Automation Anywhere AI Agent Reviews

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.

The primary use cases predominantly involve banking, including check validation, fraud management, and anti-money laundering. We extract these from the core banking system and pull them into the agent, which then provides a response that we return to the system.

I predominantly try to integrate Automation Anywhere AI Agent with ServiceNow tools and Dynatrace for data tracing. However, the data is not moving correctly onto the integration platform.

Assessing the impact of Automation Anywhere AI Agent on workflow efficiency depends on the type of use case we have, such as compliance-related matters. With agentic AI, we achieve zero error. Otherwise, we calculate the cost and return on investment for that specific area. We also check the accuracy of our output and whether hallucination is present.

The data extraction feature has helped in processing tasks, but we use our own ingestion model rather than the Automation Anywhere model. We go with Hadoop and a data lake approach, process it internally, and then push the regular data to Automation Anywhere AI Agent.

The integration capabilities ensure smooth operations. We observe what comes into the pipeline and see the differences. I feel Automation Anywhere AI Agent is better than Blue Prism in that specific context. Generally, we go for Blue Prism in banking use cases, but of late, we are moving to Automation Anywhere AI Agent.

When considering weaknesses and improvements, the platform does not give us the liberty to use our own features where we can bring out creativity. We have to map the process to whatever is available, and the bidirectional integration is missing. If we have another agent coming from another tool and we are using it, integration is very difficult.

Price is an issue because when we go with AI, we are getting more efficiency. Comparatively to other tools, it is on the higher side. However, we do not consider the price point when we choose this kind of tool for banking. We go for the compliance side, license feasibility, and other factors. Pricing is high in the market, but we are not in pricing mode. We see other advantages where we can use license solutions that can be used in the majority of cases. We give business-as-usual support to the customer, so we bring in the cost to our bandwidth and do not charge the customer. There are many combinations we employ.

Additional features I would like to see in the future include faster learning capabilities. For single use cases, we require multiple agentic AIs. If we could combine them into one and do the decision-making with a single price, that would be better. Currently, for each specific condition, we use one agentic AI and then combine another agentic AI on top of it that controls all the conditions. This creates a complex situation, but if it were simplified, it would be good.

I have been working with Automation Anywhere AI Agent for around seven months.

If Automation Anywhere AI Agent does not fit the requirement, we make our own tools. We do not generally do this because the complexity is now greater. We are in the banking area, so we cannot move the data outside the premises. If it does not meet our needs and Blue Prism is not nearby, we make our own features and then provide a Gen AI solution for that if Automation Anywhere AI Agent does not come into the picture. In that case, we have an additional margin. Our competency is only Automation Anywhere AI Agent; we do not have Blue Prism competency in RPA. The companies I have worked for are only using Automation Anywhere AI Agent after initially using Blue Prism. We use only one tool at a time.

The initial setup for Automation Anywhere AI Agent is straightforward. We take the low-hanging fruit so it comes very clearly, and then we get approval from all stakeholders and institutionalize it. The initial setup is good, but expansion or scaling becomes a bit difficult or complex. Still, compared to other tools, it is good.

I do not use the Process Analytics feature.

I would describe the role of Cognitive Automation skills in managing complex tasks as very good. However, we are not using it now. They are moving more toward Gen AI, and Cognizant is not using much of Cognitive Automation skills anymore.

I am also dealing with BotFarm by Automation Anywhere AI Agent.

I would recommend BotFarm for companies in banking, financials, and wealth management, so all in the BFSI sector. Whether small bank or big bank, that is what we do.

BotFarm is not very popular; the standard Automation Anywhere AI Agent is more widespread.

I feel that Automation Anywhere AI Agent is the best option on the market when it comes to automation. I would rate this review a 7 out of 10.

My main use case for Automation Anywhere AI Agent is to extract data from different invoices such as bills of lading, document automation, OCR extraction, and speech-to-text.

In a recent project, a client sends me documents via their respective email, and I use document automation to extract invoice details such as invoice date, vendor name, amount, currency, PO number, and delivery note. After extracting the data, I receive the values in Excel format.

