The AI That Sat In on Every Customer Call - and Learned the Business
The AI notetaker that turned every customer call into a searchable, actionable record - and quietly became the brain behind Gainsight's push into agentic AI.
Somewhere in your company's history is a customer who told you exactly why they were about to leave. They said it on a Tuesday call, between a status update and a scheduling question. The account manager nodded, meant to write it down, and then the next meeting started. The warning evaporated. Three months later the renewal quietly did not happen, and nobody could say precisely why.
UpdateAI was built around that exact moment of forgetting. Founded in Los Angeles around 2020 by Josh Schachter, the company made a conversational intelligence platform for the people who spend their days on customer calls after the sale closes - Customer Success Managers, account managers, and the revenue teams standing behind them. Its AI meeting assistant joins a Zoom, Microsoft Teams, or Google Meet call, records and transcribes it, and then does the part humans keep failing to do: it writes down what mattered, flags what needs a follow-up, and files it somewhere the whole team can find it later.
In July 2025, Gainsight - the company widely credited with naming and building out the Customer Success category - acquired UpdateAI and made its technology the foundation of a new agentic AI platform called Atlas. Schachter became SVP and General Manager of that effort. For a business that started as a note-taking bot, it was a notable place to land.
01 / The ProblemThe customer intelligence gap
Schachter has described Customer Success Managers as the heartbeat of SaaS: closest to the customer, tied directly to revenue outcomes, and yet routinely buried under manual work. A CSM might run six or eight customer calls in a day. Each one produces notes that should reach the CRM, the product team, and the account plan. Most of it never arrives intact.
UpdateAI gave that leak a name. It called it the "customer intelligence gap" - the distance between what customers actually say on calls and what the rest of the company ever learns about it. Sales tools had spent a decade recording and analyzing pipeline calls. Post-sales, where churn and expansion actually happen, was still running on memory and hastily typed notes.
02 / The ProductWhat the bot actually does
The workflow is deliberately unglamorous, which is the point. A CSM stays in the conversation while an assistant handles the record-keeping. The moment the call ends, the summary is already moving. On Zoom, UpdateAI prefers to use Zoom's own cloud recording when it can, and sends its own bot when it cannot - a small detail that tells you the product was built by people who had sat through the awkwardness of a mystery bot appearing in a customer meeting.
Bot joins Zoom, Teams or Meet
Records & transcribes the call
Smart Summary + action items
Sent to email & Slack
Notes sync to Salesforce
Once the recording exists, the AI goes to work. Smart Meeting Summaries condense an hour of conversation into the decisions, the open questions, and the next steps. The system detects follow-ups and drafts the emails to close them. Notes flow into Slack channels and into Salesforce, so the account record updates itself instead of waiting for someone to remember. When ChatGPT arrived, UpdateAI was among the first Customer Success tools to wire it in, using large language models to make those summaries read like something a person would actually send.
Over time the product climbed from single meetings to patterns across many. Reading signals across a whole book of business, it could surface which accounts sounded like risk, what product complaints kept recurring, and where an expansion conversation might be waiting. The individual summary was the hook. The aggregate intelligence was the reason to stay.
CSMs were the heartbeat of SaaS, closest to customers, directly tied to measurable business outcomes, yet undervalued and often overextended by tedious, manual tasks. Josh Schachter, Founder
03 / The CustomerBuilt for post-sales, not sales
This is the decision that shaped everything else. The obvious market for an AI notetaker was sales, where budgets are large and call recording was already normal. UpdateAI went the other way and built for post-sales. Its users were CSMs and account teams at early- and growth-stage B2B SaaS companies - the people responsible for keeping customers and growing them, not just closing them.
That focus changed what the product needed to notice. A sales tool wants to know if a deal will close. A Customer Success tool wants to know if a happy customer is about to become an unhappy one, whether a feature request is a passing comment or a dealbreaker, and which quiet account has stopped showing up. UpdateAI tuned its models for those signals rather than for deal scoring.
| Tool type | Primary user | Question it answers |
|---|---|---|
| UpdateAI | Customer Success / post-sales | Is this customer at risk, and what did we promise? |
| Sales intelligence | Account executives | Will this deal close this quarter? |
| Generic notetakers | Anyone on a call | What was said, verbatim? |
04 / The StrategyThe podcast was the marketing
UpdateAI did something most software companies talk about and few pull off. It ran the audience it was selling to. Its podcast, Unchurned, positioned itself as the number one show for Customer Success, featuring CS leaders and operators from across the industry. The product was built for CSMs; the podcast was made for CSMs; the same people showed up in both places.
That is a distribution engine disguised as content. Instead of buying attention from the Customer Success community, UpdateAI became a fixture inside it. When Gainsight acquired the company, it acquired Unchurned too. The show was part of the asset, not a side project.
The best proof that you understand a market is that the market wants to hear you talk about itself.
05 / The MoneyLean by design
UpdateAI raised roughly $4 million in total. An early round of about $1.7 million got the product off the ground, and a $2.3 million seed followed in May 2023, led by IdealabX with Stage Venture Partners, Zoom Ventures, a16z's scout fund, and Howard Morgan participating. It was also the inaugural investment of CS Angel, a syndicate of Customer Success leaders backing tools for their own field.
The Zoom Ventures check is worth pausing on. UpdateAI did not just integrate with Zoom; the platform it depended on invested in it. That is the kind of alignment that is hard to manufacture and useful to have when your bot needs to sit inside someone else's meeting.
The business ran on a familiar B2B SaaS shape: per-seat subscriptions with a low-friction entry point and paid team plans, sold to CS organizations and spread through integrations with Zoom, Teams, Meet, Slack, and Salesforce. The team stayed small, in the range of 11 to 50 people, distributed across the United States, Argentina, and Ukraine. Co-founders Richard Ferraris and Sergey Kondratiuk led engineering and data science.
06 / The ExitFrom summaries to agents
On July 21, 2025, Gainsight announced it had acquired UpdateAI. The logic was less about buying a notetaker and more about buying a capability. Gainsight was building Atlas, a system of agentic AI - software meant to take action across the customer journey rather than wait for a prompt. Agents like that need fuel, and the best fuel is a steady read of what customers are actually saying. UpdateAI had spent five years building exactly that read.
So the summary engine became the sensing layer for something larger. What UpdateAI did quietly in the background of a call - notice a risk, catch a commitment, log a signal - is the raw material an autonomous agent needs to act on a customer's behalf. Schachter framed the move as a chance to amplify the work alongside people who had spent careers on Customer Success.
Joining Gainsight means we get to amplify that impact, working alongside some of the best minds in CS to build the next generation of AI agents. Josh Schachter, on the acquisition
07 / The TakeawayWhere it fits
UpdateAI competed in a crowded field of conversation tools - Gong and Chorus on the sales side, Fireflies, Otter, Fathom, Avoma, and Grain among the notetakers. Its edge was not a better transcript. It was a narrower question. By building for the specific worries of Customer Success rather than the general act of recording a meeting, it made itself the tool that answered the one thing post-sales teams needed to know: which customer is drifting, and what did we say we would do about it.
The through-line from a 2020 startup to a 2025 acquisition is consistency. Same customer, same problem, same insistence that the words spoken on a call are data worth keeping. The product got smarter and the ambition got larger, but the bet never changed. Turns out that the least glamorous job in software - remembering what the customer said - was a good place to build a company.