That's quite a bit of infrastructure for starting out!
Are you expecting to be handling tens of millions of requests per day out of the gate, or is this more a case of wanting to not have to think about scaling for a long time?
1
Good questions, the short answer is both!
Imagine a scenario I get a few customers with very high traffic and they install ChipBot on their website. I don't want that impacting my existing customers, especially my first ones.
Infrastructure isn't fun for me so I wanted to complete that work ASAP and not do it for a while. What's not in the diagram is the data ETL process, but I'll save that for another post.
Also the cost is small for this infrastructure, I'm paying ~$150/mo on AWS for this setup. If I wanted to 2x throughput, it would only add ~$50 a month.
I think Heroku or other IaaS solutions are very powerful tools for MVPs and I could have used them early on to save even more time. But I personally like AWS because I know it and it can be cheap for me in the long run.
2
Thanks for the great post. It has really inspired me. There are couple of things I would like to know about.
How chatbots really help in someone else business: specially the product like yours.
And how it works: a rough idea
May be that would inspire me to build one ans solve few problems.
Thanks again.
1
Thanks for the reply and these 2 great questions.
How chatbots really help in someone else business: specially the product like yours.
A chatbot's advantage comes with ease of information and ability to guide the user. If you're a business owner running it, you even can find user behavior patterns over time after chatting with so many users.
Usage example: Imagine you're on the Apple.com iPad page, and you wanted to know the tech specs, without going to the tech specs page. A bot floating on the bottom right can rescue the user, inform them, then guide them to the next call-to-action.
For ChipBot, I take a different approach to bots by using conventional experiences to find information but preserve the "chatbot" experience (e.g.: floating widget, database of knowledge, call to actions).
And how it works: a rough idea
In a nutshell, ChipBot crowdsources questions from your users.
It has all the same features as a web chatbot but focuses on search rather than conversation. ChipBot records all events like searches, question views, clicks, and creates a macro view of how your users behave with your content.
2
Thanks zillion. Simple and clear. I am not familiar with AI but would love to try once bases on data collections and behaviors.
That's quite a bit of infrastructure for starting out!
Are you expecting to be handling tens of millions of requests per day out of the gate, or is this more a case of wanting to not have to think about scaling for a long time?
Good questions, the short answer is both!
Imagine a scenario I get a few customers with very high traffic and they install ChipBot on their website. I don't want that impacting my existing customers, especially my first ones.
Infrastructure isn't fun for me so I wanted to complete that work ASAP and not do it for a while. What's not in the diagram is the data ETL process, but I'll save that for another post.
Also the cost is small for this infrastructure, I'm paying ~$150/mo on AWS for this setup. If I wanted to 2x throughput, it would only add ~$50 a month.
I think Heroku or other IaaS solutions are very powerful tools for MVPs and I could have used them early on to save even more time. But I personally like AWS because I know it and it can be cheap for me in the long run.
Thanks for the great post. It has really inspired me. There are couple of things I would like to know about.
How chatbots really help in someone else business: specially the product like yours.
And how it works: a rough idea
May be that would inspire me to build one ans solve few problems.
Thanks again.
Thanks for the reply and these 2 great questions.
A chatbot's advantage comes with ease of information and ability to guide the user. If you're a business owner running it, you even can find user behavior patterns over time after chatting with so many users.
Usage example: Imagine you're on the Apple.com iPad page, and you wanted to know the tech specs, without going to the tech specs page. A bot floating on the bottom right can rescue the user, inform them, then guide them to the next call-to-action.
For ChipBot, I take a different approach to bots by using conventional experiences to find information but preserve the "chatbot" experience (e.g.: floating widget, database of knowledge, call to actions).
In a nutshell, ChipBot crowdsources questions from your users.
It has all the same features as a web chatbot but focuses on search rather than conversation. ChipBot records all events like searches, question views, clicks, and creates a macro view of how your users behave with your content.
Thanks zillion. Simple and clear. I am not familiar with AI but would love to try once bases on data collections and behaviors.