There’s a lot of noise right now about AI “transforming customer success,” but most of what I see online feels abstract. The reality in day-to-day work is much more practical…and honestly, more interesting.
As someone actively working in Customer Success, I’ve been spending a lot of time figuring out what actually helps versus what just adds another layer of complexity. Two things have stood out recently: how we’re using data more deeply through tools like Hex, and how quickly AI agents are starting to shift internal workflows.
Beyond my own experiments, I’ve been paying attention to how other customer success managers are using AI in practice. The patterns are surprisingly consistent:
1. Reducing admin overload. The biggest immediate win is time. Call notes, follow-ups, CRM updates, and tasks that used to quietly eat up hours each week are being partially automated or at least streamlined. That time goes back into actual customer engagement.
2. Improving the quality of conversations. AI isn’t just speeding things up; it’s improving preparation. Instead of walking into calls with generic account summaries, CSMs are walking in with clearer insight into what actually matters to the customer, how they’re tracking against it, and where opportunities exist.
3. Surfacing risk and opportunity earlier. Signals that used to be buried, like subtle usage drops or changes in product behavior are now easier to catch. That means fewer surprises and more proactive intervention.
Talking about signals and indicators, have you tried out Hex Threads?
My go-to for AI data has been HEX. One of the most useful changes in my workflow lately has been using Hex Threads. If your company already uses Hex and you’re in a CSM role but don’t yet have access, it’s worth asking for it directly. It’s one of those things that quietly unlocks a lot more context than you realize you were missing.
Having the ability to drill into account-level and user-level behavior changes the quality of your conversations in a very real way. Another thing I’ve learned: there are often advanced permissions in Hex that aren’t automatically exposed. If you’re in CS and only seeing surface-level data, ask about deeper access!
You can also connect Hex to Claude and start building your client facing QBRs.
Experimenting with AI agents for smoother handoffs
Lately, I’ve been experimenting with Claude to build lightweight agents that help with internal handoffs, especially from Sales to CS and AE teams. One early use case I tested was creating a simple agent that summarizes sales context into something a CSM or AE can actually use. Instead of digging through long email threads or scattered notes, the agent pulls together key details like:
- Customer goals discussed during sales cycles
- Key stakeholders and their priorities
- Any early risk signals or expectations set pre-sale
It’s still early, but even in its basic form, it has already reduced the missing information layer that usually happens when accounts move from Sales into post-sale ownership
I’m still early in my own experimentation with AI agents in Claude Cowork, and I’m very curious how others are integrating these tools into their workflows.
Are you seeing similar shifts in your CS teams? What’s actually sticking versus what’s still just hype?