Shadow 2.0 - The work your meetings create, done before they end

@rafay_farhan exactly - follow-up emails and scheduling are useful, but they’re also the obvious layer.

The more interesting part is when Shadow can take something from the conversation and turn it into a real work artifact - like a deck, proposal, NDA, quotation, or Jira ticket - while the context is still fresh.

That “can you prep a deck for the next meeting?” moment is exactly the kind of workflow we want Shadow to own.

Curious - what kind of deck would be most useful for you: sales proposal, customer recap, internal update, or next-meeting prep?

@kaushal_bundel this is exactly the right way to think about it.

Shadow should not blindly execute everything it hears on a call. The model we believe in is:

call → proposed actions → review queue → user approves → action goes out

So if sales numbers, pricing, internal context, or anything sensitive comes up, it should not be automatically broadcasted. The user should be able to review, edit, reject, or approve actions before they are sent to anyone or pushed into tools.

And yes, the knowledge base point is a big part of the product direction. The goal is not just “schedule the next meeting,” but to make the action smarter using company context - for example, attach the right deck, include relevant case studies, or add industry trends based on who you’re speaking with.

So the two principles are:

1. Human-in-the-loop by default

2. Company context to make actions more useful

Curious - would you want controls at the workspace level, or per action type, like emails can auto-draft but external sharing always needs approval?

@kaushal_bundel These are all very thoughtful questions. We're starting with an individual plan and gradually moving towards a team-based one and the edge cases you've described are exactly what make this problem hard and worth solving. Privacy will be a central pillar in how we build this out. Would love to have you as an early tester when we get there!

@sharath_kumar25 Thanks for the kind words Sharath! Great to hear from you.

@daniel_henry4 That’s the overload we’re trying to reduce.

@maali_baali That’s what we’re trying to change with Shadow: instead of ending the call with more work, the follow-ups, docs, scheduling, and next steps are already prepared for review.

Curious - what’s the most annoying post-call task for you: follow-up emails, notes/docs, scheduling, or updating tasks/CRM?

@sidharth_choraria1 yes, that’s a big part of the experience we want to build.

Shadow should not just create notes, but make the meeting context usable later - so you can ask things like “what did we decide?”, “what were the objections?”, or “what should I follow up on?”

Curious - would you use this more to recall past decisions, find action items, or prep before the next meeting?

@lipkovskiy The setup process point is well noted. We'll come up with a work around.

@lipkovskiy appreciate this - especially since you’ve tried a few tools already.

You’re right on setup too. We’re working on making that much smoother, because the product should feel useful before it feels like work to configure.

Would love to hear how it goes on your upcoming call - what part of the setup felt slowest for you?

@yuri_mihaileanu1 Great question.

Shadow doesn’t just rely on the call transcript. It needs company context too — things like your docs, past notes, product info, processes, and connected tools - so the task it creates is grounded in how your team actually works.

Also, we keep a human approval layer before anything important goes out. So Shadow can prepare the task in real time, but the user reviews/approves it before it’s sent, created, or scheduled.

Curious - for your team, would the most important context come from internal docs, CRM, past calls, or project/task tools?

@yuri_mihaileanu1 would love to hear your company's knowledge base platforms

@madalina_barbu Thanks a lot!!

@munevver_ertuncccc exactly - trust and timing are the hard parts here.

If Shadow acts too early, it feels risky. If it acts too late, it becomes just another post-call task list.

That’s why we’re building it around a review/approval layer: Shadow prepares the work while the call is still happening, but the user stays in control before anything gets sent, created, or scheduled.

Curious - what would make you trust a tool like this faster: clear previews, approval before every action, or better control over what it’s allowed to do?