As businesses are increasingly laying their focus on artificial intelligence solutions, it might be important to understand what goes into implementing these solutions.
Most business product and solutions have now taken to the tone of customer satisfaction, making every solution look the same. It’s only a matter of time, that they will all offer or at least speak about artificial intelligence. I would assume, it would become increasingly difficult for the businesses to understand the truth behind these systems.
With some of my discussions with the peers, concerns such as “what happens to the investments made on existing platforms and data silos” and notions like “Data will be plugged to an AI platform and its all about adding another application into your ecosystem” were prevalent. Another interesting concern that caught my attention was “What happens when we have less data initially where we have two different behavior data from two different customers? How does the traditional ML programs learn from this data?”
To answer these questions logically and to elaborate how Plumb5 solves these concerns, I put together a simple comparison between AI platforms, Plumb5 and the most effective natural intelligence model(Brain). I hope I have answered these concerns using 5 macro points which allow anyone to reason and evaluate the right AI platform for their business.
| #1. Is the data collection app and the data processing app integrated? | |
| Human Brain | Yes. The processing unit is seamlessly integrated with the data collection touch-points (sensory organs) for instant intelligence. They work with a simple concept of “Neurons that fire together, wire together” and are able to extract patterns to process data across touchpoints in real-time. The Spatio-temporal data is specific to objects and helps in building intelligence around objects. |
| AI Platforms | No. The data ecosystem in a business is lying in silos and is integrated manually for data processing. This creates latency. Since systems are not data-aware, they don’t wire together causing data discrepancies leading to sub-optimal results |
| Plumb5 | Yes. Plumb5 integrates data from all touchpoints to a single data-aware platform to ensure real-time processing. Working on the concepts of the brain, the platform wires together data associated with a single customer which enables in real-time pattern extraction and process data for real-time intelligence. The platform mimics the brain in terms of arranging the temporal data which helps in building intelligence around business primaries like customer and product. |
| #2. Is the data model designed for real-time association and synthesis of every data parameter collected | |
| Human Brain | Yes. The brain exhibits intelligence through real-time learning across any data parameter (but, restricted to the data collected by its sensory organs). Logically this requires data organization that facilitates quick retrieval of data without having to query large amounts of data. Simply put, the brain uses the most advanced data model which helps it to process quick insights on the fly |
| AI Platforms | No. The data is completely disorganized and resorts to querying mechanism over large datasets. Data relationships are incomplete due to this disorganization due to which businesses suffer from latent decision making and bad expenses from redundant short term solutions. |
| Plumb5 | Yes. Plumb5 is a data-aware platform which solves the process of complex data querying. The bottoms-up approach to designing the unified data model allows the business to scale based on any new data parameter collected and process insights in real-time without worrying about big data hauls. |
| #3. Is it designed for real-time learning and intelligence | |
| Human Brain | Yes. The brain is an unsupervised learning model that has the capacity to learn on the fly through observation and guidance. Using this learning, it can exhibit intelligence in real-time, even with a single record. |
| AI Platforms | Not yet. Supervised learning models are being considered. However, due to traditional machine learning practices, the challenges in achieving real-time intelligence are still not solved. Most learning models resort to larger data sets to learn and this process is currently latent. Few theories like the HTM (Hierarchical Temporary Memory) look promising to learn with minimum data sets but do not process in real-time. Since data organization is not fixed, the risk of feeding incomplete data to learning models can pose threat to productivity and returns. |
| Plumb5 | Yes. Plumb5 is designed to learn and process intelligence in real-time. Similar to the ART (Adaptive Resonance Theory), Plumb5 runs the 4 step intelligence extraction process of identification, comparison, weights and response selection. Since Plumb5 is a supervised learning platform, the platform needs pre-set instructions on weight allocation to exhibit intelligence. |
| #4. Is the platform good to use this intelligence and take next appropriate actions in real-time | |
| Human Brain | Yes. The brain is seamlessly integrated with the motor parts (arms, legs etc) which help them take quick actions using intelligence, allowing them to be completely autonomous. |
| AI Platforms | Yes at a basic level. You can see a lot of businesses using automation are now setting up triggers to carry out subsequent actions. But these triggers are based on current response states and don’t necessarily involve past intelligence. Even if past intelligence is included, the action triggers don’t fire in real-time due to pre-processing routines and latent inclusive learning. Unavailability of bi-directional data flow further hampers real-time action delivery |
| Plumb5 | Yes. In contrast to other AI processes, Plumb5 provides bi-directional data flow by connecting back to the touch-points for initiating triggers. As Plumb5 holds the entire behavior pattern of a single customer’s lifecycle, cumulative analysis of past and present behavior is computed in real-time. The supervised weights allow the system to compute states which allow the machine to initiate next action using intelligence. As mentioned earlier, Plumb5 is a supervised platform and needs instructions to map states to delivery packets.(templates) |
| #5. Can the platform learn from these actions and fine tune its intelligence in real-time | |
| Human Brain | Yes. It can. |
| AI Platforms | No. Non-availability of directional flow and incomplete data tagging poses real challenges in achieving continuous learning. |
| Plumb5 | Yes. The Intelligence extraction process allows the platform to learn as it performs and consolidate intelligence through continuous learning. |
Couple of questions that people asked when I spoke to them about Plumb5
Does Plumb5 have built-in machine vision algorithms?
Plumb5 integrates with machine vision tools for extracting face detection and product image parameters like shape and color. The extracted tags are wired to the individual customer stack which is further used in triggering contextual recommendations.
What about Natural Language Processing (NLP)?
Plumb5 has a built-in NLP parser which is used to detect customer sentiment from customer conversations. The extracted tags are wired back to the stack to generate conversational assemblies, which can be integrated back with bot frameworks.
These views strictly belong to the author and no claim made by the product or the business. Would love to hear your feedback/ comments/ opinions/ concerns.