Chat Bots

2016 was challenging, 2017 was learning, and in 2018 the Bots will break the world! by not replacing but bridging the gaps between brands and Functions, and their consumers and users.

The Taxonomy of Chat-Bots

Retrieval-Based vs Generative Models

Retrieval based model (easier) use a repository of predefined responses and some kind of heuristic to pick an appropriate response based on the input and context.

Generative model (harder) generates a new response from scratch.

Long Vs Short Conversations

The longer (harder) the conversation are, more difficult it is to automate them, as long threads tend to have multiple questions in them

Short-text (easier) conversations, where the goal is to create a single response to a single input are a lot easier.

Open Domain Vs Closed Domain

In an open domain (harder) the user can take the conversation anywhere. E.g., social media sites. The infinite number of topics and the fact that a certain world knowledge is required to create reasonable responses makes this a hard problem.

In a closed domain (easier) setting the space of possible inputs and outputs is limited as the system is trying to achieve a very specific goal. Technical support, shopping assistance.

Tech Mahindra — Makers Lab presents Entellio

A In-House developed Enterprise Chat-bot Framework. In the above terms, Entellio is a combination of:

• Retrieval based taxonomy based on word vector semantics (an unsupervised technique of word vector model)
• A hybrid of both long and short conversation in the mix
• Closed domain, centered around the Enterprise context.

What makes it special:

• Open source tech provides a unique plug and play functionality and does not require your data to be put on public cloud (IBM Watson)
• Performs Natural Language Processing and Latent Semantic Analysis
• Available both on Mobile and Web. Designed for the Enterprise

Architecture design technique used:

• Simplicity: CSV/Excel file with FAQs can be dropped into the Bot, to enable users to have semantically meaningful conversations.
• YAGNI (You ain’t gonna need it principle): if it had no value, it was removed from the design of the framework.
• Separation of concerns: A well demarcated separation f concerns is maintained between front-end, middle-ware (ML server) and DB is maintained to avoid coupling and maintain cohesion between modules (based on REST service model).

Use-Cases from within the House:

• UVO — HR, is an employee chat-bot that solves HR queries and engages with over 115,000 TechM employees.
• UVO — Targetter, a field sales assistant chat-bot.

Both built on Entellio enterprise chat-bot framework.

This experience will not be complete without a hands-on experience.

https://entellio.techmahindra.com you can register upto 3 bots on this cloud based environment, in an enterprise version, this is unlimited. Instructions Manual Link

50+ Chat-Bot ideas that will set your enterprise apart in the market:

A bot to deal with workplace harassment. Asset management Bot. Personal Assistant Bot. Grammar correction bot. KPI reporting and monitoring bot for managers. Get healthy & mindful bot. Social club bot. Expert finder bot. Code bot that helps beginners find things. Org policies questionnaire bot. Book club bots. A team voting bot. motivation and inspiration bot. notification bot. on-call support bot. stand-up bot. a bot to monitor usage of other bots. compliance monitoring bots. legal review assistant. language translation bots. PM assistant bot. Team playbook bot. Vacation assistant bot. Meeting scheduling bot. Job satisfaction bot. new person on-boarding bot. relocation bot. Expense report bot. interview feedback collector bot. approvals bot. Interview process management bot. Find a cause bot. News bot. …

Avaamo (from Tech Mahindra start-up eco-system)

www.avaamo.com A platform that includes speech synthesis, speech recognition, multi-faceted machine learning, neural networks and natural language understanding along with data categorization, NLU services, Semantic search and holds 7 patented technologies.

Plug and Play for retail

From promoting omnichannel engagements to boosting store associate productivity.

• Conversational Commerce: Discover products conversationally, receive predictive offers, and even complete their purchases in a seamless manner. Cart abandonment rates cut by nearly 90% and show a marked increase in omnichannel engagement of shoppers.
• Returns, policies, and RMA: Drive product registrations, provide post-purchase assistance, and simplify warranty lookups and return merchandise authorizations. Providing more information about return policies and process has been shown to dramatically reduce the RMA requests as well as the reverse logistics costs.
• Seasonal demand spikes: An average store associate spends 30% of his time hunting for information for customers across various terminals and every week has to memorize a 30-page PDF to stay on top of current store deals in his category. Using conversational AI through mobile messaging apps, you can provide in-the-moment sales assistance to their store associates. Similarly using in-stream AI-based guidance, you can quickly ramp up their Call center agents as well.

Personal Shopper Demo Link, Warranty Advisor Demo Link, Training Advisor Demo Link, In-Store Assistant Demo Link.

What sets Avaamo apart:

• 6 industries, 27 countries, 1M+ interactions/day.
• Connectors to 150+ enterprise systems
• Security and compliance built in for PCI, FINRA, HIPAA
• Can be deployed on existing channels including the web, mobile, SMS, Facebook, Alexa, Google Assistant or even your existing 1800 number.
• 14 language models voice & text