Delivering Responsible AI

We are an AI services business, focussed on building and implementing AI solutions that are transparent and dependable. Our experience in Symbolic AI helps us achieve decision automation using machine intelligence with less ambiguity.

Our Capabilities


Ongoing Projects

Data Discovery
A Product company wanted to discover user patterns from product usage data to personalize relevant features
The solution was to create a N:1 user-centric data from the activity log and create longitudinal record of the user's temporal pattern.

These patterns were used to compare with optimal patterns to understand deviations in user's flow allowing in identifying features that were used, unused and undiscovered. These patterns were mapped to user-roles to understand their affinity towards a specific feature. Based on these discoveries, in-product notifications were delivered to drive feature discovery and usage

These metrics also helped the business to predict subscription continuity of a specific user and use these insights to predict subscription revenues.
A Data Science company was looking to monitor aviation activity, economic health and climate impact due to COVID 19
The solution was to build dashboards to monitor multiple indicators related to travel restrictions, aviation status and its impact. A gist of what these dashboards contain is listed below

  • Travel Restrictions: Status of Air-Bridges and share of routes, which are currently operational and comparison with pre-crisis levels
  • Aviation Status: Track the current status of the aviation sector (groundings of aircraft, numbers of passengers travelling, etc)
  • Avaition Impact: Analyse the impact of current levels of aviation sector activity on the global economy, and the current climate impact of the aviation sector
Behavioral Segmentation
A Retail Major wanted to recommend products based on customer's intent and past purchase behavior
The solution was to model customer behavior to understand the intent of a customer towards a particular product.

As the data-sets came from a mature CDP, customer interaction data and transaction data was structured. Over this data-set, we built an intent detection algorithm, which would calculate the intent rank based on user's frequency, recency, time spent and their depth of interaction.

This customer-product rank was made available to the collaborative-filtering algorithm along with past transaction data.

This allowed the recommendation engine to fine tune its recommendations based on past purchases and feedback from the customer, plus customers current intent rank along with cohort's purchase behavior
Autonomous Learning
A Internal Project where we are building real-time decision capabilities for Robots
For autonomous robots, real-time learning, reasoning and decision-making is key in order to perform in real-world scenarios. To achieve this, we are designing a declarative input graph, which will organize data in real-time, as it collects data from sensory inputs (cameras, sensors or other IoT devices).

Based on semantics, data is organized by objects and sequenced by time, which gets converted to symbolic sequences, acting as Episodic Memories. Robots can use these memories to make real-time decisions based on past and current data.

This gives us the opportunity to generate knowledge representations dynamically, which will allow machines to abstract, reason and plan its next move.
Data Compression
An AI company wanted to improve performance of their NLP models without any data loss
The solution was to develop an encoding algorithm for lossless compression using airthmetic coding.

Our approach of the adaptive model was to extract the text corpus for unique entities and group them based on POS (Parts of Speech) to create the first level of compressed characters or symbols.

Using semantic labels and dependency parsing logic, the compressed sequence is grouped with associated symbols to create a merged entity with a new symbolic representation of the complete sentence.


We have experience designing, implementing and automating learning and decision processes. Here is a list of capabilities where we can be handy.

We believe that the core of any AI exercise lies in its data models. We have experience in designing expert systems for design experiments as well as hueristic models for random observation

  • Heuristic/Implicit Models
  • Explicit Models
  • Knowledge Graphs
We have experience in implementing different models ranging from detecting spatial and temporal patterns to probilistic predictions.

  • Behavioral Analysis
  • Probabilistic Attribution
  • Collaborative Filtering
  • Time-series Prediction
  • Named Entity Extraction
  • Semantic Role Labeling
  • Inference Engine
We have exposure to designing systems that can integrate decision outputs to automate state changes. This capability allows in building autonomous machines that can learn and make decisions

  • Markov Decision Processes
  • Finite State Automation
  • Rule based Triggers
We also have experience in build data visualization layers, allowing users to not only visualize predicted outputs but also to visualize the inner decisions of the network at an individual node level.

  • Data Dashboards
  • Data Simulation
  • Search
  • N-dimensional Reports


Digging deep into Symbolic AI techniques

Artificial Intelligent State Machine

The article covers how we may integrate a node-based data structure with learning models to create a state machine that might have the capacity to embody super intelligence.

Autonomous AI

This article explains how a self-learning machine can exhibit autonomous classification, pattern detection or output prediction, using a simple data organization technique.

Handling Visual Parameters

The article covers the approach of extracting visual parameters to a knowledge ensemble incorporating shape, depth and color with nested relationships.

Embedding Language Processing

The article is a high level article on embedding language to the knowledge representation and the use of dependency parsing and hierarchies to abstract.

About Us

RM's proprietary techniques allow in providing decision automation solutions with great control and transparency. Our strengths lie in designing

  • Declarative Memories and Inference Engine
  • Object Oriented Data Structure for Behavioral Study
  • Integrating NLP Techniques (Named Entities) to Declarative Memories

Started in 2016, Responsible Machines(RM) started out with a vision to enable robots or machines to learn, speak, make decisions and exhibit general intelligence.

RM's Symbolic AI techniques which solves the symbol-grounding problem allows in exhibiting machine reasoning and abstraction, which acts as a foundation to generalized learning concepts, a more promising approach to human like intelligence


percent confident in building zero-redundancy data models which help in reducing processing required for AI workloads, thereby reducing carbon footprint.


Proprietary Algorithms in the area of converting data to Symbolic Sequences, Unique Entity Extraction and Machine Reasoning


Years of Experience combined, in designing, implementing and delivering Analytical and AI products and projects


percent savings guaranteed on any AI implementation by way of optimized processing, reduced storage and a smaller data engineering team

Working with

We are collaborating with multiple partners to provide integrated decision automation


Manoj Khanna

With over 20 years of senior leadership and deep consulting experience at major consulting firms, Manoj leads the vision at Responsible Machines. Having worked with HP, Mphasis, EPAM in the past, he has a reputation for implementing strategic competitive transformation programs.


Director-Data Sciences
Veer has more than 15 years of experience in building analytical and AI products. His previous venture "Plumb5" solved unstructured customer data challenges and was used by large enterprises like Walmart, Unilever, Aegon, Commonwealth Bank in solving their single customer view problem.

Margaret Morris

Margaret handles operations at Responsible Machines. She brings a rich experience of working with deep science projects. She devised the GEO-DMF System for Outer Space research and worked as an Assistant Director for the former Institute for Applied Archaeological Sciences at Barry University.

Frank Boehm

Frank Boehm handles market development and strategic partnerships for Responsible Machines. He is a Research Associate at the University of Guelph and brings his rich expertise of developing advanced nanomedical diagnostic and therapeutic technologies.

Contact Us

Responsible Machines

ResponsibleMachines(RM) is a deep-tech AI company and was started to make AI more transparent and perform general intelligence tasks using machine reasoning and abstraction.

511, Ave of Americas, #4047
New York, NY 10011

(Manoj) +1 (917) 740-2269

Your message has been sent. Thank you!