The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
I just finished building my own open-source farm bot called Farmbt š±
It's a team of 13 AI agents. I asked them one question, and they came back with a ranked list of opportunities nobody has fully cracked yet.
You describe a problem area in plain words, and the agents take it from there:
š Researchers search the web in parallel and pull out findings, each with its source
ā
A verifier re-opens the pages and checks every claim against the real text
š³ļø A gap hunter runs three independent passes, and only gaps that show up repeatedly survive
š„ A critic tries to prove each gap is already solved
š ļø Solver agents design fixes, check them against existing products and papers, and write an experiment plan
The hard part wasn't getting AI to produce answers. It was making it not make things up. So every claim is graded by evidence, confidence ratings are capped by what the source actually supports, and everything is saved, so a run that hits a rate limit picks up where it stopped.
It runs on free-tier models today and gets better with every model upgrade, with no code changes.
First test: crop-disease detection apps for smallholder farmers. 39 sources, 16 findings and 5 opportunities, ending in a concrete first experiment to run.
Open source: github.com/Samthnalyst1234/Farmbt
If you need an AI pipeline that's reliable enough to trust, for research, automation or anything in between, let's talk š
The step where a verifier re opens every page is a solid guard against hallucinations, and persisting state lets the run survive rate limits without missing a beat
Thanks Johnson, Those two were the hardest parts to get right. The verifier goes after the riskiest sources first (anything the agent only saw in search results), and a claim is only marked unsupported if a second check agrees, so one bad read can't sink a good finding. And...
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Related posts
ā¢ Oct 1
Rebuilding the moment an inbound inquiry turns into a confirmed booking for MEND Appliance Care, a local repair shop in Spalding, Lincolnshire.
The Problem Solved: Replaced 2.5 hours of daily owner phone-tag, vague fault descriptions, wrong van parts, and customer no-shows with an instant AI-powered diagnostic intake, fixed upfront pricing, deposit collection, and auto-generated dispatch scheduling.
Key Features Built in Lovable:
ā¢ Multi-Modal Intake: Customers describe the issue using text, photos, model numbers, or video.
ā¢ Instant AI Diagnosis & Transparent Pricing: Autodetects fault & exact part (e.g., Bosch Serie 6 drum shock absorbers) with clear labor/part cost breakdowns.
ā¢ Flexible Resolution Paths: Home Repair, Workshop Drop-Off, DIY Part Purchase, or local stockist routing (Toolstation/Screwfix).
ā¢ Owner Safety Net & Dispatch Dashboard: Real-time job calendar, "Ask Maya" human-in-the-loop review queue, and 1-click customer comms.
Built entirely with @Lovable for the #lovablechallenge.
Shop owner dashboard : Email: maya@mend-demo.co.uk
Password: MendDemo2026
The "Ask Maya" human-in-the-loop queue is one of those details that separates a clever demo from something you can actually ship to a real business. Most AI intake flows skip this ā they either automate everything and break on edge cases, or hand off to a human and defeat the...
What I built & why
I built PawRoute, a smart booking and automation experience for a fictional solo mobile dog groomer, Maya, based in Austin, Texas.
I chose mobile dog grooming because booking isn't as simple as finding an empty calendar slot. Every new inquiry can require the owner to check the customer's location, understand the dogās needs, recommend a service, calculate pricing, find a time that works with the driving route, confirm the appointment, send reminders, and handle cancellations.
I wanted to answer one question:
How can I turn āCan I book?ā into āYouāre bookedā with as little work from the business owner as possible?
So I designed PawRoute around a simple 3-step customer booking experience. Customers provide their location and dog details, receive a recommended service with clear pricing, and choose from available appointment times.
I also created Best Route Match, which prioritizes appointments when Maya is already working nearby instead of treating every available time equally.
Behind the customer experience, PawRoute handles the repetitive work: booking confirmations, reminders, customer information, route updates, rescheduling, and automation activity.
I also designed an owner dashboard where Maya can see appointments, customers, her driving schedule, and what PawRoute has automated.
