I've used Amazon Connect for various purposes. It's usually a combination of Amazon Connect, Lex, and other AWS services.

Primarily, we use it for call center operations – both inbound and outbound. Amazon Connect's recently released outbound features are great for our marketing and dealer teams.

For inbound, we use Amazon Lex for conversational voicebots and Amazon Connect for call routing. I've implemented these solutions for fleet management, a banking POC, and customers in areas like capital management, holiday management, and ticketing.

What I like most is how well it integrates with other AWS services, the ease of implementation, and the robust partner network that supports it. This has helped Amazon Connect gain over 3000+ customers in just six or seven years.

It's fairly easy for an agent with no prior experience to learn. A short training session is usually sufficient. Here, I'm assuming that the agents, while not tech-savvy, can follow instructions. If someone has experience taking calls for several years, the transition will be smooth.

Some adjustments are needed as most of Amazon Connect functions via a computer interface. Additionally, there's the option to route calls to traditional phones, but that's less common. Agents will need to understand how to navigate the interface, what the different icons represent, how to use the call controls, and how to work with any integrated systems.

So, a 15-30 minute training session should be enough to get agents up and running on the basics of Amazon Connect itself. If there are complex integrations with CRMs or other additional tasks required of the agent, then, the training would need to be more extensive.

There are quite a few areas, even though the platform is new and user-friendly. Here's what could be improved:

• Agent Desktop: Amazon Connect should enhance its native agent desktop. Intuitive UI should be present for the agent desktop. Currently, it forces users to rely on third-party CRM integrations. Predominantly, there is no support for full-fledged customization for the native agent desktop. More customization options for the native desktop would be beneficial.
• Channel Support:

  The platform was designed primarily for voice and chat. It relies heavily on third-party integrations (not native to AWS) like Lambda or Kinesis for additional functionality. The user ends up using multiple AWS resources, which end up adding costs and a lot of tech that needs to be known for basic usage as well.

  And there is no multiple-channel support. When I say the channels, it's only voice and chat, but no WhatsApp or other native channels, like Facebook or Twitter; it needs another integration to integrate with Amazon Connect. Native support for channels like WhatsApp or Facebook would streamline things.

• AI Features: The built-in AI language models are somewhat limited. While some languages are supported, highly customized prompts or AI versions aren't available without extensive feature requests and wait times.
• Outbound Campaigns: Outbound campaign features are still basic. More integrations for data sources, lead management, segmentation, and support across different regions would address regulatory needs and improve functionality.
• Reporting & Conversations: Amazon Connect has recently added conversation analysis and Contact Lens support. However, historically, third-party tools have provided more insights, especially when looking for quick insights or any sort of customization. The native solution doesn't give any predefined templates or reusable pieces, which can actually have a lot of customization potential for reporting. They give very basic stuff with no way to plot or integrate reporting with external data. Ultimately, these things have to be developed separately.
• Global Architecture: There is no global support. One instance usually doesn't talk with another instance; it has to be integrated and use a PSTN pull. There could be a global customer, but an Amazon Connect solution would force everything into one instance, maybe in US East, or maybe in Europe. If you wanted a multi-tenant setup, with calls routed from different locations and agents placed globally, that kind of support is something they don't have yet. Given the capabilities of AWS networking, they should build this out. Native support for multi-tenant global architectures would be helpful for businesses operating across locations.
• SIP Support: The lack of native SIP (Session Initiation Protocol) integration is a major drawback. It forces users to purchase or port numbers to Amazon. Native SIP integration is missing, though there are some beta workarounds. However, it's been years without official support. That's a major drawback, as a lot of use cases are dependent on SIP.
• Generative AI: Amazon Connect's reliance on Amazon Q and Amazon Bedrock for knowledge bases limits options. Any knowledge base that customers want to use has to be integrated through Amazon Q or Amazon Bedrocka. There's no support for OpenAI or other external language models (LMs). If a customer wants to use their own LMs, they have to bring them to Amazon's native LM space first.

These are some of the key areas where I see room for improvement in Amazon Connect.

I've been using Amazon Connect for about six and a half years, almost since its launch. I've been exploring its features and implementing it across different geographies.

I have contacted customer service and support for deployment and support activities.

In my experience, getting timely responses can be a challenge unless you have an account executive to help. That's because Amazon Connect support is handled by the same team as their other services, and they don't always have dedicated Connect specialists.

An account executive can help you get dedicated assistance, which generally leads to faster responses.

However, response times can still vary. Some issues have taken up to a month or two to resolve.

For a small, straightforward Amazon Connect contact center, setup is easy. However, if you need extensive back-end integrations – like external authentication, pulling data from third-party sources, complex data presentation, or integrating with other systems – the process becomes more complex.

In these cases, you'll need expertise beyond just Amazon Connect engineers. You'll likely require developers familiar with Lambda, other AWS services, and potentially additional programming and scripting languages. So, while a basic setup can be done quickly (even within a day), more complex integrations with Amazon Connect will certainly take longer.

The good news is that it's self-service. You can log in, create an account, and start working right away. You'll find icons in your Amazon settings to get started.

In my previous experience deploying Amazon Connect, we typically had a team of three or four engineers depending on the project. These engineers had expertise in cloud technologies, coding, telephony, and Amazon Connect specifically.

