Blog

What I wish I’d had: a revenue leader’s case for a workflow layer

I ran an $800M+ renewals business with more technology than we could use. The thing we never had was the layer between the conversation and the work.

Kenny Ngo

Kenny Ngo · · 6 min read

Before Pondros, I led an $800M+ global renewals business inside a high-growth, $1B+ B2B SaaS company.

We built the strategy, policies, tooling, and operating rhythms that sales reps relied on when working with customers. Different parts of the job looked different on paper. The underlying challenge was the same: keep a lot of moving work on track across a lot of people, without letting the things that quietly matter most slip.

We weren’t short on technology

Salesforce was the CRM and the source of truth for our sales teams. That part was never really in question. But the work that actually moved things forward happened everywhere else.

A rep would make a commitment to a customer on a call. Someone would ask for pricing approval in Slack. Legal would request a redline. A VP would ask for an update. A customer would need one more piece of information before moving forward.

Those commitments lived in email, texts, phone calls, Zoom, Slack, and sometimes even a note scribbled on a piece of paper. We had plenty of technology. In fact, we probably had too much. Every quarter, it seemed like we were evaluating another AI tool promising to summarize meetings, surface activity, or generate another report. Most solved a real problem in isolation.

The problem was that they didn’t connect the work, and it was not scalable in practice. What we got was more data, not necessarily more clarity. Meeting summaries still needed someone to read them and decide what mattered. Activity reports told us what happened, not always what needed to happen next. Dashboards could accurately tell us what happened last week while remaining surprisingly quiet about what was about to slip this week, or the next best action.

The job increasingly became maintaining the systems that were supposed to help us run the business.

The real gap was between the conversation and the work

I’d start a Monday with renewal data, meeting summaries, a stack of Slack messages, and a calendar already filling up. None of those systems were necessarily wrong. They just didn’t know about each other.

A rep says, “I’ll get that to you by Friday.” That’s a commitment. But unless someone deliberately transferred it into Salesforce, a project tool, or a spreadsheet, it remained buried inside the conversation where it happened.

By Thursday, the conversation had scrolled away. We managed because good people compensated for the gaps in the process. Experienced operators developed a sixth sense for which accounts were drifting, which reps were overcommitting, and which internal requests were stuck.

“We managed” isn’t the same as “nothing slipped.”

This was especially painful across a field sales organization

Our team was supporting hundreds of sales reps working directly with customers. That meant building processes and policies that could actually work in the field, where things move quickly, conversations happen across channels, and reps don’t have time to stop after every customer interaction and update five different systems.

When hundreds of reps are making commitments across hundreds or thousands of customer relationships, small misses compound. One pricing request doesn’t sink a renewal. One missed follow-up doesn’t necessarily lose a customer.

But hundreds of small commitments, spread across a quarter, can quietly become a meaningful business problem.

The missing layer was the workflow

That’s what I’ve come to believe since building Pondros. The answer wasn’t another system of record.

It was a layer connecting the systems we already had to the conversations where work actually happens. And increasingly, that conversation was happening in Slack.

Somebody agrees to something in a Slack thread. Someone asks for an action. A customer commitment gets discussed after a meeting. Unless a human deliberately captures it, it disappears into the conversation. CRMs weren’t designed to manage every small commitment a sales team makes every day. Project management systems weren’t designed around the “can you send me that by end of day?” requests that happen constantly in a sales organization.

That’s where Pondros fits. We’ve written separately about why Slack project management is worth treating as a discipline. This is the version of that argument I arrived at from the other side of it, as the person being asked why a renewal moved.

Capture, chase, close

Capture

Capture the commitments where they’re actually made, in meetings, Slack threads, and day-to-day conversations. Instead of relying on someone to remember to log the work afterward, the commitment becomes an actionable task with an owner and a date. How capture works.

Chase

Follow up automatically when something hasn’t happened. We spend a considerable amount of time manually asking, and asking again, for things that should already have been moving across teams scattered over different timezones, languages, and geographies.

Instead of opening Salesforce, checking meeting notes, scanning Slack, and digging through documents to reconstruct the state of the business, I’d want the answer waiting for me.

What is slipping? Who needs a nudge? Which customer commitments haven’t been closed? Where is my team blocked? What requires my attention today?

#Priya · direct message
P

Pondros 10:02 AM

Quick chase: you own send the updated pricing deck to Alex,” due Friday. From Kestrel Sync on Tuesday. Still on track?

On it ✓Snooze
The chase, in a direct message. The task, its owner, when it is due, and the meeting it came from.

Pondros can take that work off their plate. How chasing works.

Close

This is where the AI project manager in the multiplayer era comes to life. Capture gets the commitment into a system. Chase makes sure nobody forgets it. Close is the difference between AI that summarizes work and AI that helps execute it.

Closing the loop means checking that the deliverable actually landed, rather than assuming it did because nobody complained. That’s a different job than a better dashboard. A dashboard is somewhere to put things. An AI project manager is the layer that does the asking, on its own, until the work is actually closed out.

Pondros · agentBeta

Pull the Q3 numbers for Acme and build the client deck.

Done

  • recallhow Acme decks are formatted
  • board_commitments4 open tasks, 2 overdue
  • deep_research6 sources, all cited
  • create_google_slides12 slides, house theme
  • share_google_fileshared with the Acme channel

Q3 Review — Acme

12 slides · in your Drive · shared

A job handed to the agent, and the trace of what it actually did.

The next generation of AI isn’t just going to tell us what happened. It will increasingly help make sure the work actually gets done, in a context shared by humans and agents. How the agent closes the loop.

Where this leaves revenue and technology leaders

If I could go back and add one thing to my operating rhythm, it wouldn’t be another tool.

It would be a workflow layer that connected the conversations happening every day to the work that needed to happen afterward. Because the best revenue organizations aren’t necessarily the ones with the most technology.

They’re the ones where the gap between “we agreed to this” and “this got done” is as small as possible.

And where closing that gap doesn’t depend on someone’s memory, inbox, or willingness to chase people for the fifth time. That was the gap I felt every week running a large revenue organization and supporting the sales teams working with customers in the field.

It’s the gap Pondros is built to close. The product tour walks the whole loop.

Start in Slack.

Free to start, no credit card. It works in the Slack you already have.

Free to start, no credit card