How Software Robots Work
What Do You Mean – My System Can’t Do That?
The number one comment we hear from customers is,
“I thought my systems were smarter than this!”
Well, they often are not.
The leading CRM and Marketing Automation platforms are not very good at correcting or repairing data. They are really optimized for one-on-one record editing by your employees. They lack the power, for example, to correct the spelling of “Flarida” to be correctly spelled “Florida”.
Does that matter? Yes – because if the data comes in with errors, the default solution for a CRM is to ‘drop’ the data – that is, the system fails to add the record, regardless of the value of the rest of the record.
Using, Znu Robots, we bypass this failure to correct the data before the CRM can reject it, improving the acceptance of cross-platform data and overall business results.
Traceable, Auditable Methods
Some AI methods are not similar at all to the human equivalent. Image recognition, for example, uses a set of techniques that do not resemble a human process at all.
Unlike image recognition, the AI and ML used for data enhancement closely mirror how humans do the same task, except with far more information, faster processing and a much better error rate.
These ‘human similar’ robotic processes give Znu Robots the power of complete auditability; every operation can be traced to a logical set of steps that explain the outcome.
Machine learning (ML) is the study of computer algorithms that can improve automatically through experience and by the use of data. It is seen as a part of artificial intelligence. Machine learning algorithms build a model based on sample data, known as training data, in order to make predictions or decisions without being explicitly programmed to do so. Machine learning algorithms are used in a wide variety of applications, such as in medicine, email filtering, speech recognition, and computer vision, where it is difficult or unfeasible to develop conventional algorithms to perform the needed tasks. [Full credit to the Wikipedia article on Machine Learning]
What do typical full deployments of Plan9 AI Robots look like in mid-sized businesses?
You can start with a single robot, and if your needs are fully met, you can stop there. Many businesses want a 360° view of all data and customer behavior inside their product and/or systems. In those cases, a full deployment similar to the below diagram is typical.
How long does all this take? Usually a few months from beginning to end for a large-scale deployment.
Single-purpose RPA Robots can normally be deployed in one day or less.
What About ‘Self-Service’ RPA?
We are aware that several industry analysts have tried to define RPA as a set of self-service (end user driven) workflow tools.
Au Contraire
‘Self-Service’ RPA will be even less successful than self-service marketing automation:
- Data integration requires API level expertise – period – not a visual mapping tool view to ‘connect field a to field b’ – the work requires code level expertise. If you search for the API docs for your platforms, you will see what we mean.
- ETL and data transformation are sub genres of data engineering – not exactly a common set of skills.
- The ‘Transform’ logic in ETL requires code. Sometimes small code, and sometimes medium or large code. An automation builder or workflow tool can only perform rules already built. What about when you need new rules?
- Race conditions; Connection errors; API errors; Data encoding; JSON; Corner cases – all require years of experience.
- Unlike manually run tools, RPA Software Robots run silently when needed, usually every 15 minutes; that is 35,000 times per year. You can’t have errors and ‘opps’ problems with these robots. You need rock solid, 100% reliable RPA, not a self-service kludge.
At Plan9, we build RPA Software Robots to solve business problems. We have never seen a workflow tool that can build ANY of our Robots.