Finally, I added a cancellation-recovery workflow. When an appointment opens up, PawRoute can help identify suitable waitlisted customers and refill the slot instead of requiring Maya to manually message people.
The goal wasn't simply to build another booking website. It was to design a system where the customer sees simplicity while the complexity is handled behind the scenes.
3 steps for customers. Less scheduling work for Maya.
Built with Lovable for the Lovable Challenge.
TYKHO Care ā sensory-friendly haircuts for children Ā· Odesa, Ukraine
TYKHO turns a parent's worried "can we come to you?" into the right kind of visit at the right time ā booked, prepared and refilled on cancellation, without a single scheduling message for the stylist.
The client:* Anna "Anya" Kovalynska, a solo children's hairdresser in Odesa. One chair, one child at a time ā a quiet studio in Arcadia plus home visits along Fontanka Road and in Tairova. Many of her clients are children who find salons hard: the buzz of clippers, mirrors, waiting rooms, being touched.
Quirks I designed for:
- quiet hours only on Tue/Wed mornings
- high-support visits need a 15-min buffer
- some children need a 20-min Meet & Greet first
- home visits are grouped by district
- "tough morning" cancellations burn the rare quiet slots
- parents re-explain their child's triggers every time
- Anya can't answer the phone mid-cut
For parents (no account, no app store):
- 5 calm questions ā the gentlest kind of visit, with the reasons
- only the times that fit their child
- instant confirmation and a "what will happen" story to show the child
- reminder with one-tap confirm, move or cancel
- none of the times fit? Join the waitlist and get an offer by email
For Anya (installable owner app):
- her whole day on one screen
- a brief for every child built from the parent's answers
- reminders and a one-tap nudge
- cancellations offered automatically to the best-fit family on the waitlist, held 2 hours, then passed on
- a day off asks the affected families to pick a new time ā nothing is cancelled
- book for a family by phone with the same 5 questions
- a note after each visit that the next visit starts from
- Outbox of every email sent for her Ā· services, prices and hours Ā· QR poster for the studio mirror Ā· light and dark theme
Under the hood: Lovable Cloud ā server-validated slots with no double booking at the database level, row-level security, scheduled jobs for reminders and offer expiry, transactional email via Resend on its own domain, two PWAs.
Removed from Anya's day: scheduling DMs Ā· repeated care questions Ā· manual reminders Ā· chasing cancellations and the waitlist Ā· calling every family on a sick day Ā· preparing each child from memory.
One demo week: 14 bookings confirmed without a message Ā· ~3 h of messages saved (estimate) Ā· 2 cancelled slots refilled automatically Ā· 0 care questions repeated.
Note: TYKHO Care, Anna Kovalynska and the families are fictional; the studio photo is AI-generated.
Links:
Site: https://tykhocare.com
Demo video: ŃŠ°Š¹Š» Ā«01 Demo videoĀ»
Bonus: Ā«02 What got easier for AnyaĀ», Ā«03 How I built itĀ»
For Test:
1. Be a parent (no login). Open https://tykhocare.com on your phone ā "Start the comfort check" ā answer for a made-up child ā pick one of the times ā enter your real email. You'll get the confirmation email with a "Manage visit" link (reschedule / cancel in one tap). None of the times fit? Tap "Join the waitlist".
2 Ā· Be Anya (owner). https://tykhocare.com/owner/login ā email: usertestanya@gmail.com Ā· password: Anya0508. Your booking is already in her day with a brief built from your answers. The envelope icon (Outbox) shows every email the system sent. "ā¦" ā Services & hours: prices, hours, a day off. "ā¦" ā Studio QR poster.
3 Ā· Try the waitlist. Cancel your visit from the email ("Manage visit") ā if the freed time fits a family on the waitlist, Anya's Outbox shows the "A quiet time just opened" offer that went to them ā no call from Anya needed.
Tip: both sides install as separate apps (TYKHO / TYKHO Owner) ā Chrome ā® ā Install app, or Safari ā Share ā Add to Home Screen. Demo data resets every night. While signed in as Anya, you can also restore it yourself: "ā¦" ā Reset demo data.
Great job šš„ā¤ļø