I'm secretly building an AI agent to run my fantasy football team against my family.
Nobody in the league knows.
I have a mechanical engineering degree, not computer science. So I'm building it by arguing with Claude until the code works.
Let's see if a bot can beat my relatives
Pigskin update before bed. It texts me, but I still make the moves myself.
When asked, it flagged McConkey as questionable and said to hold until morning.
Tomorrow I'm watching for the new gameday alerts: morning digest and lock notices before kickoff, especially McConkey.
Making progress in some places. Not as much as I hoped in others.
Week 1 AI scoreboard is live.
Two points only:
1) Did it make life easier
2) Did the team win
Grok Bot: 2
OpenClaw: 0
Full breakdown on IG + TikTok.
Week 1: I lost to my brother-in-law.
Wife's team finished with 2nd-highest points.
Her bot managed and reported. Mine went quiet because the data looked fine. Silence isn't what I wanted.
Pigskin still makes me open ESPN.
TushPush is making her week easier.
More work is needed
Posted the first DeYoBuilds video last night to kick off Week 1.
Late nights are turning into a habit.
Family still doesn't know about my secret AI agent.
We'll see how Pigskin does.
I wondered if Grokbot could stand up a fantasy agent faster than my 3 weeks in OpenClaw.
Last night on my wife's team: under an hour to live ESPN lineup write and email notify.
WOW! 🤯
TushPush is live. Pigskin is still not ready. Kickoff tomorrow.
Back at it tonight.
Read-only scripts to pull data: done. Verified against live ESPN.
Real time messaging alerts: not started.
If a starter gets downgraded Thursday afternoon, Pigskin won't tell me yet.
Good data. No delivery.
I've got until Thursday for week 1. Going to need all of it!
2.5 weeks working on the fantasy football agent in OpenClaw in the evenings. Based on all the videos I'm seeing online, sounds like Grokbot could be set up in 30 minutes to run my team instead.
I'll need to put that to the test this season. Maybe I'll run that on my wife's team.
Not as bad as my initial reaction, but not as good as I'd hope for from a bot and an algorithm built by AI.
Hopefully my live GM agent, Pigskin, can right the ship by staying on top of lineups, injuries, waivers, and trades. Time to get him ready.
Ran the numbers on last night's draft. The valuations of my picks were mostly right. However, the order I picked in, and when I stopped trusting it, was not.
Overall? Looks like I'm in the middle of the pack and have a lot of work to do to win the league.
By round 12 I had quit using the tool I built for gut calls on names I didn't know. Final few rounds felt like slop, but still turned out ok on valuation.
Final grade compared to ESPN and my custom rankings: B-, 4th of 9 when all team's picks are analyzed the same way.
@jezamancenido Agree that this is a great application! I'm working on setting up mine right now. Wasn't quite ready for the draft (last night), but trying to get it to monitor injuries, waivers, and lineups for me this season.
What system would you have used if you did build it?
Two picks I do like (I think): McCaffrey at #8 in round 1, Hurts in round 7.
Everything else is the starting point for this season-long experiment now.
By round 12 I was no longer taking the live bot recommendations and had defaulted back to the overall ranking algorithm printout I had made. CTRL+F couldn't move fast enough in 90 seconds so made some split second decisions. Not exactly what I had in mind when I started this...