step into Coda docs, how am I going to move them into “agentic AI”? Unless it gets done in a way that they do not even notice it. And I think that that is what is going to happen over the next 10 years and following - AI will become invisible in daily knowledge work. And then it will take off.
Other data points - Excel is most famous for being “the accountant’s tool”, but that did not stop it from being used for statistical analysis, engineering, and plain old lists. IBM’s PC got the Personal Computer accepted in businesses all around the world, except in IBM. For too long IBM let the PC Division be an also ran in the company, and they subsequently lost the leadership in PC’s, but also all computer hardware.
Earlier you mentioned that 80% of Spreadsheet relevant to major business decision have major errors in them - can you imagine the havoc that uncontrolled/ unsupervised AI could cause? We have already seen high profile examples in some court cases, and in American Government healthcare reporting.
How do I bring the weave together?
@Shishir ‘s challenge, over and above the technology, is to manage the Venn diagram you mention. As the integrated solution is being developed and rolled out, it will be important to not ignore the Grammarly, Coda and Superhuman populations that have made the individual products successful. Don’t let the technological progress run away from the customer. Don’t do the opposite of IBM, and focus so much on the new stuff that your existing revenue streams shrivel up.
Do take advantage of your existing enthusiastic Coda community, and address the various pain points that have been raised. And the same for Grammarly and Superhuman. Find and promote the people that identify the C Auris fungus, make room for their contributions to expand the Venn diagram center. And my hobby horse - do also pay attention to the soft issues - do well managed product launches that do not alienate large portions of your existing user base.
In the 80’s strategist Michael Porter divided the business world into high quality, low volume, high margin businesses on the one hand, and low quality, low margin, high volume business on the other hand. In the 21st century successful business string together many core competencies. In this case, it has to be AND Coda AND Grammarly AND Superhuman AND Superhuman Go AND large data bases AND retaining the no-code citizen developer ethos.
But, it’s just a ramble
Rambling Pete
ok thank you, and how does the “team aspect” work.
for example, I work on 2 teams or separate companies - would all the invites be under a single team? would the coda docs all be a single team?
is it possible for me to create 2 different teams under my main sign up account and invite the user separately?
You can only be a part of one organization or team at a time. If a user joins another organizations’ license they will be removed from the previous one.
Could you share your use case for two different teams? I want to share it with our team so they can hear from our customers on why switching between accounts or teams could he helpful.
Here’s what I would like to be able to do:
My main account (gobehere.com)
__ personal email
__ work email team 1
Invite 4 people from domain A under my parent account (team 1)
Invite 3 people from domain B under my parent account (team 2)
No access to information between A and B
Alternatively, and I assume this is the way I should set it up
Create org in team A
Create org in team B
Have each invite my account to their org, or to individual coda forms, etc. if I can’t do org.
Ultimately I’d like to be an admin on 3 different accounts (my account, team A, team B) without having to have a separate account for A & B
Thanks,