Mostly, I use Automation Anywhere AI Agent for document automation where there is a GenAI prompt to extract both unstructured and structured data in a structured format. The GenAI prompt helps me extract only the relevant fields, not other fields.

The best features Automation Anywhere AI Agent offers include the ability to extract data from unstructured data, such as invoice document formats that are not always the same, sometimes being handwritten or perhaps a low-quality scan.

Automation Anywhere AI Agent positively impacts my organization as it reduces manual effort and delivers output very quickly, allowing me to process documents for their intended purpose very quickly and effectively, cutting down the time significantly.

It saves me approximately four to five hours per day as I automate repetitive processes.

Automation Anywhere AI Agent is quite good overall, and I believe the main area for improvement is working on different OCRs to extract data properly.

The OCR and document understanding of Automation Anywhere AI Agent still struggle with complex and unstructured formats such as invoices with multiple layouts, poor scan quality, or handwritten fields. Even after training IQ Bot, I sometimes see accuracy issues and a need for manual validation.

Another improvement area is error handling and debugging since when a bot fails in production, the logs are not always detailed enough, increasing troubleshooting time.

I have been using Automation Anywhere AI Agent for approximately one year.

Automation Anywhere AI Agent is quite stable in its operations.

Automation Anywhere AI Agent scales very well in an enterprise environment as the Control Room handles multiple bots, users, and environments efficiently, allowing me to scale from a few bots to many in production without major performance issues.

The customer support provided by Automation Anywhere AI Agent is the best, being available 24/7, and they solve problems most of the time with a very high success rate.

I would rate the customer support of Automation Anywhere AI Agent a 10.

Positive

I am purely using Automation Anywhere AI Agent and did not use a different solution before.

I have definitely seen a return on investment with Automation Anywhere AI Agent. By automating repetitive processes such as data entry, report generation, and transaction processing, I reduced manual effort significantly as tasks that earlier took hours now take minutes, resulting in clear time savings.

My experience with the pricing, setup cost, and licensing of Automation Anywhere AI Agent is that the pricing and licensing model is suitable for large enterprises, but it can be expensive for organizations with high automation volume. However, the cost is justified because of its stability, scalability, and governance features.

For small and mid-sized businesses, the licensing cost can feel high compared to other RPA tools, and the license structure is not always very transparent at the beginning, requiring proper planning to avoid over-licensing.

Before choosing Automation Anywhere AI Agent, I did not evaluate other options because there are many features and options available to me, and I do not need to choose another AI agent.

My advice for others looking into using Automation Anywhere AI Agent is to start with a well-defined, stable process rather than a very complex one, as it works best when the process is standardized and repetitive. It is also important to plan for licensing, infrastructure, and scalability early to avoid cost or performance issues later.

Sometimes it does not capture what I am trying to extract, such as the date format, and it becomes confused with the date or currency. For example, if the amount is AED and there is something such as a bracket or a line, it may be confused and provide AFD or other incorrect values.

Overall, Automation Anywhere AI Agent is a strong enterprise RPA platform that delivers good value when implemented correctly, working best for organizations that plan automation strategically and invest in proper governance and training. With continued improvement in AI accuracy and usability, it can become even more competitive. I would rate this review an 8.

I am currently in POC mode implementing one of Automation Anywhere AI Agent solutions for a customer in the insurance industry. My customer wants to build Automation Anywhere AI Agent for making underwriting decisions. We are building the LLM for the underwriter team who gives the final confirmation on insurance policy processing. The underwriters review all the rules being run and if all the rules are passing properly through all stages, then the system gives the decision. This entire flow is being built with Automation Anywhere AI Agent.

During the POC mode, we built a small use case where we could see potential difference in the underwriter team workflow. After the new business branch and all the processes, once the policy is passed through, an underwriter gives the decision whether this policy is confirmed to proceed or if there is any further investigation, medical checkups, or other requirements needed. The decision is completely with the underwriter. Currently in the manual stage, we run all the rules for checking whether the policy is proper to hand over to the customer or if it requires more checks. The underwriter performs all this process manually. We are now building all these checks through Automation Anywhere AI Agent which will do all the AI checks and then give the decision whether the policy is to be issued or not.