I started BotLogix in 2018 because small businesses were losing customers to voicemail and slow response times. So I built bots that answered. Then voice agents. Then full-stack products.
In the last 8 years, I have built 6 products. Two are live, one is an MVP, and three are still in development. This article is what I have learned.
BoxBuddy: my first commercial App Store app
BoxBuddy was BotLogix's first commercial App Store app. It is an iPhone and iPad app for managing a move. You photograph what goes into a box, put a QR code on it, and search for anything by keyword later, even halfway through unpacking. Voice dictation and offline capture are there because moving-day Wi-Fi is usually spotty.
It was not built for a client. It was built for the App Store, for anyone who is moving.
What I learned: Solve one specific problem really well. Do not try to be everything. BoxBuddy does one job: it keeps track of every box and what is inside it. It does boxes, and it does them better than a marker and a roll of tape.
AEC Benefits: the platform behind a construction benefits broker
AEC Benefits is an Ontario group benefits broker that works with construction businesses. The product is the platform behind it: 24/7 lead capture and outreach, a CRM that runs on a schedule, 50-plus articles written for the searches construction owners actually make, and email follow-up that stays CASL-compliant on its own.
Here is what it actually does. A construction business owner searches for help with group benefits and lands on an article written for that search. They fill out a form or take the benefits quiz. The system captures their details, sends a welcome email, and starts a follow-up sequence that only sends during business hours. The broker steps in for the sales conversation. Everything before that runs by itself.
Lesson: Depth beats breadth. This platform is not built for every broker. It is built around one broker's funnel, from the first search to the sales call. If you deeply understand one business problem, you can build something people will use every day. If you try to solve problems for everyone, you solve problems for no one.
The other four: one MVP, three in development
BenefitShield, AdvisorShield, TalkAbout, and Huddle are not finished. Huddle is an MVP. The other three are in development. I would rather say that plainly than dress them up.
BenefitShield V3
BenefitShield reads insurance documents for benefits brokers and HR platforms: booklets, rate sheets, census files, old quotes. It turns them into clean, structured data. The part I care about most happens before any AI runs. Rules sort the documents first, so a billing statement gets set aside and a benefits booklet goes to a two-pass extraction that scores its own confidence on every value. Not every document needs AI. Sort first, then send the hard stuff to the model.
AdvisorShield
AdvisorShield is case management for Canadian financial advisors who handle insurance, investments, and group benefits. What exists today is the foundation: sign-in, sessions, protected routes, the database, Docker deployment, and a design system. Case management itself, document uploads, audit logging, and billing come next. Foundation first is on purpose. It is not about shipping fast. It is about shipping right.
TalkAbout
TalkAbout is a local-first desktop app. You talk, it transcribes, and it turns what you said into a clean email, a summary, a list of action items, or a social post. The speech-to-text and the language model both run on your own computer. No cloud round-trip. A lot of the most useful thinking in a business happens out loud, and some of it should never leave the room. Privacy has to be a design decision you make at the start, not a setting you add later.
Huddle
Huddle is an invite-only hub for BNI-style networking groups. It pulls members' Instagram, YouTube, and LinkedIn posts into one feed, with a member directory, a support queue, digest emails, and participation scoring. Social platform APIs change constantly, so Huddle has fallbacks: RSS feeds where they exist, and any public post link when they do not. The first phase was built so a group has something useful right away, while longer platform reviews like Meta's App Review run in parallel.
Huddle is also the clearest example of the lesson that runs through all six. It makes engagement visible. It cannot make members show up for each other. The tool is not the system.
Lesson: Software does not work without organizational change. You can build the perfect system. If the team will not use it, it is worthless. The best projects are ones where people are committed to changing how they work, not just buying something new.
What I got wrong
I thought better software solved problems
Launch day is the start of the work
I thought features were the differentiator
What I got right
Each one aims at one specific problem
I shipped something useful before everything was done
Charging is the real test
What I am doing now
BotLogix has evolved. I am not just building products anymore. I am helping other Ontario businesses adopt AI and automation. The core lesson is the same: solve specific problems for specific customers who are willing to change.
The six products were the workshop. Now I am applying those lessons to help other businesses do the same thing: identify their specific problem, solve it quickly, measure the impact, and expand from there.
FAQ
Do you regret any of the products?
Why AI and automation now?
What is next for BotLogix?
If you are thinking about solving a problem with software or automation, the playbook is: understand the customer's specific situation, solve that specific problem, ship fast, measure the impact, and iterate. If you have a problem like that, let us talk.
