Register for our upcoming webinar → Introducing the Agentic Support Platform
Today, we’re introducing the new Pylon, the first platform built for Agentic Customer Support, available in beta.
For the past 50 years, the customer support role has not changed much.
An issue arrives. Someone reads it, searches for context, checks past conversations, investigates what happened, asks another team for help, categorizes the issue, drafts a reply, and updates several systems.
The tools have changed. The work has not.
That way of working is over.
──────────────────────────
Software engineering has already gone through this change.
A few years ago, engineers wrote nearly every line of code by hand. Today, many engineers describe what they want, direct agents to build it, review the result, and make the decisions that require their judgment. The engineer is still responsible for the outcome. But the way the work gets done is completely different.
Customer support is next. Agentic Customer Support changes the basic working model:
Before: you did the work.
Now: you direct agents to do the work for you.
Humans remain in control of the customer relationship, the judgment calls, and the final outcome. Agents take on the work around those decisions: gathering context, investigating issues, drafting responses, coordinating handoffs, updating systems, and following through.
This is not another feature inside the existing support workflow. It is a new workflow.
──────────────────────────
Most support issues arrive with very little useful information. A customer says something is broken. They may not know which part of the product is involved, what changed, or which details the support team needs. The person assigned to the issue starts from zero.
They search past tickets. Check the customer’s account. Look through Slack, CRM notes, product usage, logs, Linear, and the codebase. Ask an engineer whether the behavior is expected. Find out that the customer is onboarding, that they reported the same problem last month, or that the person writing in is the executive sponsor.
This work can take longer than writing the answer.
With Pylon, agents begin working as soon as the issue arrives. They gather the account context, check connected systems, search for related issues, investigate the product, and recommend what should happen next. When a human opens the issue, the investigation is already there.
The team can review the evidence, ask follow-up questions, and direct the agent to keep working:
Instead of spending the first part of every issue digging, the team begins with work already completed.
──────────────────────────
Traditional support tools organize work into a queue. Humans take issues one at a time and move each one forward manually. Agents do not need to work that way.
Background Agents can start from a trigger and work across the queue without waiting for someone to open each issue. One agent can investigate a technical problem while another checks the account history, another prepares an escalation, and another looks for related issues.
The work stays visible. Humans can see what each agent is doing, review its output, redirect it, or take over. The result is not simply faster ticket handling. It is a support team that can keep multiple streams of work moving at once.
──────────────────────────
Every support team has processes that live in someone’s head. The strongest technical support engineer knows which systems to check before escalating a bug. The support lead knows how to handle a high-value customer during onboarding. Someone else knows how to investigate a billing discrepancy or prepare a clean handoff to engineering.
Pylon turns those processes into Skills. A Skill is a reusable set of instructions written in natural language. You describe what the agent should investigate, which systems it should use, and what actions it can take. Agents can run those Skills automatically when the right issue arrives. Humans can also invoke the same Skill when they need it. The team defines the process once. Humans and agents use it everywhere. Over time, every investigation, decision, and action can improve, or suggest how to improve, the agents, Skills, workflows, and knowledge behind the next issue.
──────────────────────────
An agent cannot make a good decision if it only sees the customer’s latest message and a help center. B2B support requires a broader understanding.
Who is the customer? What are they trying to accomplish? Are they onboarding or renewing? Who is writing in? What have they reported before? How is the product configured? What changed recently? Is this an isolated issue or part of a larger pattern?
Pylon has spent the past three years connecting the systems where that information lives: support conversations, CRM records, calls, emails, product usage, internal conversations, issue trackers, logs, and codebases.
But connecting raw data is not enough. Pylon turns that data into three forms of usable context:
That context exists before an issue arrives. Agents use it throughout the investigation, and humans can access the same understanding when they need to make a decision. This is what allows an agent to recognize that a customer is still onboarding, that the same issue happened before, or that the person asking is the executive sponsor. Those details change the right response.
──────────────────────────
“But what about the existing AI support agents?”
Our approach is different.
AI support agents were built for deflection. They answer the easiest questions before a human sees them. That is useful, but those questions represent a small part of the work. The majority of support work remains manual: gathering context, investigating difficult issues, coordinating across teams, escalating, drafting replies, and taking action in other systems.
That is our focus.
The goal is not to deflect a few more tickets. It is to change what support teams do all day.
──────────────────────────
We gave 20 existing Pylon customers, including Cognition, Hex, Canvas Medical, and Nominal, early access to the agentic platform.
Across the early-access group:
Teams are using agents to investigate issues before escalation, make technical knowledge accessible outside engineering, answer internal questions in Slack, and turn their strongest operators’ processes into repeatable Skills.
The benefit is not only faster responses.
Engineering receives fewer escalations, and the escalations it does receive arrive with better evidence. Support teams handle growing volume without adding headcount at the same rate. Customer context becomes available to Success, Product, Sales, and Engineering without routing every question through the support team.
Support becomes a source of leverage for the rest of the company.
──────────────────────────
Agentic Customer Support does not remove humans from support. It changes where humans spend their time.
Less time searching through systems. Less time moving information between teams. Less time repeating the same investigation. More time applying judgment, communicating with customers, improving the product, and deciding what should happen next.
The support professional remains responsible for the outcome. They just stop doing every step by hand.
That is Agentic Customer Support.
The new Pylon is available in beta today.
──────────────────────────
Join our upcoming webinar here
Media Contact
Pylon Workforce Management is available now. See it in action with a live demo.
