I REALLY hope that Coda management is taking note and appreciates Community Members like @Max_OBrien , @Bill_French , @Christiaan_Huizer , @joost_mineur. And several others, some of which I havenât seen in the community for a while? Their passion and knowledge is a major asset and should be appreciated and utilized. They leave us mere mortal Coda makers in awe.
Over the years there have been upwellings of frustration from the community with the level of communication. Coda eventually responds, only for things to slip back. With each upwelling there are fewer and fewer people joining. The Coda community on Reddit is dead for all intents and purposes, this community has a fraction of the engagement it used to have. What is happening on YouTube, I do not really know, but a quick look at Coda HQ shows nothing new in literally years.
Newcomers look at the lack of engagement and deduce that Coda is dead or dying.
I hope that this time things will be different and that SuperHuman will provide engagement with the different customer groups in its admittedly very diverse customer universe. **Donât let the pursuit of perfection lead to the death of great.
**
But, itâs just a ramble,
Rambling Pete
Quick heads up: Iâm onsite with the team this week, so responses might be slower than usual. Back to normal next week.
Piet â not a ramble at all. This is spot on.
Youâre absolutely right to call out the depth and generosity that folks like @Max_OBrien , @Bill_French, @Christiaan_Huizer, @joost_mineur, and others bring here every day. Communities like this donât work without people who consistently show up and help others level up. Yâall are the backbone of this community, and I hope we can bring more folks into the fold as well.
I hear you on the recent patterns. Iâm here to be more visible during the everyday moments, not just the big ones. A steady presence matters, and as I get started, my focus is on continuity and consistency.
On Reddit and other platforms: not my direct responsibility at the moment, but definitely on my radar. Iâm meeting with those teams soon to figure out how we can show up better in more places.
Part of my job is making this more consistent, not just when things blow up, but as a regular thing. Less spiking, more steady. I appreciate you flagging this!
Iâll let the others reply directly to your comment, but I can say that itâs nice to see some engagement from the Codans. Hoping things will continue and ramp up in the coming weeks / months.
Find & Replace is for Documents. Agents are for Data.
In a traditional word processor, Find & Replace is a simple pattern-matching utility. But Coda isnât just a document; itâs a relational canvas. The very essence of find and replace is a big ball of dealing with uncertainty in a document that spans all sorts of nuanced representations. Whatever Coda might have provided, you wouldnât like it.
When a user asks for âfind and replace,â they are asking for the ability to evolve their content. With Coda MCP, it doesnât just match stringsâit orchestrate intent.
Here is what Agentic Find & Replace looks like when we move beyond RegEx:
1. The Semantic Upgrade
The Prompt: âFind all mentions of âlast yearâs pricingâ and replace them with the actual price from our Current Catalog table.â
Beyond RegEx: RegEx can match characters, but it canât match intent. An Agent understands context and performs live data lookups automatically.
2. The Structural Refactor
The Prompt: âFind any unstructured meeting notes and replace them with a bulleted list inside a Callout block, bolding any names mentioned.â
Beyond RegEx: RegEx can match patterns, but it cannot intelligently identify a ânoteâ or interact with Codaâs rich-text block types (like Callouts) on the fly.
3. The Everyday âSuperpowersâ (Small wins, big impact)
Standard find-and-replace often fails on the little things that are trivially easy for an Agent:
Date Normalization: âFind all dates like â02/04â or âFeb 4â and replace them with the standardized âFebruary 4th, 2026â format.â (RegEx handles this with pain; Agents handle it with a sentence).
Reference Linking: âFind any email addresses in the doc and replace them with @Mention links to the corresponding person in my Team table.â
Contextual Acronyms: âFind âMVPâ and replace it with âMinimum Viable Productâ, but only the first time itâs used in each section.â (Nearly impossible for pure RegEx).
4. The Smart Upsert (Page to Table)
The Prompt: âFind all paragraphs that describe customer feedback, and âreplaceâ them by moving them into the Customer Feedback table, leaving a summary link behind in the canvas.â
Beyond RegEx: This is a cross-surface data operation. RegEx stops at the character; Agents work at the structural level.
The âSimple & Elegantâ Bottom Line
In the legacy world, you use a tool. In the Coda world, you use a Collaborator.
Donât search for text. Instruct your Agent to evolve your doc.
The Visionary Prompt:
âHey Agent, find all mentions of âVersion 1.0â and update them to â@Version 2.0â references, but only if they appear in the Status section.â
That is the power of Coda + MCP.
@Bharat_Batra1 When are you going to open up the MCP beta!
@AJM â just a friendly tip for next time: starting a fresh thread and tagging it as Ask The Community can help your question get more visibility. Sometimes great questions can get buried in longer conversations, and a dedicated thread makes it easier for folks to jump in and help!
I wasnât asking for something new. I was responding to your comment about the importance of coda as a writing space.
@Bill_French respectfully, not everything needs to be ai or agentic. When I ask for find and replace Iâm really just expressing a need for find and replace while using coda as a writing doc.
