[Un]Churned · 15 Apr 2026 · From the week of 13 April
Inside Google's AI-First Post-Sales Playbook ft. Brady Bluhm (Gainsight) & Diane Wu (Google)
These are notes on the conversation, checked against its transcript. The episode itself has the full discussion.
In brief
Diane Wu, Global Head of Customer Success & Experience at Google Cloud Security, and Brady Bluhm, a Gainsight senior product manager and former CSM, discuss how AI changes the customer success role. Diane describes her team's use of call transcripts, Gemini notes and NotebookLM, and the adoption resistance from her strongest CSMs. Brady describes his own AI workflow, including agents, meeting transcripts as memory, and using the Gainsight MCP to close old CTAs. The central argument is that once knowledge is no longer scarce, the CSM's edge becomes curating context for each customer, with human judgment kept in the loop.
For founders
- Brady Bluhm says that within about two years he expects fewer people to need to log into his product's UI as LLMs evolve, which has made him question building UX now, though he still does because not everyone works in LLMs all the time.
- Brady Bluhm says that after using the Staircase MCP for a day or two, he questioned why users would keep logging into his product, and after testing the Gainsight MCP he now sees Gainsight's structure as operational bones that an LLM can work on.
- Diane Wu describes two phases of AI adoption: first getting people to prompt and validate LLM output, then using an MCP so one natural-language interface can replace multiple tools.
- Diane Wu says her team measures post-sales success by how well it delivers hyper-personalized value across customers using 8-9 products, which she says is hard to do when the portfolio is large.
For revenue leaders
- Diane Wu says her best CSMs were harder to get onto NotebookLM because they had built their own playbooks, whereas a new CSM with no playbook would likely find it immediately useful, so their adoption curve is less steep.
- Diane Wu says her team's easiest adoption came from putting Gemini notes and transcripts into NotebookLM, while advanced uses such as success plan templates and EBR slides adopted much more slowly.
- Diane Wu describes AI changing coverage models, where a high-touch enterprise CSM has traditionally covered about 10 to 20 accounts, and she expects AI to change how many accounts one person can cover.
- Brady Bluhm says new CSMs can ramp much faster with AI because knowledge is available on demand, where he took about 6 months to feel he knew the role and 12 months to feel good at it.
- Diane Wu says her team turns on transcripts and Gemini notes for every customer call where the customer permits it, and that customers have embraced this because it consolidates their context.
What was said 19, most useful first
Time saved is the wrong measure for him, because faster output has led to more tasks being assigned to him. Listen
Brady Bluhm says AI saves him a lot of time on each task, but that he now accepts more work because he can complete it quickly. He gives an example of a task that used to take 4 to 5 hours now taking about 15 minutes. He says he is working harder than ever, partly because the pace of the industry has increased.
“it's going to take me 15 minutes to do it where it used to take me 4 to 5 hours to do some of these tasks”
Listen to the episode Leadership & culture Link to this Report a problem
Diane Wu's team keeps one NotebookLM notebook per customer, fed by call transcripts, slides, office hour notes and action items, and shares these with product and engineering. Listen
Diane Wu says every customer should have a NotebookLM notebook with presentations, call transcripts, technical office hour notes and action items as source files. She says a script can keep these updated, which turns a physical notebook into a dynamic record of every conversation. She says CSMs share these notebooks with product managers and engineers when translating customer needs or troubleshooting bugs, which she says has an outsized impact.
“every customer you should have a notebook LM around.”
Listen to the episode Retention & customer success Link to this Report a problem
Diane Wu runs three to five parallel customer data analyses in Gemini, which she says takes about a quarter of the time of doing them one by one. Listen
Diane Wu describes pulling up one customer's dashboards and spreadsheets and spending hours on cross-dataset analysis. She now uses one Gemini chat per customer with the same prompt structure and runs three to five analyses in parallel. She says the output still needs tweaking, and that the time spent shifts from analysis to creating insights faster.
“That takes me 1/4 of the time, and obviously I have to go back and tweak it.”
