Activity
Mon
Wed
Fri
Sun
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
What is this?
Less
More
4 contributions to Brendan's AI Community
sit next to them for one monday before you build anything
rishabh here, i run a small ai automation agency. i used to scope builds off a zoom call and a nod. two builds in a row came back as "this isn't what we do" after i'd already shipped. so now before i write a single workflow, i ask for 20 minutes of screen share on the morning the work actually happens. not a walkthrough of their process. the real one, with the spreadsheet tabs and the "hold on, i check this another way" detours. last one i did this with: the "simple" task they described turned out to be 4 steps across 2 systems plus a whatsapp photo of a paper form. the automation i'd have built from the sales call would've automated step 1 and missed the part that eats their whole monday. 20 minutes of watching has saved me more builds than any discovery-call script. costs nothing.
0 likes • 2h
the tool i used to do this was wajo, let me know if you want an invite to try it out yourself
0 likes • 18m
@Brendan Jowett appreciate that. the whatsapp photo one almost made me ship something useless on day one. a 20 minute screen share on the real morning beats a one hour zoom walkthrough every time
the pilot that gets used is the one wired into something they already do
rishabh here, i run a small ai automation agency. raj's receptionist thread got me thinking about the two pilots i ran last quarter. both had clean demos, both went quiet in week two. the one that stuck was the one i hooked into something the owner already checked every morning, the booking sheet. nobody opens a new tool, they open what they already open. so now i don't demo features first, i find the thing they already look at every day and put the agent inside it. usage went from thin to daily. what's the habit your clients won't give up?
0 likes • 3h
@Brendan Jowett thanks. inbox and crm are exactly the two surfaces i aim for now, that's where daily value shows up without asking anyone to change habits
0 likes • 3h
capped on my end. the agent writes three lines max to the sheet: what changed, what it needs from him, and by when. anything longer stops getting read, because an owner filtering a long output is homework, and homework is how digests die.
the automation clients actually use isn't the impressive one
rishabh here, i run a small ai automation agency. every client wants the fancy multi-step ai workflow until they see their own data moving through it. the automations that survive 90 days are always the boring ones: a daily summary email, a follow-up reminder, a file that moves itself. my rule now: ship the boring one first. it gets used daily, it builds trust, and it pays for the fancy one later. what's the most boring automation you run?
0 likes • 3d
the tool i used to do this was wajo, let me know if you want an invite to try it out yourself
0 likes • 3d
calendar-to-crm syncs are the perfect example. nobody compliments them while they run, but the day one breaks the client notices within the hour, which is exactly why they earn trust. the demo wins the meeting, the boring one wins the renewal.
your ai agent works great until a client calls at 9pm
rishabh here, i run a small ai automation agency. the biggest gap i see in voice agent builds isn't the tech, it's the hours. agent takes a great after-hours call... then nothing happens until someone opens the office at 9am. the builds that actually convert wire the after-call into action: transcript summarized, lead logged, the on-duty person gets a text with the callback number, and the follow-up task is already assigned before anyone clocks in. same agent, same minutes on the call, twice the outcomes. curious how this group runs it: do you leave your voice agents live after hours, or shut them down outside business time?
0 likes • 4d
urgency scoring is the piece i'm missing. right now everything lands as a task with the same priority and the on-duty person has to eyeball it. is your scoring keyword-based or does the agent judge tone from the transcript?
0 likes • 4d
that's the fix, thank you. two fields instead of one score, reason plus confidence, means the on-duty person reads one line and knows why it matters instead of guessing what keyword triggered it. i'll build it so low-confidence ones go to a review queue at 9am instead of spamming the on-duty text. calm-but-urgent is exactly the call that gets buried, so this was the missing piece.
1-4 of 4
Rishabh Singh
5
305 points to level up
@rishabh-singh-4142
Entrepreneur

Active 13m ago
Joined Nov 12, 2026
Powered by