Indeed. You were reminding everyone of something so essential that has always been missing. And itâs a rational complaint. I canât explain why they never added this. However, I have a hunch itâs because an all-in-one document is a technically nuanced beast. Any attempt to appease, would make more than 50% of users begging for more and the remaining underwhelmed. Thatâs my hypothesis. Iâll let the Codans defend their choices.
I agree. It doesnât. But when you have zero options, no prospect of feature relief on the horizon, and you need to increase your productivity, perhaps such a seemingly radical approach might benefit your work. I offer these alternatives because I thought diversity of workarounds mattered in this community.
Itâs not that ai is not my jam! But building a whole agent to do find and replace will take me much longer than it will take to ctrl f and scroll!
In any case, Iâve been trying to transition away from coda because it isnât working for me and it requires too much setup knowledge and time that I cannot do on my own.
I feel it is unfair to divide the world into âpro AIâ and âanti AIâ in this way.
I understand that to the guy with the big hammer, every problem looks like a nail.
And we (the pro-AI antagonists) are promolgating a message here against significant opposition (or at least indifference), butâŠ
Using an AI Agent to burn down a lot of trees (or coal or uranium or whatever) to melt a ton of tokens in yet another water-sucking, CO2-pumping, power-guzzling, data center - just to do some simple text processing - well, it just feels kinda wrong.
But then I must recall that when MS Windows first started showing little animations of pages flying from one folder to another during a file copy operation, I was also âoffendedâ by the sheer waste of âpreciousâ CPU resources.
I was also shocked the first time a fellow engineer sent me a webcam image holding up a page with âLUNCH?â written on it - all those megapixels to send a 6-character message? Argh!
And then we set up our own âcoffee camâ so we could remotely check the coffee perculator to ensure it was full before going to the coffee corner - to avoid being the shmuck who had to refill the darn thing. The waste!
But still - using AI Agents (especially the screen-scraper / mouse-clicker kind) to do something so basic as search & replace - feels like âthrow another whale on the barbie and grab a beerâ kinda attitude.
Itâs official, so - Iâve become a grumpy old man!
I understand this reaction. It is a common and rational way to think about tooling choices. You know these features are far more powerful, but you have a deep focus on getting stuff done using muscle memory, perhaps crafted over decades.
Let me clarify - I built no agent. Google and Coda engineers performed the heavy lifting. I simply connected the Coda MCP service to Antigravity, then started doing things like finding and replacing with great precision, using features Iâve never had before.
I also didnât build agents to do the dozens of different use cases Iâve written about. I DID have to learn about using these technologies. Nothing new comes at zero cost.
You had me at âCoffee Camâ. This is a great mental reset.
I agree. It does feel wrong. This is especially true if you frame the underlying technology in a constrained and negative way. How long do you think it will feel wrong? For me, it was an AI minute. Iâve refactored lengthy multi-page Coda documents with several complex processes created by agents that plan, review, execute, test/validate, and document every change. As I said, it only took me an AI minute to realize itâs not wrong.
We tend to avoid these seemingly abhorrent things until they become more energy or task-efficient. But such calculations of effort and energy are often wrong, too. Today, almost a trillion images will be generated and moved about the world. These are extremely inefficient vehicles for conveying data. Yet, financially, they are justified. Humans, it seems, quickly get comfortable with waste, or at least the assumption that we are wasteful. Often, we are not wasteful. We are productive.
As to an agentic find-and-replace, you are measuring the investment and cost required to perform one simple aspect of the adjacent possibilities given this very broad agentic ability to sculpt text with precision and intelligence that can be human-driven, programatically processed, or agentically executed by other agents. Feel free to assume users only want the essence of yesterdayâs find-and-replace, while I assume future requirements are significantly more complex.
If Coda were to provide a find/replace feature, the next complaint would be that it is not represented in CFL. Than another, and another, and another. The heat loss would grow. The complexities and possible exceptions to rule after ruleâall expanding to fill several conversations here in the community.
Given what I know, I canât support a demand for a feature designed in the previous century. I think Iâll lean into a future where features are a function of thought.
Totally understand. I donât have the time either. I still love almost every aspect of Coda, especially Coda with agents because I donât have the time or knowledge about every aspect of the tool. My agents know this and transform my requirements into manifestations of Coda beauty, leaving me to buff out the wrinkles. This is the time I can afford.
If they donât ship the MCP service soon, Iâll probably be right behind you. But will I move to a competitor? Nope. I will simply move everything that Coda does for my team and me into an agentic platform where the spirit of Coda will live on.
I remember the flying paper between folders when copying in Windows. My dad commented that if they moved the folders closer together, it would copy faster. It made me laugh. I told him they did that to entertain you while you were waiting for files to copy. If the flying paper wasnât there at all, though, it just might have copied a little faster.
I hated Clippy; he was so annoying. I saw a video once that compared two office staff membersâ attempting to complete a report. One was using MS Word, and the other was using WordPerfect. The one using WordPerfect finished her report with no problems, the one using MS Word kept getting bothered by Clippy and never finished her report because she was driven mad by Clippy. It made me laugh. I tried to find the video without any luck.
