LinkedIn Skills for AI agents

Fictional example data. [ASK: which number was stuck, at what level, for how long?] Customers kept telling us our delivery fees were too high. Then we read 30 cancelled first orders from September: 21 had been promised next-day delivery we could not meet.

“Remote” does not establish France eligibility; this mandatory criterion is unknown.

Fictional example data. - N3 "Maybe 70%": hedged, base unknown. Not usable as a number until checked against the project tracker. - L8 is the interviewer's suggestion ("halved delivery time across the board"). Ines rejected the general claim at L9 and confirmed it only for one project. It is not a position of hers and must not appear as one.

Those observations are useful, but narrow. They tell us that ownership needed clarification in those five requests. We did not compare teams or track completion speed.

Fictional example data. The old rule fired on any soil-moisture reading below 18%. It sent 1,134 alerts. 352 were acknowledged within 24 hours, or 31%. The new rule waited until moisture had stayed below 18% for six hours. It sent 252 alerts. 197 were acknowledged within 24 hours, or 78%. I think the six-hour rule is why. But this was not a controlled test, and the last three weeks were drier than the first three.

I worked alongside Ivo on the Orchard Relay documentation project from January to April 2025. He was careful about the details that make a handoff usable.

Fictional example data. It is early. Eight clients is a small sample, and October is not December. We will know more after the year-end close.

Fictional example data. Did resolution time or reopen rate move along with first response? When we added automatic ticket tagging in March, our first-response time dropped, but reopens went from 8% to 11% over the next two months because some tickets landed in the wrong queue.

No follow-up is drafted: the user has no new information to add. Stop if Ari declines or asks for no contact; silence does not justify another pitch.

I'd separate cost per attempted task from cost per accepted outcome, and include human review and rework in the latter. For a reviewed contract, what would your team count as an accepted outcome?

Em: no bank-change action and no drafted acceptance. Verify the request through a known independent contact channel before deciding what to answer.

Fictional example data. Not today: Kestrelpane has no HubSpot integration. You can export contacts from Kestrelpane as CSV. On the tools, the pilot agencies dropped [ASK: which tools?].

Tagline draft: Project handoff workshops for small design studios

Fictional example data. - Prep time: 30 + 40 + 25 + 60 + 15 + 10 = 180 minutes, equal to her capacity. There is no slack, so if Thursday runs long, drop the Saturday replies or move the poll.

Fictional example data. - Weekdays computed from the dates: 2026-11-16 is a Monday, 2026-11-29 (day 14) a Sunday, so the review meeting is on Monday 2026-11-30. - C-items used before their share date: none. C3 first appears on day 8, 2026-11-23. - Load per participant if they take every slot: week 1, 50 of 60 minutes; week 2, 55 of 60 minutes.

Fictional example data. Next year, I plan to start fractional finance work for seed-stage startups. [ASK: what services do you plan to offer?] If you're a seed-stage founder, I'd welcome a conversation about the finance work you need. This is a plan for next year; I am not offering fractional work yet.

Fictional example data. Tallowbrook Freight has launched Dock Window. Dock Window lets carriers book dock slots in real time. During our 8-week pilot at the Leeds depot, average gate-to-dock time fell from 47 to 29 minutes. Drivers and dispatchers loved it. [ASK: what did they say, and how was it collected?] Thanks to the team who made this possible. Want a demo? [ASK: add the booking link or how to request one.] #logistics #supplychain #freight