Skip to content

Affiliates

Seeking folks involved in the AI space, and in contact with people who deploy conversational AI agents.

Problem

To understand user input many ping LLMs with a multiple choice question, and request a JSON object. This is fraught with issues including unpredictable responses, hallucinations, latency, privacy loss, monthly bills, and most importantly will always miss out on things the user says due to the narrow focus of a single multiple choice question.

Solution

The Sophia NLU Engine is a natural language understanding engine, and provides an alternative way to process user input:

  • Developed in Rust with key differential being it's self contained and lightweight nature. No external dependencies or API calls, Processes about 20,000 words/sec.
  • No bulky models, only needs the simple vocabulary data store of 145k words (79MB) or 914k words (177MB).
  • Massive boost compared to Python counterparts, which are bulky multi-gigabyte installs that process at most 300 words/sec.
  • Contains POS tagger, named entity recognition, phrase interpreter, anaphora resolution, automated spelling correction, and more.

Premium Licenses

Standard dual license model, base NLU engine is free and open source with premium licenses available that include:

  • Binary app that acts as localhost RPC / CLI server allowing integration with any programming language.
  • Custom vocabulary / entities.
  • Selector based phrase matching, akin to custom intent recognition.
  • Built-in user feedback pipeline with full statistics to gauge your user's experience.
  • Full usage statistics by time and category.

Details on how people can integrate Sophia into their operations including basic code examples available on the Integrations Guide page.

Contextual Awareness Upgrade

Shortly within a few weeks hopefully, a new upgrade will be available that contains:

  • Contextual Awareness - The ability to understand what it's reading and differentiate between phrases like "visit google.com", "visit the store", "visit Amanda's idea", "visit church", and "visit my parents"
  • Advanced Phrase Interpretation - Novel training regimen that will greatly increase its phrase parsing with custom, small and accurate models for the various language constructs. Aside from getting list of phrases with respective verb / noun clauses broken down, get full breadth of user input including questions, imperatives, declaratives and sentiments, so nothing is ever missed.
  • Multi-Turn Dialog - Define conversation schemas (eg. book flight) and Sophia will continue tracking multiple messages as a conversation whole until all necessary details are collected from user.
  • Multiple Language Support - Ability to easily load and run multiple languages at the same time from one instance

The Offer / Earn Money

Connect with people frustrated pinging ChatGPT for JSON objects to understand their users, and want a better solution. Sophia is already great, and in a few weeks with that upgrade in place will be far superior to the pinging LLMs for JSON object solution. Everything local and private, blazingly fast, custom vocabulary, 100% reliable and trustworthy results each time, get the full breadth of the user input without missing out, etc.

As the Purchase page lists, currently Premium licenses of $295, $1495, $2495. For everyone you refer, you get 20%. So sell a Professional license at $1495, that means $300 in your pocket.

If interested, reach out to me directly at [email protected] and I'll be more than happy to get you setup.