I'm not a chatbot developer, and I'm not trying to become a "chatbot builder" as a career.
But building and shipping 3 AI chatbots so far has made me a better consultant.
My specialty is productivity and digitization 💪
Still, I spent serious time learning how chatbots are actually built and operated.
Not because the process is fun and feeds some fantasy of an alternative career path, but because I want to widen my clients' options and sharpen their decisions about digital solutions.
The Three Chatbots
📏 The first chatbot was personal (not for a client); it helps companies assess their readiness to adopt AI technologies.
⚖️ The second was an AI legal advisor powered by thousands of pre-processed legal articles, designed to answer with precision and without hallucination. Accuracy so far exceeds 95%, validated by specialist lawyers.
📋 The third is an accountability officer for a charity organization. It reminds team members to submit daily reports, then refines and merges them to produce automated weekly and monthly reports and update tracking dashboards. It even chases late submitters via private Telegram messages to ensure the latest accomplishments and challenges are recorded in a central database. (We basically automated the manager's nagging 🎉)
What I Learned
- The non-smart parts account for 80% of what makes an AI system successful and efficient
- There are cheap smart solutions and even cheaper ones (building AI systems is not expensive at all when you have the right expertise)
- Co-creation is great, but a client who doesn't know what's possible with today's technology isn't an effective partner
- A consultant's duty is to educate the client about the full range of technical options; engineering, financial, and operational; before jumping to the newest or fanciest solution (the priority is the client's interest, not the consultant's portfolio)
If you're considering a digital investment for your team (AI or otherwise), reach out and I'll help you choose the right technology for your requirements and budget.