Community seeding list

The person who’d open this is whoever is building Ramp’s finance-ops community, at the point where they need a first batch of people to invite and don’t want to guess. They get fourteen practitioners who posted publicly about spend management, month-end close, AP and T&E in the thirty days to 26 September 2026, ranked and split into three asks: invite to the community, invite to the advisory board, invite onto the podcast. Each row opens to show what the person posts about, their most-engaged recent post with a link, who actually replied to it, and a one-line call on whether they’re a practitioner or a vendor. A “screened out” panel shows who I dropped and why.

The usual way to do this is to scroll LinkedIn, sort by follower count, and paste names into a spreadsheet that’s stale within a week. Follower count would have given the wrong list here. The top three posts by raw comment count all belonged to vendors selling against Ramp, so ranking by comments alone would have made the vendors the first names on the list. Reading the replies is the part that takes hours by hand, and it’s what separated a real thread from a pile of “Interested”.

I started with trayo_get_workspace, which returned the ICP that defines the category. It’s a saved Ramp buyer profile, not a placeholder, so I didn’t have to invent the category myself. Four trayo_search_posts calls returned 200 public LinkedIn posts across the 30-day window. I opened the 12 most promising threads with trayo_get_post_engagers, which returned 34 named commenters with their titles, companies and the text of what they wrote. From those I scored each person on four things: what share of commenters wrote a real reply, how many substantive replies there were, how many of the repliers hold a finance title, and whether the post carried first-hand specifics. Followers don’t enter the score. trayo_research_person then returned career histories for 11 of the shortlist. Last, I wrote the fourteen chosen people back with trayo_add_people and saved them as a Trayo list, “Finance Ops Community Seeding — Sept 2026”, with 14 members. Every name, title, company, photo, post excerpt and quoted comment in the app came back from one of those calls.

The comment data changed the ordering more than I expected. One practitioner with nine comments was reposting a template that four other accounts published word for word that week, and eight of his nine comments were “Interested” or a pasted CV, so he’s in the screened-out panel and not on the list. Jeremy Salles at Marvell had six comments, five of them job-seeker noise, which dropped him from a would-be top five to eighth. Trayo exposes comments and commenters but not reactions or impressions, so reach here is measured on replies only. That has a cost. The people with the most specific first-hand stories, like the Head of Finance who cut a four-country close from twelve days to six, had zero comments, so they score low by construction. Five of the fourteen are in that position, and the app says so on their row rather than hiding it.

The data is real and the judgement is mine. The ordering of the cohort, the practitioner/vendor call on each person and the four score weights are my calls layered on Trayo’s data, and the method panel says so. Export CSV is real and downloads client-side. The three outbound buttons, Push to CRM and the per-person invites, are stubs. They raise a toast tagged “Demo — nothing was sent” that describes what would have gone out, because this is a demo and nobody should get an invitation from it.