If there’s one thing I love about Coda, it makes me look good!
On Friday I discovered an urgent need to assess proposed automation processes - automate or continue to perform manual steps?
Is it possible to build an app in Coda that tells us if it’s wise to automate a process?
So I did what any Maker would do - I built the app in about 4 hours using the new Pack Studio and a few simple formulas to make the results pretty.
I was most concerned about tracking the break-even point and needed a dynamic chart showing relative outcomes over the estimated life of proposed automation.
The next chart shows a case where there are significant development investments (dives deeper into the orange early in the lifespan), the likes of which this my firm will often encounter with this tool.
There’s nothing ground-breaking about this app; anyone with a spreadsheet can do this, but we wanted a solution that could be embedded in a proposal and integrated as part of a document process. We also wanted mobile agility, the need to create a polished proposal while sitting in an airline seat. Yet again, Coda nails the landing.
The spreadsheet app that this approach displaces is fraught with issues, difficult to run without a laptop, and prone to mistakes along with copy/paste, etc. Huge time-saver and best of all, it can compute the saved time.
I would love to move this into the marketplace, but I have no time to focus; happy to entertain a partnership if there’s a developer who can run with this idea. I have lots of automation guidelines, good Pack code, and en entire manuscript draft on the topic inspired by Elon Musk. This doc could be very valuable to a lot of people.
@Bill_French also contributes a lot on Airtable script! But this Coda doc he created reminds me the potential on what Coda can achieve compared to other tools.
So cool! Did you run your Pack building process through the doc and should it be automated? Knowing what I know about building Packs, perhaps best not to.
In this solution, I surmised that no one likes to create simulation data. In dynamic modeling systems the time-series data is typically generated based on the rules and the inputs. As such, I decided that a pack could play two roles in this analysis:
To generate the outcome data points needed such as total opportunity costs and savings.
The time-series table that shows the progression from initial automation investment to benefits over the life expectancy of the automated process.
By using a simple schema, I was able to simplify the computations (in javascript) while exporting the time-series data. This grid serves as the basis for all visuals and outcome data points.
I’m not sure I fully understand the question, but I did build the pack in the Coda editor; it was really quite effortless and the javascript editor performed well. I prefer developing inVSCode because I’m lazy and CoPilot (GPT-3 AI Pair Programmer) is always writing code for me based on simple code comments.
Should it be automated? I don’t know - enlighten me.
Thanks! Cool, yes. Amazing? Um, it’s actually a testament to how amazing Coda is. I wish I could take credit for the amazing part, but what makes this amazing is some amazing machinery that lay just beneath the veneer of a Coda document.
Let’s start with the UI. The ability to build a simple collection of inputs that work well (cross-platform) on mobile and desktop devices. That’s amazing.
How about those inline formulas and other data-centric controls which transform a document into a clear and uncompromising assessment dashboard experience. That’s amazing.
And custom Packs that are designed and built integral with the Coda development process. That’s truly amazing.
How about the simple schema object that magically transform the output of a Pack into a complex table object. Yep - this is pretty amazing.
Lastly, there was a team of Codans who aligned all of these amazing things and actually thought carefully about how one might chain features to produce solutions that address essential business requirements. Amazing.
While agreeing on basically everything you said (I also love Coda, literally), the artwork is made by the painter; even if someone else has produced colors, canvas and brushes.
Oh my, @Bill_French, once again you nailed it. LOVE this approach you took!
I think the way you used packs for the data modelling is great, but I think oftentimes the developer would have left it at that. The result for the client would have been “Hmm, seems difficult, so I’m sure it’s valuable, I personally just don’t know what to do with it”.
By translating the results into a visually intuitive result, you solved that problem for the client, and thereby released the real value.
“We want to democratize this process with a truly horizontal product that every knowledge worker can use…"
What they’re describing, of course, already exists. While Coda is not specifically suited as a generalized replacement for Excel, it can be configured as a proxy for many complex business model challenges.
Causal says its formulas “read like plain English” such “as Profit = Revenue – Costs” and also claims it typically takes “100x fewer formulas” in Causal to build exactly the same model in Excel.
We do this all the time in Coda, so this idea is far from novel. Abstraction of terms and encapsulation of business logic by reference is also possible in Excel itself which suggests to me that they aren’t likely to build a disruptive solution. Disruption often comes by changing the game entirely; attacking the 8000 lb gorilla on the hill is a bad idea.
Coda was shaped to climb Mt Improbable by building a scaffolding that avoids the dominant aspects of the number-crunching, word processing, and data visualization segments. Another kinder, gentler spreadsheet is not what comes after the spreadsheet - it’s something entirely not like a spreadsheet.
This is why I have been so excited about the democratization of systems modeling, data science, and decision support tools. Business problems are not addressed by formulas regardless of how easy they may be to create. Rather, they are solved by dynamic models.
And as this thread makes clear in my comments, Coda is not prepared to address many aspects of complex modeling, but its architectural DNA makes it a candidate for doing so at some point in the not-to-distant future.
