@dimakhanarin All three labels describe the same engagement: someone sits with the client, works out how the job actually runs today, then automates it. That middle step is where most of the fee goes, and it's still mostly done by hand in workshops.
ProcessScout shortens the dig: it drafts process maps from your SOPs, policies, docs and workflow recordings, with each step linked to its source. https://t.co/QnY6gmJI2O
The first 90 days of most AI projects is archaeology: digging up how the work actually gets done.
Teams that skip it pay for it later, with interest.
How long did discovery take on your last automation project?
@mfranz_on Treat a skill like an SOP handed to a new hire: every step should trace back to a source someone approved. A step that can't point to where it came from is the one to audit before it touches real data.
@rohanpaul_ai@humynlabs Office automation has the same problem. Agents get built from the SOP, which is the tidy-lab version of the job. The workarounds people use on a messy Tuesday only show up when you capture the work as it's actually done.
@jordan_ross_8F Knowing the numbers shows you where the constraint is. Knowing the process shows you why it's there. Founders who can walk through how an order actually moves from sale to delivery tend to spot the fix before the dashboard does.
@DataChaz Using recorded outputs as the spec is the smart part. Docs say what a tool should do, recordings show what it actually does. Business processes work the same way, and most automation projects skip that check.
Ownership gets easier when the first draft isn't a blank page. ProcessScout drafts the map from the SOPs, policies and recordings you already have, each step linked to its source. https://t.co/QnY6gmJI2O
Unpopular opinion: most companies don't have a documentation problem. They have an ownership problem.
Nobody owns the process, so nobody owns the doc, so the doc dies.
Who actually owns process documentation where you work?
@chamath Hi Chamath, I agree with you. Sometimes you need to know the business context (analyse the current state from email / slack / ERP / CRM) and create the future (org, process, system and data architecture). Our product does this and could be a nice input into 8090
@kevinkern Agree the board is human reporting. What carries over is the stage definitions. An agent moving from review to fix still needs to know what passes review, and that usually lives in someone's head, not on any board.
@FellowAInotes@sspaeti@jrdntgn Seems we could collaborate - check out https://t.co/QnY6gmKfSm Meeting intelligence is a white space in the venn diagram. Who can we talk to to progress this?
@sspaeti@jrdntgn Good read. The implicit stuff is the hard part. In my experience it only surfaces when you watch someone actually do the work and ask why at each step. Then it needs to stay tied to where it came from, or the company brain drifts like the old SOPs did.
@HamzaAlsamraee Exactly. Redundancy feels like waste right up until the day it saves you. The trick is knowing which slack is deliberate and which is just nobody having looked.
The most expensive consultant deliverable is a process map that's out of date the week it ships.
If a step can't point back to its source, it's an opinion.
Agree?
@IvelisseArroyo The undocumented process point is the one I keep seeing. Teams buy the tool, then spend months arguing how the work actually gets done, because three people already run three versions of it.