←
Advith Chelikani965 posts
Advith Chelikani avatar

Advith Chelikani

@advith_c

cofounder @usepylon, building the full stack for customer work

◷ Joined October 2023▧ 398 Photos and videos
Advith Chelikani avatar
Advith Chelikani@advith_c·Jul 15

Today, Pylon becomes the Agentic Support Platform. We’ve spent the last several years preparing for when models were good enough to build the workflows we always imagined and now we’re launching a set of new primitives that we think will change the way you do customer work. With Background Agents you’ll be able to work on your whole queue at once and show up to each customer issue with an investigation already complete. Combined with Skills, you’ll be able to build and deploy your own custom workflows to reduce tedious work: file feature requests automatically, update your knowledge base when new features are shipped, and escalate to engineering with full context. With Assist Agent you’ll be able to go into interactive mode, pushing an investigation forward, digging into the logs and codebase, and learning automatically from past investigations. With Slack Agent you can give customer insights to your whole org - anyone in Slack can get their questions about your product or your customers answered all from Pylon’s growing knowledge of how you operate. We’ve been lucky to work with some great customers as part of our initial beta, including Mike from Cognition, Bre from Hex, and Regie from Goodtime and we’re excited to expand our beta to even more existing and future Pylon customers :) See how it all fits together on July 20. Register here: bit.ly/4pkHrqG This is just the start of many powerful new features that will help you operate at lightspeed.

1:00
Cognition avatar
⟳ Advith Chelikani retweeted
Cognition@cognition·20h

Meet Devin Bots. From writing firmware to merging thousands of PRs a week, a Devin Bot leads a team of agents to get the job done.

1:33
Advith Chelikani avatar
Advith Chelikani@advith_c·Oct 13

Excited to partner with the @cognition team to launch a native integration for our customers. The future of customer work is agents cooperating through the whole stack and we're excited to take a step toward that vision. @usepylon's agents triage and investigate customer requests - when a bug is identified that needs a product change, it hands them to Devin with context to put up a code change. Once the change is shipped, progress flows back into Pylon and we can close the loop with the customer. For any Pylon customer already using Devin - codebase search through Pylon is free forever AND those who set up the integration get 5k in Devin credits. For new Devin users who set up the integration, you get 28 free seats in year one, 15K in credits, and 50K in matched spend on the startup teams plan. We’re walking through it live on Oct. 27 at 10:30 am PT: ordnl.link/5TuNEWi

Post attachment
Advith Chelikani avatar
Advith Chelikani@advith_c·Sep 15

top 1% spend is huge and dominating by coding agents, median spend is $12 and has a lot of room to grow as most knowledge work is not using much AI yet (under .1% salary). that’s still a fraction of the cost of a software seat and ramp data is skewed to early AI adopters. you can safely expect a 10x+ increase in AI spend per employee at the median company over the next few years as there is more diffusion of agents into the economy and more app companies successfully building the full stack for their domains if you’re building application layer AI, the market is still very early

Post attachment
Advith Chelikani avatar
Advith Chelikani@advith_c·Sep 13

the future of app layer companies is owning the record, owning the harness, and eventually owning the model

I was just at YC demo day yesterday. Besides hardware/physical things, everyone is just basically just building a domain-specific harness

Post attachment
Advith Chelikani avatar
Advith Chelikani@advith_c·Sep 9

One of the more useful changes we’ve made internally is giving our support team a better way to investigate product issues before they become an engineering interruption. Before Agentic Support, answering a customer question often meant support or an engineer starting from scratch: finding the account, checking logs, searching past issues, and figuring out whether the behavior was expected, a known bug, or something new. Now, much of that work starts as soon as an issue arrives. Background Agents gather relevant context, and support can use Assist Agent to search logs, the codebase, other tools like Linear or Notion, and dig deeper without leaving the issue. When engineering gets involved, the account context, relevant logs, and investigation are already there. In the four weeks after rollout, support ran 773 codebase searches. Questions in our internal helpdesk Slack channel dropped by about 50% compared with the four weeks before. We’ve been running this inside Pylon as we build Agentic Support. The full write-up, including the response-time, adoption, and capacity data, is here: ordnl.link/XAJW45o

Post attachment
Robert Eng avatar
⟳ Advith Chelikani retweeted
Robert Eng@rengrenghello·Jul 15

Today we’re launching the new Pylon, the first platform built for Agentic Customer Support. Earlier this year, we had a thesis: most AI support products were optimized for deflecting simple questions, but those questions represent only a small part of the work support teams do. The rest is investigation, coordination, judgment, and action: figuring out what happened, who the customer is, whether this happened before, what other teams need to do, and what should happen next. The limiting factor is not the model. It is whether the agent has enough context to do the work. It needs the customer’s history, account context, past issues, internal conversations, product data, logs, Linear tickets, and sometimes the codebase. It also needs to understand how your team handles that situation. So we built agents on top of Pylon’s context layer: support, account, and product intelligence assembled from every system your team uses. Then we gave those agents the Skills and tools to investigate and act. Here’s how it works: When a ticket arrives, Background Agents begin an investigation before anyone opens it. When a human steps in, Assist Agent lets them review the evidence, ask follow-up questions, send the agent into another system, and approve what happens next. While one agent is working, the human moves to another issue. The team stops processing tickets one at a time and starts directing several streams of work at once. We deliberately have a human in the loop. The team can inspect the work, correct it, and build confidence before allowing an agent to act on its own. Skills capture how your team works, so the same processes can run consistently and become more autonomous as confidence grows. We call this new way of operating "Agentic Support": agents do the work, humans direct it. Cognition, Hex, Canvas Medical, Nominal, and other Pylon customers have been helping us test this model in early access. Today, we’re opening it to more teams. The new Pylon is available in beta now. See how it works live: bit.ly/44X1qCi

