The October reply
A LinkedIn reply that says “interested, but not until our budget resets in October” is worth more than most of what’s already in the CRM, and it’s the thing least likely to get there. Logging it means copying the text, opening HubSpot in another tab, searching the name, choosing between the two Dan Moraleses who both apparently work in logistics, pasting, dating it and setting a reminder for late September. Two minutes. It gets skipped whenever the afternoon fills up, and by October nobody remembers the conversation happened.
It’s one of the jobs people doing their own outreach actually pay dassi for. dassi runs in the browser you already use, where LinkedIn and the CRM are just two tabs you’re signed into, so there’s nothing to connect.
What “log this” does
With the thread open you type “log this to HubSpot” (Salesforce, Pipedrive and a Google Sheet pretending to be a CRM all work the same way). dassi reads the conversation, plus the person’s name and company, from the page and searches the CRM in the other tab. If two records could fit, it asks which. If none does, it offers to draft a new contact. The note it writes has the date, a two-line summary and the exact phrase that matters (“after the October budget reset”), with a follow-up task for October 1, and you see all of it before anything is saved because CRM writes sit on your approval list.
After a handful come out right, save it as a workflow. We like running it at the end of the day over that day’s replies, one conversation at a time, approving the session as a whole rather than each note.
Why nobody built the integration
LinkedIn doesn’t give ordinary accounts API access to messages, and the connectors most people can buy never see what was said. If your company pays for a Sales Navigator tier with CRM sync, look at what it already logs first; that may be all you need. Most small teams don’t have it, and for them the conversation only exists on the page.
The part we find more interesting is the judgment. Whether “sounds good, circle back in Q3” is a warm lead or a polite no used to be decided by whoever typed the note, in about a second. dassi guesses, shows you the guess, and keeps your correction for next time, since its memory carries across runs. We’re not sure yet how well that holds up on a team where five people read the same phrase five different ways.
Staying on the right side of LinkedIn
LinkedIn’s user agreement prohibits scraping and bots, and accounts that look automated get restricted. So this job is small on purpose: conversations you’re part of, read at about the speed you’d read them, inside your own session. dassi isn’t meant for sweeping years of inbox history or pulling profiles in bulk, and if that’s the job, we’d rather you didn’t do it at all. The LinkedIn and Reddit pipeline operations page covers the rest of a small pipeline (catch-up notes, a lead sheet kept current) under the same approve-each-step rule.
The thread and the CRM record go to whichever AI provider you picked, on your own key or through dassi’s relay with included credits. The two Dan Moraleses remain your problem.