If you have a skilled engineer with comprehensive knowledge, one or two people can manage even a thousand-seat call center. The team size depends on factors like use cases, the complexity of integrations, and whether the deployment is global. Support needs generally require fewer people.

It is worth the money. Compared to a traditional on-premise call center, which can be costly and resource-intensive, Amazon Connect can offer significant savings. This is especially true if your use cases are straightforward, call volume is manageable, and you have an existing CRM to integrate.

Sometimes, the total Amazon Connect bill could be on par with, or even less than, what you'd pay just for SIP trunks or PSTN lines in a long-term on-premise setup. So, in many situations, moving from on-premise to cloud-based solutions like Amazon Connect can absolutely provide a good ROI.

Amazon Connect has a pay-as-you-go model. But I have a slightly different perspective on it.

When an on-premise setup migrates to Amazon Connect, it's possible for a 2,000-seat call center to have a lower Amazon Connect bill than a 300-seat one. Why? It's not just about the pay-as-you-go model.

Call volume matters, especially outside the US. But most importantly, implementation is crucial. A 300-seat center with heavy integrations across various AWS services, third-party tools, and potentially misconfigured free features can rack up unexpected costs due to a lack of transparency. Bills can be complex to analyze.

On the other hand, a 2,000-seat center with simple use cases (like backend authentication, language selection, and direct agent routing) and a separately purchased CRM integration through the marketplace might have a much lower bill.

Ultimately, it depends heavily on how you plan and which features are essential. Extensive use of Lambdas, customer profiles, and other services will increase the cost.

At a high level, the pay-as-you-go model is attractive— no agent licenses and scalability is simple. But actual costs are tied to your specific use cases. The more integrations you have, the higher the potential Amazon bill. The various services used, not just those within the 'Connect' category, factor into the cost. Even things like exceeding queue limits add up.

My advice is to thoroughly calculate the cost before introducing any feature. If cost is a major concern, consider alternatives.

When Amazon Connect was initially released, it was quite limited. Now, it's grown with better features and documentation. For me, setting up a basic system is straightforward. For instance, I configured a small 15-seat call center in just 20 minutes. Naturally, complexity increases along with features and requirements.

My suggestion for anyone considering Amazon Connect is to start with a strong understanding of basic telephony and contact center terminology. This will make it much easier to grasp the concepts.

Amazon provides a lot of documentation. I recommend reviewing that thoroughly before diving into the system itself. Understanding the 'why' behind Amazon Connect will make the hands-on experience much smoother.

Overall, I would rate the solution a seven out of ten. In my experience, there's still a lot of room for development and expansion to support even more use cases.

Amazon Connect is a totally cloud-based contact center technology.

Amazon Connect is similar, but it's a totally cloud-based contact center solution like Cisco, Avaya, or Genesys.

I primarily use it for high-level contact center services like DynamoDB, Kinesis, Lambda functions, and Amazon Connect. But Amazon Connect continuously introduces new features. In 2023 alone, they released over 200 new features for Connect.

There's no maintenance cost. It's totally managed by Amazon. We only have to pay for the services we actually use.

Also, scalability and reliability are the best features of this tool.

It's a reliable and scalable tool. No maintenance and pay-as-you-go are the most important features.

Moreover, it is easy for a beginner to learn to use Amazon Connect for the first time.

With proper training, an agent can learn the basics in one or two days, maximum one week.

Earlier, we had to integrate Salesforce with Amazon Connect for any contact center to support customer case details and other information. We needed to integrate Salesforce as the CRM tool.

But Amazon recently released a new feature called 'Customer Profiles.' It's intended to be similar to Salesforce. However, feature-wise, it isn't equivalent to Salesforce yet. Amazon is continuously improving Customer Profiles and similar features.

I have been using it for six to seven years now.

It's very stable and reliable. In the last eight or ten years, there were only two or three breakdowns. One lasted for four or five minutes and another for ten to fifteen minutes.

It has been a very reliable solution for us.

It is a scalable product. We have multiple contact centers across the company. Globally, we support Europe, the UK, the US, and Asia-Pacific regions like Singapore, Malaysia, the Philippines, Australia, and India.

So, the number of agents we have using Amazon Connect is more than 5,000. We're an enterprise customer of Amazon.

The quality of customer service and support depends on the specific case.

We also have Genesys.

The initial installation and setup were easy.

You need to set up your infrastructure first - your network connections, security, etc. If you're using an internal network, consider an EDI VPN. Otherwise, you can create an Amazon Connect instance and start using it directly.

It will be worth the investment. I'd recommend it, especially if your employees are located globally or use a hybrid work model. It's very easy to use. Agents can log in and connect with customers from anywhere.

It provides a good overall experience.

There are costs involved, like $0.18 per minute for incoming calls.

The licensing is a pay-as-you-go model.

My recommendation depends on your exact requirements. It involves the following aspects:

• What's the size of your operation?
• What specific purposes do you have in mind?
• Could you elaborate on how you intend to use it?

For example, if I have a call center with 30 people, and I primarily want to use it for call center operations. I wouldn't recommend Avaya or Cisco. I'd suggest considering either Amazon Connect or Salesforce Service Cloud (voice services).

For Amazon Connect, I'd rate it a seven out of ten. There are a few features they still need to improve, particularly with Customer Profiles and case management. If these features become as robust as Salesforce, then Amazon Connect could be a complete contact center solution without the need to integrate third-party CRM tools like Salesforce.