Diane Wu found that her most productive CSMs resisted new AI tools more than she expected, because they had built their own playbooks. Listen
Diane Wu says she initially assumed her top CSMs would adopt NotebookLM first, but found it was almost the opposite. She says her best CSMs have built institutional playbooks, spreadsheets and tools that work well for them, so changing those habits is difficult. She suggests a brand new CSM without an existing playbook would find the tool immediately useful, so their adoption curve is less steep.
“your most productive and best CS NS have built and curated their own internal playbook of how to drive operational success that works for them”
Listen to the episode Leadership & culture Link to this Report a problem
AI adoption is change management, and it should be driven by small wow moments rather than large changes to existing behaviour. Listen
Diane Wu says adoption depends less on any individual's technical skill than on how much of their existing work will change, and how to move them across that gap. She says the team's easiest adoption came from using transcripts and Gemini notes in NotebookLM for conversation history. She says advanced uses such as success plan templates and EBR slides disrupt more institutionalized behaviours and have seen less adoption.
“you have to do it with small wow moments”
Listen to the episode Leadership & culture Link to this Report a problem
Brady Bluhm predicts that in about two years people will not need to log into product UIs, because LLMs are improving faster than other software. Listen
Brady Bluhm says he had a product manager's existential moment after playing with Staircase's MCP and asking why anyone would log into his product in a year and a half or two years. He expects that change because LLMs themselves are improving faster than other software. He still builds for the UI today because not everyone works in LLMs all the time.
“in two years from now, I think that's going to change because the LLMS themselves evolved so much faster than any other product will too”
The traditional application UI will eventually fade, with LLMs becoming the main workspace regardless of the back-end application. Listen
Diane Wu says that in the old world CSMs used multiple tabs and tools to get the context they need, and she expects this to go away. She describes the LLM as the new workspace, with the interface being natural language rather than visual, at least today. She adds that this might change very quickly in the future.
“I'm a believer of the traditional application UI is going to eventually fade and go away and LLMS will become the new workspace”
Brady Bluhm used the Gainsight MCP to find and close five CTAs he had opened in Gainsight years earlier, with a single prompt. Listen
Brady Bluhm says that while testing the Gainsight CS MCP, his LLM found five CTAs he had opened as a CSM a couple of years earlier, and he asked it to close them. He says the MCP can also write to the timeline, and he sees CTA fatigue, a known pain point in Gainsight, as something the LLM may help solve. He describes this as an example of how software will change.
“I found five Ctas that were still open that I had opened in gain site a couple of years ago.”
Listen to the episode Retention & customer success Link to this Report a problem
Once knowledge is no longer gated by access, the CSM's differentiator becomes curating the context of a specific customer's use case. Listen
Diane Wu says AI has made access to knowledge less of an advantage, so a CSM no longer adds value by having best-practice guides that customers lack. She describes the differentiator as how the CSM curates context for that customer's use case to add operational value. Host Josh Schachter ties this to Gainsight's theme that CS is moving from software as a service to retention as a service.
“It's really about how do you curate the context associated with that customer's specific use case to add that differentiated operational value.”
Listen to the episode Retention & customer success Link to this Report a problem
AI can shorten how long it takes new CSMs to ramp because product and customer knowledge is available on demand. Listen
Brady Bluhm says that as a CSM he took about 6 months to feel he knew what he was doing and 12 months to feel good at the role, because he had to collect knowledge himself. He says enablement of new CSMs can now be much faster because answers are at their fingertips and they do not need to know whom to ask internally. He notes this also changes internal team and project management.
“I think like enablement of new CSMS can be way faster because that knowledge is at their fingertips.”
Listen to the episode Hiring & team building Link to this Report a problem
Meeting transcripts can serve as a personal memory, and an AI assistant can search them for who discussed a topic and pull exact quotes. Listen
Brady Bluhm says he has a poor memory and now uses meeting transcripts as a complete record of meetings he has attended. He says he can ask the AI who he spoke with about a given topic and it will find the conversation and return the exact quotes. He describes this as changing how he works as both a product manager and a former CSM.
