Why we created Chatbot Coach
Originally, the project was meant for business use. Passionate about coaching, we realized how underused it was in business setting, mostly because of a lack of accessibility: traditional coaching is very expensive and companies would therefore only invest in it for their executives, but other employees would rarely have access to it. After this initial step of testing our bot out with companies, we realized that it would also be beneficial to open it to individuals, therefore allowing anyone to work on their goals, professional or personal.
What it does
Chatbot Coach is a conversation robot currently integrated in Slack and Microsoft Teams, that focuses on bringing coaching via text rather than (or alongside with) face-to-face coaching. The bot includes 50+ frameworks from coaching, cognitive psychology, behavioral sciences, mental models, systems thinking, and strategy.
How we built it
We decided to slowly phase out the human component: Starting with human-led coaching, then trying out hybrid 2+2 (2 human-led session and 2 bot sessions), then reducing it to hybrid 1+3 (only 1 human-led session for 3 bot sessions). For each iteration, we paid cautious attention to the user feedback and progressed from taking it into consideration.
Challenges we ran into
- People want illusion of control: added buttons to give them more choice
- People need a quiet space to focus: sent reminders in the first session to prompt the user to isolate themselves, turn off notifications etc.
- People expect to get mentoring: set expectations on coaching in the first session
- Businesses buy slowly and cut HR budgets due to the Covid crisis: started selling B2C
Accomplishments that we're proud of
- 0% dropout rate from the recent projects
- Big initial demand for the webinar: 115 people expressing interest after posting in just 3 FB groups working with the biggest companies in the world (clients here)
- Not giving up and iterating: the company is 5 years old, we started with a completely different idea but pivoted to where we are today
- We won a number of hackathons and awards (see the deck for examples)
What we learned
- Full automation of coaching is possible
- Instead of using NLP to generate content, we should use NLP together with a recommendation engine AI to recommend next exercise, content written by expert coaches structured in a modular way
- B2C and B2B can reinforce each other - we sell B2C to HR, they will promote us in the future
- Coaching is a big need, but is very expensive - friction for adoption
- It’s good to start manually and slowly automate.
What's next for Chatbot Coach
- Investing in product development
- Reaching full automation
- Scaling through B2B and B2C, which reinforce each other
Log in or sign up for Devpost to join the conversation.