AI Agent - Technology Copywriting

Positive

The main use case for Automation Anywhere AI Agent  in the organizations I work with is bank reconciliation and PO invoicing. The use cases depend on the segment. In one of the use cases, bank reconciliation was implemented, along with awarding POs or creating invoicing and raising it based on the PO that they used to receive from the customer. All of that documentation or work which was bulk in number and very repetitive for the organization normally fell into banking or manufacturing. These were two of the use cases which I have promoted or implemented at multiple places.

For PO invoicing in manufacturing or banking, Automation Anywhere AI Agent  was ingesting all invoices from email extracts and reading the line item data using the IQ bot. It was validating this data against the PO and GRN data from the SAP in the organization. It was autonomously resolving common mismatches, such as price variance within the tolerance or missing PO reference, by querying ERP  and vendor master data. Exceptions such as quantity mismatch were routed through the automation bot with a recommended action. With respect to bank reconciliation, the agent compared the core banking transactions with the statements from multiple partner banks. It used artificial intelligence to auto-match transactions with date, amount, variations, and narration. The unmatched items were categorized and posted automatically where rules allow.

I worked with a customer called SNS Group, which was my customer. I onboarded them on Salesforce . In both PO invoicing and bank reconciliation, the agent was limited to data capture. It applied policy-based reasoning to manage the tolerance limit and vendor SLAs, aging rules. It was learning from the historical resolutions, such as which vendor usually sends partial invoices and which banks post delayed fees. There are many use cases across industry, but these are two which I was part of the implementation team.

The positive impact of Automation Anywhere AI Agent on organizations is noticeable, and cost saving was definitely the best outcome. Considering the current environment after 2022, the world is moving towards a cost saving environment with vendor consolidation. The only objective is how to save costs. At the same time, companies are also focusing on efficiencies in processes such as invoice processing, reconciliation, and cash application. These all require at least 80 to 90% accuracy and efficiency, which Automation Anywhere AI Agent was helping to achieve. Cost saving is something the entire world is moving towards. The third benefit is proper accuracy and control. Error rates literally dropped when Automation Anywhere AI Agent comes into the picture because decisions were constantly and consistently applied based on patterns and policies. The audit trails, tolerance checks, and exception reasoning were built into the process. Since all three benefits are achieved—accuracy, efficiency, and cost saving—the business impact and dollar impact is really high.

The improvements SNS Group saw included a reduction in manual hours, error rates, and faster invoice processing times due to Automation Anywhere AI Agent. Invoice processing time literally reduced to hours within the same day, which is eight hours. Initially, invoice processing used to take four to five days. After the salary disbursement, vendor payments used to be done in the mid of month and took around four to five days. Now it is within eight to max 24 hours, and the entire invoice processing cycle time is reduced. Bank reconciliation went from T+3 to same day. Initially, it used to take more than three to five days, and now the bank reconciliation is completed on the same day. The manual effort reduction was 70%, which is a huge number. Reconciliation was reduced by almost 80%, and invoice processing by almost 70 to 85%. The cost reduction was almost 25 to 30% for SNS Group. With respect to error reduction, since the bots were literally following the policies and learning so quickly, the error reduction was almost 90%, which is a good number.

The best features Automation Anywhere AI Agent offers include OCR and reading the OCR. I personally felt the OCR technology was one of the best features that Automation Anywhere  had. At that time, I was also competing with UIPath, and UIPath was struggling with this particular requirement. The top features that stand out include the AI logic and autonomous decisioning. It was not just the data extraction, but the AI agent interpreted the data, decided, and acted upon it based on policies. Reading natural language was another valuable feature that I loved. Users can interact using plain language through chat, email, or ticketing. For example, a user could ask to help with all POs valued more than half a million dollars where approval is pending. AI-driven document processing was definitely valuable, as OCR and semantic understanding were very good. The system was able to handle invoices, contracts, bank statements, and delivery receipts. Self-learning was another thing which I would talk about, as the system was good enough to learn itself and improve the accuracy over time.

The self-learning aspect of Automation Anywhere AI Agent is significant. Nowadays, it is not a big thing because all the AI tools are doing it. However, when I was promoting it or being part of the implementation, what it really meant to me is that it was not unsupervised AI randomly changing behavior. It was guided learning from outcomes into workflows. The system observed how humans resolved exceptions and got better at doing the same next time. When a ticket or case was received and the knowledge facts did not have the response, a human intervened and responded with the right resolution. The system was smart enough to understand the resolution and update the knowledge fact, so the matching logic got smarter without new rules. Exception recommendations became more relevant. The system was also able to understand vendor and counterparty behavior modeling. There are many things, not just one, but a lot of things which can always be talked about.

