Willowfinch
Love where you live.
One of the beauties of being self-funded is that I don't have anyone I feel obliged to pay back. One of the worst drawbacks of being self-funded is that you can't do everything yourself, so everything takes more time to execute because you're working under budgetary constraints.
I succeeded in getting connections between replit and my algorithm database yesterday. Authentication succeeded on the replit side and I was able to populate the algo database with tokens!!!
I've also started "coming out" about how Willowfinch was wrecked last fall. I'm still only really writing about recovery here. The past three years have been so frustrating because of dev. Dev and feature deployments when you're bootstrapping & paying for dev is just draining. Worse, Willowfinch is so complex that most devs have not been able to get their heads around it or the code base. FFS, I even paid one "30-year veteran of tech" to learn the architecture so they could guide the redeployment & fix some bugs. She couldn't do it. I literally went through 10 devs on this product and only two really "got it" got it. J., who then handed it off to F and if F was busy with other things, I was dead in the water.
That's where the advancements in AI just this past year in code support and app building have become demonstrably significant. They're no longer just toys to make pretty pictures of people with too many fingers and marketing copy. I'd try an AI engine to help with SQL Server on Azure and it would give me horrible advice. By late summer, I was able to build tables referencing remote databases. I was still having problems using python's flask - which was irritating as all get out, but was my "ultimate test."
That was because last summer one of my Code Sister friends who's also in management suggested I just drop the WF I've developed and start fresh. That about made me cry... okay, if I had any tears left I would have cried, but I don't, so I didn't. Starting over from scratch was daunting because just getting a developer to understand what was happening consumed weeks of meetings and hours of explanations and work. I knew it would come down to me.
I've started rebuilding using replit (hit me up for a referral link, I could use the credits). It occurred to me this morning that this ability to harness AI to build the code I couldn't have built in the past without a lot of pain and suffering or expense is huge for me and for Willowfinch. It just might keep the idea alive.
While AI can gather some of the data, it can't gather all of it and we already know it's notoriously bad at math. In any event, rebuilding Willowfinch from the ground up without using a dev studio or having to hunker down and take a bootcamp seems possible this morning.
And I'm feeling like the "bad thing" which happened last fall might not have been such a bad thing after all... with the AI improvements, I might just be able to streamline a lot of my data acquisition processes. I'm also building the recommender I never would have been able to fully integrate before without substantial cost.
The optimism I'm feeling is literally highlighted by the chirping of birds / birdsong coming from my back yard on a sunny, blue-sky spring day here in Seattle. The wreckage of winter just might be behind me now. It could be that the Willowfinch which needed to be built, couldn't have been built until this time because - again - the technology didn't exist. The tech is still nascent, but I'm cobbling it together.
Well, it's been two weeks and my SIXTH developer still hasn't sent me a proposal on migrating from Okta to a 0Auth account. The latest AI iterations have nicely laid out plans, so am gonna start to try to at least fix my PROD deployment.
I swear, "back in the day" you had to buy 8-10 books of 1,000 pages apiece to learn a new programming language because there were ALWAYS syntactical errors, so you needed to compare version. Plus, the changes could be monumental between versions, so just to keep up you needed another book or two. Not to mention that corporate would only allow certain languages or programs to be used or you could be fired.
The internet changed all that. I remember being able to access the Oracle 8.0 documentation online and whoo-pee!! I didn't need another book on my bookshelf. I had an engineering co-worker who actually crushed his bookshelves with overload so that's how really "heavy" (get it? 'heavy' instead of 'big'?) a deal this kind of thing is.
Still, learning a new software language is tedious and time-consuming. I went from 0 - Perl program developed (which lasted 10+ years on coworkers machines!) in five weeks, but those five weeks were concentrated work. In one big respect, I just don't have that in me anymore, I'm looking to hire. BUT... what I've found is that because I'm so used to working with data, functions, algorithms, whatever... the complexity of what I expect to be built and what people are used to building, is - well, lacking. This is a primary reason why development of my app in the first place took 3x as long for the first iteration.
In any event, here I go. And if anyone has a build platform they like, let me know. I'm hoping to keep my current codebase at the moment and just fix the authentication problem and learn a bit as I go. There's one monster algorithm I REALLY don't want to replace because, well, the logic is just nasty ugly. I think I calculated at one time something like 1.X million use cases. Yeah, that kind of ugly uck. It exists right now and it works. Replacing it has defeated multiple developers. No joke.
