Small business

Tech Solutions: digital services for small businesses

A studio I founded that builds rule-based WhatsApp chatbots, NFC business cards with landing pages, and full websites for small businesses.

Pilot ✓ Oct 2026

Cloudflare Workers
WhatsApp Cloud API
Astro
Playwright + axe

Try the booking bot

The studio builds WhatsApp chatbots that follow rules, with no AI. This one runs in your browser for a made-up business, with the same mechanics as the real ones. Tap the buttons or type, for example “book”, “hours”, “person” or “stop”.

Studio Demo

made-up business

Hi! I’m the booking assistant of Studio Demo. What do you want to do?

What is behind it

It follows rules: a menu, buttons and keywords, even inside a sentence. It has no AI, so what it does not understand gets the options again.
The real bots run on the WhatsApp Business Cloud API and Cloudflare Workers, with the bookings in a Postgres database for each client.
Every message is checked before it is processed: the webhook signature, duplicate deliveries, a limit of 20 messages per number per minute, and an opt-out word (STOP).
Bookings have a capacity per branch, date and time slot, and are confirmed with one tap. The real one is in beta with a beauty salon.

The business in this chat is made up, and the engine is a simplified version that runs in your browser. The real bot lives on WhatsApp, and a client’s conversations, texts and brand are not shown here.

What I built

Founder and product engineer, since September 2026.

Stages reached

Pilot ✓ Oct 2026

Result. Chatbot beta for L'Boutique Estética: bookings by branch, date and time slot, with capacity limits and one-tap confirmation. Pilot website for Embajadora de Boyacá with Lighthouse scores of 99 to 100.

Inside the product

Project showreel (26 s).

Business plan in seven phases

I wrote this analysis before building the studio. It decides what to sell, to whom, and what not to build. Status says how much each phase is worth today.

Business plan in seven phases
Phase	What the analysis says	Status
1. Starting point	The goal is a main income and quick sales, not a product startup. One NFC product does not pay a salary: at 80,000 COP a month per client it takes 60 to 75 recurring clients to reach 5 to 6 million COP. So the business is a packaged, repeatable service. Paipa gives an edge (tourism, fairs, three hours from Bogotá) and a ceiling: hundreds of businesses, not thousands, so Duitama, Sogamoso and Tunja come next.	Inference
2. Critical review of the offer	NFC is an entry hook, not the product: the hardware is a commodity and Google already gives a free review link and QR code. Two features I wanted, review coupons and filtering who is asked to review, break Google's rules, so I dropped them. Websites become a standard component, not custom work, so the studio does not turn into an agency.	Policies checked in secondary sources
3. Niches and first client	Eight niches scored from 1 to 5 on six criteria: pain, ability to pay, ease of contact, recurrence, cross-sell and the first small sale. Small lodging led with 25 out of 30; restaurants, barbershops and B2B shops had 23. The scores are my inference, to be tested in phase 5.	Inference
4. Portfolio and model	I ranked twelve candidate services: three to sell now, two folded into other plans, three postponed and two discarded. A booking engine was ruled out because hostels already have software for it, so the edge is installing and maintaining what exists. I also corrected my own break-even: after payment fees, selling direct saves about 12 points against the 15.5% a booking platform charges, not the full 15.5.	Verified in secondary sources
5. Validation in 28 days	Seven hypotheses ranked by how fast they can kill the business. The only signal that counts is a paid deposit of 50%. Decision at day 28: continue with 4 or more deposits from about 15 proposals, pivot with 2 or 3, abandon with 0 or 1. Pivot rules were written in advance. The conversation script avoids hypothetical questions and asks about the last lost sale.	Defined, outcome not published
6. MVP and growth	The MVP is a repeatable delivery process, not software: seven pieces in about 30 to 40 hours, from a digital card template and a tap counter to a monthly report. A client login panel, a booking engine, an app, a CRM, a free-form AI chatbot and a multi-client SaaS stay out until paying clients ask for them.	Plan
7. Final deliverable	A roadmap from two weeks to twelve months, the five cheapest experiments and the five biggest risks: a monthly fee rejected, becoming an agency, dependence on Google, Meta and Airbnb rules, a small market, and data and invoicing rules. The rule of thumb: with fewer than 4 deposits at day 28, build nothing more and pivot.	Plan

From my seven-phase business analysis for Tech Solutions. The price ranges in it are hypotheses and are not published here. Market facts come from secondary sources and were to be confirmed in the official policies.

Analysis and product skills shown on this page

Analysis and product skills shown on this page
Skill	Where it shows
Ranking options with explicit criteria	Eight niches scored on six criteria and twelve services sorted into sell, include, postpone and discard.
Checking the rules before designing	Google's review policy removed two planned features before any code was written.
Correcting my own estimate	The break-even was recalculated after payment fees, from 15.5 to about 12 points of saving.
Cutting scope	A written list of what not to build, and a delivery process instead of software.
Experiments before opinions	A paid deposit is the only signal that counts, with thresholds and pivot rules set before the 28 days.
Turning the plan into products	The two client pilots on this page, a booking chatbot and a website, are the first products built from the offer.