We provide a sample chat web app that uses the standard React.js library, so you can host all chat assets within your website.  If you prefer, we can provide a single web URL for your production chatbot, so that you can host your chatbot interaction on our hosted web service.  Finally, we provide a command line interface (CLI), so you can create your own proprietary chat interface to suit your specific needs.

You can use multiple types of data to provide the knowledge facts for your chatbots.  We currently provide three main types of plugins to provide facts for a chatbot.

• SQL Database Query Results.  You can have a chatbot that makes database requests to corporate databases or third-party application databases.  The queries use industry standard SQL, so you can integration with Oracle, MySQL, Microsoft SQL Server, IBM DB2, etc.  Your chatbot plugin can query your databases in real time to get up-to-the-minute results.
• REST API Results.  If your knowledge sources provide REST API access, we can help you create a custom plugin to query that data using Python. Examples of applications with REST APIs are Salesforce, PeopleSoft, ADP, ZenDesk, and hundreds of other enterprise applications.  By using a REST API, your plugin has access to live information from your applications.
• Facts Expressed in YAML.  We can help you import other types of fact data stored in PDF files, Excel spreadsheets, JSON data, and other formats by translating that data to industry-standard YAML format.  By automating an update process, your fact data can be kept up to date for use by your chatbot.

Our professional services team can also create custom plugins for dealing with other types of data in a secure and auditable manner.

The administrative interface displays a log of all of your chatbot interactions, and your admin staff can download your log data for routine analysis. We also log the interactions between the Large Language Model (e.g. ChatGPT) and our plugins that implement your fact base, so you can see what information is passed between the LLM and our software.

We train your admin staff to add data to your fact base.  Once the designed data is loaded, your team can test the chatbot before it is deployed.  Your data can either be accessed live using SQL from relational database, or via queries to your enterprise applications using a REST API.  We can also work with you to create your fact data in a standard YAML representation from a variety of data sources including PDF documents and Excel spreadsheets.

Your data will all be converted into an industry-standard YAML format.  Converters exist to translate a wide variety of data representations into YAML, such as PDF, XML, JSON, Excel, and many others.  You can also build live links to SQL databases and REST APIs.

Absolutely not.  Your data resides solely on our secure servers, so it can never become training data for someone else’s machine learning models.  Our service works as a plug-in to a generative AI system like ChatGPT, so ChatGPT has to call our server for facts to incorporate into the responses our chatbot gives.  This keeps your data out of the hands of the large machine learning companies.

If you want ultimate protection for your proprietary data, we can assist you in deploying the open-source version of Smarter.sh onto your own servers in your own data center.  In this case your proprietary data need never leave your own servers.

Smarter is a hosted service.  You only pay for what you use. Your cost for the service is based on the amount of text that is sent between our servers and your users.  You can use your account information on our server to monitor your costs.  There is currently no annual contract to use Smarter.

If you use our open-source version running on your own servers, our software tracks only your volume of chat queries, so you still pay based on your usage volume.