Time flies when you're self-funding and you can't build fast enough. I suppose Willowfinch should be considered a "failure", but I haven't stopped working on it - at least not enough to bury it. And, I still use it for myself. To. This. Day. I use the tool to help me solve my budgeting problems and explore the world around me.
I have three years of stories I haven't told, but I'm back here now. I think this is the best platform to talk about the Perils of Building. There are bright spots, I promise. But I write this a day after I severed yet another YC Co-founder match relationship, which again, lasted a year. When my product is dead in the water for automated access because my freemium authentication software blocked admin access. When my DEV is so far out of whack from my PROD that I don't really know what's deployed. When I've gone through (count 'em) SIX front-end developers and only one who knows how to make things work & he's disappeared. When the overwhelming pain of arthritis made it impossible to sit at the screens for more than 30-minutes a day. When said arthritis in my right arm made typing impossible & I went for surgery (am now recover(ing/ed)). When I just found out my FA made an error in the withdraw rate of our investments & - welp - you guessed it, our financial plan collapsed.
Still, the bright spots pull me forward. Whenever I open up WF, users hop on fast & furious. I learn fast. People LOVE to talk about the problem of housing and how much they can afford. They're confused by budgeting. Personal finance calculators have very high hit rates on google search, so I've learned something about my strategy. AI has gotten sufficiently competent in programming that I don't have to learn yet another language, I can hack at building / rebuilding my prototype. I've made some good friends and met some amazingly smart and talented folk. My SO still supports and believes in me.
My brain is still scrambled from the past few months, since I got locked out just as I was about to deploy my V1 Recommender. I'm resetting my goals right now after four months of healing from the shoulder surgery and the shock of the lock-out. I'll be back here to journal about the challenges of the past three years because that's what I plan on writing here - the agony & ecstasy of working on Willowfinch. It ain't pretty, but it's my experience. And while I might be a "Loser", it's just because our story isn't at its end yet.
As a "Wealthies" finalist, @WillowfinchHQ shares the category with 3 other companies. All of which have significant funding, the largest over $100M. One is on the NASDAQ.
Our tech helps #advisors help their clients make more informed decisions about housing. Our pay wall goes up in July, so sign up now for the free beta at www.willowfinch.com
Received notice that Willowfinch has been named as a finalist for a "Wealthie" award from WealthManagement.com. Our proprietary algorithm stands up to other, more seasoned finalists in the field. It's a huge deal for a pre-launch startup.
https://www.wealthmanagement.com/industry/wealthmanagementcom-announces-2022-wealthies-finalists
Met with my personal financial advisor today. The market's gone down, so the money I was able to access last year for Willowfinch no longer exists. I've been funding out of my personal retirement savings, frankly, because my husband and I live comparatively frugal lives. He's a coupon clipper, Costco shopper and I've worn out my t-shirts until he inspects them for holes & disallows me from wearing those outside the house. We have a mortgage, but at a fraction of what one pays for an apartment in Seattle these days and we love the house.
Still, part of the Willowfinch story, in fact the Willowfinch journey, is about selling our home for the money. However, the story really has changed since 2018. Both of us have crossed the bridge of 59 1/2, so hubster didn't have an early retirement. Still, we flat out asked Adam if we needed to sell the house to live with basic needs met until 92 (whatever).
Even with the down market, we don't have to sell our home and I can keep funding Willowfinch at this low level a bit longer.
Willowfinch announced a new free beta for financial advisors. It is a recommender system which allows them to build a heatmap of properties for their clients without the client doing all the "legwork" of property searching. It covers three scenarios and requires ZERO personal information to launch.
Offer ends July 1st, 2022
This year's data refresh of the ACA health insurance found even more states breaking from the national & making their own health insurance marketplaces. We also created an additional sub-layer of modeling down to the county/ZIP level. This increased data complexity resulted in the creation of a dataset of over 54 MILLION records. We reduced this to just under 2.5M records for only the 2nd Cheapest Silver Plan in the ACA.
Our future goal is to eventually have a dataset which covers the Platinum, Gold, Bronze, and Catastrophic levels as well. That could well tally above 216,765,648 records depending upon the number of carriers and how they increase or decrease County/ZIP relationships. heh
Comment