This blog captures my frustration with AI tools for work. Microsoft and Google are pushing tools to help you write email or documents but what I need is a help reading and combining information across documents, email, source code and chat. That would be really useful. Not CoPilot for email. - ThreadSky

carnage4life.bsky.social • 35 days ago

This blog captures my frustration with AI tools for work. Microsoft and Google are pushing tools to help you write email or documents but what I need is a help reading and combining information across documents, email, source code and chat.

That would be really useful. Not CoPilot for email.

Comments

aarongrote.com•35 days ago

And they're not even getting the Email stuff right! Do I want an auto reply? No. Do I want THanks to be auto corrected every time? Yes! That would actually save me time!

arachno.capital•35 days ago

I was really excited about using dashworks to find incident-related events and artifacts across slack, linear, and github, and basically make a basic mise-en-place for a post-incident review coordinator

then they got bought out and jettisoned all their customers. back at square one

thatsjustlikeyouropinionman.com•34 days ago

We’re building something kinda like this; tell me more about your need?

arachno.capital•34 days ago

basically what I want to do is have an LLM do a first pass on the drudge work that a post-incident review facilitator would do:
- timeline of the incident response, calling out what look like questions, claims, instructions, etc
- look at the rest of Slack around the same time for "it's down" and &c

arachno.capital•34 days ago

- dig one layer into PRs linked, jira tickets, etc to characterize the work being discussed, add that stuff to the timeline too
- look for surprise, links to dashboards / queries / etc
- probably a whole bunch more stuff
...and present this all while trying to do as little meaning-making work as pos

thatsjustlikeyouropinionman.com•34 days ago

This is actually a very close fit for what we’ve already built, just a rather specific use case I hadn’t yet considered. Let me know if you’d be open to an alpha in the next month or so if we can find time to tune for that application.

arachno.capital•34 days ago

I'm just an IC, so I can't green-light an alpha on my own say-so, but I can definitely get the ball rolling. happy to discuss further over DMs (or in public) too

fallysombo.bsky.social•35 days ago

💯

machineplanet.bsky.social•35 days ago

It's all just descendants of Clippy.

samguckenheimer.bsky.social•35 days ago

If you're drafting in a foreign language, this is useful. Otherwise, it just adds noise to the signal.

apsv.bsky.social•35 days ago

btw email and docs assistants miss that we spend way more time reading and cross-referencing than writing

sorta opposite situation to coding assistants, which they're mimicking

johnnyappleseed46.bsky.social•35 days ago

Try Google Notebook LM. It will look across different documents to synthesize its understanding of their relationship

icoloma.bsky.social•34 days ago

That would hardly cover email or chat, which would have to be mirrored to GCS buckets. The use case sounds more like Agentspace to me, which works through connectors.

d3miurg.bsky.social•35 days ago

In the customer care industry it goes very much in this direction. Data from tickets is structured and combined with data gathered via API. The biggest ROI here is the structuring of the data.

bdowney.bsky.social•35 days ago

Have you tried Glean?

mike.puterbaugh.net•35 days ago

Two definitions of “information synthesis” and the industry is focused on the wrong one

Comments

Posting Rules

Reply