Automation Anywhere AI Agent could be improved in a few ways. Since I am currently working with Salesforce  and I understand that to compete Automation Anywhere  as a middleware product with Mulesoft or Informatica, the implementation phase needs to be quicker. The POC is always successful, but the implementation phase has to be quicker. The business needs to understand this better. Automation Anywhere should get into the business team and focus on explainability at the business level, not just a technical thing, but in the language or discussion where the business team can understand. There is clear business reasoning missing right now. The learning is strong within the process for sure, but there is limited reuse of learning across processes, which I think is very important. Simple model governance for business users is needed so that anyone should be able to use or create bots on Automation Anywhere. Right now, consultants and partners are needed in place. Additionally, Automation Anywhere should have native integrations with core platforms such as Oracle, SAPs, Salesforce, and ServiceNow . It should have out of the box integration rather than going for APIs because the deals which are not getting closed with Automation Anywhere or UIPath are purely because these organizations do not have out-of-the-box connectors available.

Automation Anywhere AI Agent is stable in my experience, and I have not really seen any issues with downtime or reliability. It is improving day by day.

The scalability of Automation Anywhere AI Agent has been impressive. The work my previous company was doing was with a lot of big banking companies across the Middle East. The scalability was definitely a good thing with Automation Anywhere AI Agent.

I have no experience with customer support for Automation Anywhere AI Agent. Since we ourselves were the implementation partner, I do not remember taking more support from the Automation Anywhere team unless it was regarding the licenses part.

Negative

I did not previously use a different solution before Automation Anywhere AI Agent. It was my first RPA  solution.

My experience with pricing, setup cost, and licensing for Automation Anywhere AI Agent indicates that the pricing with Automation Anywhere is not great. UIPath is reasonably better with competitive pricing. There are also a lot of players in India, including Automation Edge, which I think has a bit more competitive pricing. Licenses are usually purchased per bot, per named user, and per AI agent seat depending on usage patterns. However, the cost was not a major issue. The major issue was setup cost. The implementation cost even for a POC was very high, and that was a pain for all the customers.

I worked with a customer called SNS Group, which was my customer. I onboarded them on Salesforce. In both PO invoicing and bank reconciliation, the agent was limited to data capture. It applied policy-based reasoning to manage the tolerance limit and vendor SLAs, aging rules. It was learning from the historical resolutions, such as which vendor usually sends partial invoices and which banks post delayed fees. There are many use cases across industry, but these are two which I was part of the implementation team.

I have seen a return on investment with Automation Anywhere AI Agent. The operational cost has reduced by 25 to 40%. The overtime staffing which they used to do has gone down to almost 50%. Invoice processing, which used to take three to five days, is now done within 24 hours. Cost avoidance has gone by 40%. Overall, there are very good numbers to speak about.

My experience with pricing, setup cost, and licensing for Automation Anywhere AI Agent indicates that the pricing with Automation Anywhere is not great. UIPath is reasonably better with competitive pricing. There are also a lot of players in India, including Automation Edge, which I think has a bit more competitive pricing. Licenses are usually purchased per bot, per named user, and per AI agent seat depending on usage patterns. However, the cost was not a major issue. The major issue was setup cost. The implementation cost even for a POC was very high, and that was a pain for all the customers.

The improvements SNS Group saw included a reduction in manual hours, error rates, and faster invoice processing times due to Automation Anywhere AI Agent. Invoice processing time literally reduced to hours within the same day, which is eight hours. Initially, invoice processing used to take four to five days. After the salary disbursement, vendor payments used to be done in the mid of month and took around four to five days. Now it is within eight to max 24 hours, and the entire invoice processing cycle time is reduced. Bank reconciliation went from T+3 to same day. Initially, it used to take more than three to five days, and now the bank reconciliation is completed on the same day. The manual effort reduction was 70%, which is a huge number. Reconciliation was reduced by almost 80%, and invoice processing by almost 70 to 85%. The cost reduction was almost 25 to 30% for SNS Group. With respect to error reduction, since the bots were literally following the policies and learning so quickly, the error reduction was almost 90%, which is a good number.

The pricing with Automation Anywhere is not great. UIPath is reasonably better with competitive pricing. There are also a lot of players in India, including Automation Edge, which I think has a bit more competitive pricing. Licenses are usually purchased per bot, per named user, and per AI agent seat depending on usage patterns. However, the cost was not a major issue. The major issue was setup cost. The implementation cost even for a POC was very high, and that was a pain for all the customers.

I would rate this review as a 7 out of 10.