"SPEAK TO A REPRESENTATIVE - SPEAK TO A REPRESENTATIVE!"
We've all had the frustrating experience of getting lost in the maze of automated customer service phone menus. They ask you to "describe your issue", and then try and shove your call into one of several generic buckets. Have a more nuanced question/concern? Good luck with that.
Once you have managed to fight through an endless series of menu options, you're met with temporary relief when the robotic phone operator says they'll be glad to connect you to someone to help.
The wait should be only 45 minutes. But hey, at least you can listen to their elevator music loop.
These customer support phone systems have been in use since the 1970s, and it's safe to say that while they have improved, no one would prefer an automated menu system to a real human when they have an issue.
Everyone has heard the conventional wisdom that artificial intelligence revolution will change how corporations operate. But what form that change will take in specific industries is still unclear. . With ever-expanding computational capacity and language models that can process massive datasets, it can be an overwhelming proposition to begin re-imagining your department's workflows.
Today we're going to discuss how automated customer service teams can leverage AI to transform how companies operate. Here's what we'll cover:
- Why your customers will prefer an AI agent to a human
- Phone agent infrastructure (what's needed to handle enterprise call-center level volume)
- The most common business use cases for AI phone agents
What are Artificial Intelligence Phone Agents
An "Agent" is a system that follows a multi-step structure in order to achieve a goal. They are autonomous in nature, meaning that they are designed to make their own decisions in pursuit of the defined goal. Agents can be built to with infinite levels of complexity and are being used to solve some of the world's most complex problems.
A "phone agent" is an AI-powered agent that can speak on the phone. Think of it as a robot that can be given goals (prompts) to achieve during a phone conversation.
These are not the dreary voice menus used by the support lines of big companies. They sound and respond like real people.
Here's a quick clip of an AI agent on the phone talking with a member of our team.
How Phone Agents Work
Phone agents powered by AI are the product of a combination of technologies. They must process and understand speech (speech recognition), they must make decisions based on prompts (large language models), and then they must speak (text to voice).
These technologies are not new. But recent advancements in all three have enabled engineers (like the Bland.ai team) to build lifelike phone agents that can operate in real time. A typical call process would look like this on the backend:
- The customer says something or asks a question
- The text goes through a large language model (LLM),accompanied by a "prompt" to tell the model what to do
- The LLM decides what the agent should do next (e.g., make an appointment in the system, transfer the call, or reply back with speech)
- The agent’s response is translated from text to speech
- The agent responds to the customer
The conversation will continue based on scripts that are defined for the AI phone agent to follow until the natural end of the call.
How AI Phone Agents Use Scripts
In a typical call center setting, scripts and flow charts are written for operators to direct them on what to say in all possible scenarios. These training documents can be incredibly long and complicated, depending on the specificity of the tasks and how much "freedom" the operator is granted.
Training a human agent takes a significant amount of time and resources. Furthermore, after the training period, it is necessary to monitor and review calls to ensure they adhere to their scripts. This process must be repeated with every new hire.
AI phone agents only need to be trained once. Instead of messy flowcharts and endless written documents, you can build elegant call flows in the Bland web dashboard (called Conversational Pathways).
A script for a phone agent is made up of "nodes," which are the smallest unit of decision for a phone agent. A node has a goal (prompt) and has to meet certain conditions to move on to the next node(s). You can watch the call proceed in real time and see how your phone agent decides to use the call script to navigate the nodes.
Other Capabilities of AI Phone Agents
External Service Access Via APIs
Agents have the ability to access any external system via structured API requests. This could be used to search the internet for relevant information, record booked appointments in a shared company calendar, or update a CRM with notes and action items.
Knowledge Base Search and Respond
Bland agents are able to access, search, understand, and reply based off of shared knowledge base information. Need to train your agents to handle a new product line or a change in company policy? It can be as simple as updating the central knowledge base that your phone agents will have access to during their calls.
Benefits of AI Phone Agents
Imagine a call center team that:
- Is available 24/7
- Can answer and conduct thousands of calls simultaneously
- Never deviates from your defined service processes
- Costs a fraction of what your current call center teams are charging
At Bland we believe that AI phone agents are the answer to the challenge of reducing support team costs while increasing customer satisfaction.
You can have an army of highly-trained and cost-effective phone operators ready to scale up at a moment's notice to service your customer requests.
Like any automated customer service tool for support teams, there are challenges and important trade-offs to be considered. However, the trade-offs are not quite what you may be expecting. In the next section, we'll walk through some real use cases to illustrate this.
Use Cases for AI Phone Agents
A. Automate Customer Service
Tier 1 customer support is often the biggest initial impact for organizations starting to leverage AI phone agents. Agents can be built from pre-existing call scripts and trained using successful call transcripts from human operators.
Phone Agents as Automated Customer Service Tools
Call centers are among the most signficant support-related expenses for companies that deal with a large number of customers. As many executives have found out over the last decade, poor quality support calls lead to even more customer complaints and, ultimately, shrinking customer bases.
And yet, with increasing pressure to slash budgets, most have no choice but to pursue cheaper and cheaper labor to fill their call centers.
Up until recently, automated customer service experiences were a slippery slope to reduced customer satisfaction. But with l the latest advances in generative AI and large language models customer requests can now be met easily with "machines".
B. Sales and Lead Qualification
Many industries struggle with calling and qualifying their leads. The larger the industry, the greater the manpower needed to personally qualify online leads. The result? Most sales funnels try to use email or text automation to determine which leads are worth pursuing via a phone conversation.
With AI phone agents, you can call every single lead directly. Time to first contact is everything-- but even more importantly is using your teams time as efficiently as possible.
Companies like Better.com are using Bland to make sure their sales teams are only talking to people who they can help. No leads misunderstanding their services, no one lost in the shuffle, no one wasting time.
C. Appointment Scheduling and Reminders
AI phone agents are infinitely scalable and customizable.
Have other ideas on how to put them into production? Probably doable. The team at Bland can help.
.png)