“meeting transcripts are my memory.”
Listen to the episode Retention & customer success Link to this Report a problem
Brady Bluhm runs several AI agent chats in parallel, assigning one to research context while he works in another. Listen
Brady Bluhm describes his day as juggling agents across different chat windows. He sets one agent to work on context and research, then moves to another chat to work on his next task. He says this is still task and project management, but at a different scope and speed.
“I'm juggling agents. A lot of the time agents serve chats at least right in different chats.”
AI changes the number of accounts one CSM can cover, and she gives the traditional enterprise high-touch ratio as about 1 to 10 or 1 to 20. Listen
Diane Wu says CS leaders traditionally plan coverage with a tiered model: enterprise high-touch CSMs, mid-tier, and digital. She says a high-touch CSM has typically covered about 10 accounts, or about 20 in a highly productive model. She expects AI, and later agents, to change how many accounts a human can cover.
“you have your sort of enterprise level high touch CSM at a one to 10 or maybe 1 to 20 if you're a highly productive model.”
Listen to the episode Retention & customer success Link to this Report a problem
Google's CS team turns on transcripts and Gemini notes for every call where the customer agrees, and Diane Wu says customers have embraced it. Listen
Diane Wu describes this as the easiest, low-hanging fruit her team has done in the last six months to a year. She says customers recognize that transcripts and action items let the team consolidate their context and needs more effectively. She adds that her team respects customers who do not want to be recorded.
“turn on transcripts and recording and Gemini notes for every call if the customer permits us”
Listen to the episode Retention & customer success Link to this Report a problem
AI output should never be taken at face value, and that the human in the loop provides curation and context. Listen
Diane Wu says AI is a strong source of raw information and analysis but needs a CSM or CS leader with curiosity and technical fluency to prompt it well and ask the right discovery questions. She says this human judgment is what turns output into differentiated value. She frames the human role as curating and adding context, not replacing the CSM.
“when you use an AI tool, never ever take it for face value, right?”
Brady Bluhm asks whether a task can be done with AI, then decides whether to turn the working output into a repeatable agent or skill. Listen
Brady Bluhm says that whenever he does any task, he first asks whether it can be done with AI. After getting a usable output, which he says takes curating context and multiple rounds of feedback, he asks whether he will need to do the task again. If so, he considers training an agent or skill so the next run is faster.
“I ask, can I be doing this with AI?”
Brady Bluhm spends at least a few hours each week during work days tuning his AI setup while doing his work. Listen
Brady Bluhm says he tunes his AI setup as he works whenever the current mechanism is not getting him to the result quickly enough, such as adjusting his folder hierarchy. He says he now works in a command-line tool because it unlocks more for him, and that the chat interfaces are moving in the same direction. He also says he builds AI tools on nights and weekends, mostly for work.
“there's at least a few hours every week that I'm spending during my work days, like tuning some”
Diane Wu describes two phases of AI adoption: learning to prompt and validate LLM output first, then using an MCP to connect multiple tools behind one interface. Listen
Diane Wu says the first phase is getting people used to working in an LLM, including asking intelligent prompts and validating answers. The second phase is understanding what an MCP is and how it unlocks multiple tool sets, so a user needs one interface rather than ten open tabs. She describes the MCP as enabling context enrichment across data sets stored in different places.
“I think there's the two phase adoption of”
Diane Wu measures post-sales success by whether it delivers hyper-personalized value to each customer across a growing product portfolio. Listen
Diane Wu says she thinks of customer success not as just an organization but as how her team drives success for customers more effectively, accurately and faster. She describes customers using 8-9 products across the Google Cloud security suite, with Wiz added through acquisition, and says the challenge is delivering value that is not diluted by the number of products. She hopes tools such as Gainsight, MCP and AI will allow this personalization to happen at greater scale.
“how do we start building hyper personalization for our customer needs”
Listen to the episode Retention & customer success Link to this Report a problem