Launched this week
Katalyst is the AI sales agent for teams on Salesforce. Hang up a call and it's already done: notes summarized, records created, fields updated, follow-up drafted, next step set. It runs 24/7, reading every call, email, and calendar, surfacing the right signals, prepping you for each meeting, and prompting you to act: follow up here, this one’s slipping. What's new: AI Resolution on every account, meeting recorder, hygiene scores, deal patterns. Built for enterprise sales teams.
Curious how it handles calls when reps have thick accents or talk over each other, does the summarization still hold up or does it need cleanup? Also wondering how it decides which fields to auto-update without butting into the rep's manual edits.
Katalyst
@cemrediuc On audio: holds up well on normal crosstalk, but I won't oversell it, tough audio or heavy accents can still need occasional cleanup, and since the summary shows its source you can catch anything that drifted. On fields: it never fights manual edits, you scope what it can touch and it runs in approve-mode early, so it fills gaps rather than overwriting what you've entered yourself.
How does Katalyst distinguish between meaningful sales signals and routine activity to ensure its recommendations improve deal outcomes instead of creating unnecessary noise for sales teams?
Katalyst
@satyam_raina1 It scores signals by how much they actually move a deal and how time-sensitive they are, so a new decision-maker or a slipping renewal surfaces while routine logging stays quiet. And since it runs on each rep's own pipeline, relevance is tuned to their deals, not a blanket rule.
Katalyst
@satyam_raina1 To add, the noise problem comes from tools that treat every event as equal, Katalyst scores by deal impact and urgency instead. A new decision maker or a slipping deal rises, routine activity stays quiet. And because it runs on your own pipeline, relevance is tuned to your deals rather than a one size threshold.
the meeting recorder plus automatic field updates is a really thoughtful combo — feels like they actually watched how reps work between calls instead of just bolting AI onto Salesforce.
Katalyst
@cihanlrgq a lot of hours watching where reps actually lose time between calls, not starting from the AI and looking for somewhere to put it. Glad that comes through.
Katalyst
@cihanlrgq You picked up on exactly the thing we cared about, the recorder and field updates aren't two features, they're one loop, the call becomes the update with nothing in between. That only works if you start from how reps actually work, not from bolting AI onto the CRM. Glad it comes through.
The pending-actions queue is the design choice I personally love! Building an autonomous agent myself, the hardest line to draw is which writes the agent commits on its own vs which wait for human review. Great launch, congrats!
Katalyst
@artstavenka1 that autonomous-vs-review line was the thing we went back and forth on most, and landing on per-field earned trust rather than one global switch is what finally felt right. Appreciate you, fellow builder.
Katalyst
@artstavenka1 As a fellow builder you'll appreciate the nuance, we didn't want one global autonomy switch because trust isn't binary, it's per field and earned over time. Low stakes updates can flow, high stakes ones like forecast category stay review gated until the accuracy proves out. That line is the whole design. Congrats on your build too 🙏
The pain you're describing from Datadog is exactly right — enterprise reps rebuilding context that already lived in their email is a brutal tax on selling time. Curious: when a rep has 20+ open opportunities, does Katalyst prioritize which ones to surface in any given week, or does it process everything equally and rely on the rep to decide what matters?
Katalyst
@sabber_ahamed Nothing writes blind, updates show their reasoning and the rep approves. Forecast-level fields can stay approval-only indefinitely, so an inaccurate read gets caught before it touches the pipeline, not after.
What happens if the AI does make a wrong suggestion? Is there a safety net?
Katalyst
@nitish_gautam1 Yes, there's a human in the loop by design. Katalyst surfaces field updates and follow ups as suggestions you accept or reject, nothing writes to Salesforce silently behind your back. So if a suggestion is off, you just decline it, and the rep stays in control of what actually lands in the CRM.
Katalyst
@nitish_gautam1 Building on Avneet, every suggestion shows its reasoning, the call or email it came from, so you can see why one's off before declining it, not just catch it after. The safety net is visibility, not just a reject button.
Curious how it handles overlap when multiple reps are touching the same account - does it prioritize one owner’s notes or merge everything into a unified timeline? Also wondering if the AI resolution stuff can be turned down or off per team, since some of our AE’s get uneasy with auto-updating fields on closed-won opps.
Katalyst
@atlanla11386 On the control point, yes, this is configurable per team and per field. Auto-updating isn't all-or-nothing; you set what Katalyst can touch, and plenty of teams keep it in suggest-and-approve mode rather than letting it write on its own. Locking down closed-won opps specifically is exactly the kind of boundary you can set, so your AEs stay in control there.