1:00
Marty Kausas avatar
⟳ Advith Chelikani retweeted
Marty Kausas@marty_kausas·Jul 15

Today we’re launching the new Pylon, the first platform built for Agentic Customer Support. It’s available in beta now.

1:00
Advith Chelikani avatar
Advith Chelikani@advith_c·Jun 5

The role of a support engineer is evolving. It used to look like... → Long, manual issue investigations → Lots of back and forth with customers to get context → Ping an engineer for help with a ticket and wait Now teams can... → Have AI run initial investigations → Ask AI to gather account context → Search the codebase directly with Claude Code or Cursor The goals of support haven't changed (provide fast and high-quality help, make it easy for customers to reach you) but the tools are making it so much easier to achieve those outcomes now. Next week we'll show you how we've been using AI tools for support, product, and GTM. Robert and Nikolaj Pedersen from our support team will demo three AI workflows we actually use, including projects from our internal hackathon last month. Join us June 10 @ 10 AM PT. Link to register in the comments 👇️

Post attachment
Advith Chelikani avatar
Advith Chelikani@advith_c·Jun 3

New workflow: Send and set status ⚡ If you ever want to manually override AI auto-statuses in Pylon, you can now respond to an issue and set its custom status in the same action. → Draft your customer reply → Use the "Send" dropdown to set issue status on send → Use keyboard shortcuts to do it even faster Make sure tickets land exactly where you want them.

0:05
Advith Chelikani avatar
Advith Chelikani@advith_c·Jun 1

Here's how our first ever AI support hackathon went 👇️ What we learned from the event... 1/ For a lot of people, networking and just getting to chat with other attendees was super valuable in itself 2/ Our current customers got a ton of value out of meeting their Pylon CSMs in person (even if it was just a 15-30 min conversation in the middle of the day) 3/ It was awesome to see folks get excited about beta products we announced, but many people wanted deeper dives on use cases and how to set things up We had a great time and got such good feedback that we're planning a similar series of customer-focused workshops for this summer. Some of what we'll add based on your input... → Dedicated sessions with your Pylon CSMs → Event tracks so attendees can go deep on specific Pylon products or workflows they want to learn about → Small group sessions to facilitate different teams meeting each other and sharing their work It's going be a very fun series so we can't wait to get more of the Pylon team and our customers involved. Huge thank you to everyone who came out on May 21. It's still mind-boggling to us that people flew in from as far as the East Coast and the UK to be there! See you again at the next one 🫡

0:50
Advith Chelikani avatar
Advith Chelikani@advith_c·May 28

Usually the logic is that customer onboarding gets worse as you scale. Tighter CSM bandwidth → less personalized setup But our scaled CSM Donal Woods (who manages hundreds of accounts himself) knew that shouldn't be true anymore with AI tools and good account context.... ...so in the past few months he built a Pylonboarding experience that outputs personalized + interactive onboarding plans for every customer. This is how the flow works: 1. New customers take a 90-second quiz on their setup status, pain points, and product needs 2. Depending on their inputs, customers automatically get placed into 3 product tracks: Foundation, Growth, or Scale 3. They get a personalized email with track-specific next steps 4. Pylon auto-creates a project for the customer with track-specific tasks 5. Customers get directed to interactive product walkthroughs so they can start exploring Pylon on their own Built with Claude, Pylon, Typeform, Customer.io, and Arcade. Huge shoutout to Dónal for helping scale Pylonboarding while improving the experience for all our new customers. If you're curious to check out one of the interactive product pages: pylonboarding.usepylon.com/a…

0:21
Advith Chelikani avatar
Advith Chelikani@advith_c·May 27

Teamwork makes the dream work... ...so we're making it easier to collaborate on tickets. You can now see who else is viewing an issue directly from your list and kanban views. Now it's easier to work through your team's queue together in real-time and there's no more double-handling. Happy collaboration 🤝

Post attachment
Advith Chelikani avatar
Advith Chelikani@advith_c·May 26

The AI tools we ship at Pylon come from real needs we've noticed in our own customer work... ...which means we really do use all the agentic workflows and features we've built into our product. How? In a few weeks @rengrenghello and Nikolaj will show you. They'll demo 3 AI workflows our support, product, and GTM teams use every single day: → Tech stack we used → Patterns we applied → Prompts we wrote → Improvements we made → Outcomes we got And you can expect to leave with workflow patterns to copy and tweak for your own team. 📆 June 10 ⏰️ 10 AM PT Register here for a spot: bit.ly/3Rh0e9t

Post attachment