Spellar 3.0 - AI Meeting companion with cross-meeting memory

@lakshminath_dondeti Yes! Transcription runs fully on-device by default - we use a local Whisper model on your Mac, so audio never leaves your machine. Cloud transcription is opt-in for users who want it (e.g. for speed on older hardware). Summaries / AI notes currently use cloud LLMs, but on-device summarization is on the roadmap.

@lakshminath_dondeti but, just fyi, local engine requires M1+ laptop. Foreigh languages needs larger models - which you can easely donwload/install models in the app's settings

@igal_kalnitsky Hi, we do have a free tier, so people can try the product without jumping straight into a paid plan 🙂 After that, there are different paid plans depending on usage.

@igal_kalnitsky in addition to the free trial, we have ProductHunt deal - 20% which expires in the end of May 😜

Thanks @artstavenka1  — "useless two weeks later" nails the pain. Memory search runs across every meeting in your account, not just the ones you attended yourself: in a personal account that's all your recordings, in a team workspace it's everything that's been shared with you. So "what did clients actually say about pricing last quarter" works even if the call was a teammate's

What makes it really sharp is pairing it with Templates — you define a custom summary structure per meeting type (e.g. a Sales Call template that always pulls out pricing objections, budget signals, and next steps as their own fields). Two weeks later when you query across meetings, the AI has structured, comparable data to work with instead of free-form summaries — recall is dramatically better

Got a specific recall pattern you're trying to solve for? Happy to suggest a template setup 💜

@igorsorokinua Hi Igor, thanks a lot for the congrats!

We’re always open to feedback from our users and actively add new integrations based on their requests.

We actually already have a Linear integration, and I personally use it to save time when creating tasks after calls 😊

@igorsorokinua yeah, Linear was one of the first released integrations

We're very open for user feedback / requests. Just let us know what's missing for your workflow?

@j_che Yuliia, very proud we’re launching this together 🚀

@zahle_khan honestly, we were considering it, but it would be challenging since we have many infrastructure components for that.

Is your concern privacy? If so, Spellar has several privacy-aware configurations, such as audio retention and local transcription, among others!

@rus1ankova1enko Yes, exactly that! Spellar has built-in templates for the most common recurring formats — including 1:1, Stand-up,

  Sprint Planning, Retro, Weekly Sync, Project Sync, plus ~15 others (customer discovery, board, investor update…).

  You can also create your own: define the structure, add custom context (project background, what to listen for), pick

  the AI model, and attach it to specific recurring calendar events — so every Monday standup or weekly 1:1 gets

  summarized with the same shape automatically.

@rus1ankova1enko as Andrii mentioned, templates are your key feature here!

You can set up the type of meeting, add additional context, and establish structure (folders, tags, AI provider)

This is essentially your way to customize your meetings to suit your cases!

@rus1ankova1enko 

Yes 🙌

So recurring meetings can automatically follow the same summary structure every time, which makes everything much easier to scan later 🧠

@richa_das24 Totally fair question 🙌

Honestly, the biggest difference is that Spellar goes way beyond just transcription — it’s all about memory and making connections across your meetings.

You get an AI assistant that actually remembers the context from every previous conversation. So, even right in the middle of a meeting, you can ask things like:
• “What did we decide last time?”
• “Did the client already mention this?”
• “What were the action items from our earlier calls?”

The idea is for it to feel less like just a notes tool, and more like having a second brain for your meetings 🧠

@richa_das24 Hey Richa, that's a really good question. I understand that switching to another productivity tool can be challenging if you're accustomed to one.

However, Spellar is definitely worth trying. Its templates and Obsidian integration demonstrate the full customization available to elevate your productivity to 100% - give it a try, and you won't be disappointed 🤙

@avikshit_lp Yes — no transcript-hunting needed. Spellar pulls action items out automatically as a dedicated section, with owner

  and context attached when the conversation makes it clear ("Anna will send the deck by Friday" → task with owner +

  due).

  You can review them on a separate "Action Items" tab, then push each one straight to Linear, Jira, Notion, Todoist,

  Asana, ClickUp, or Trello with a single click. The transcript is still there if you want to jump to the exact moment

  something was decided, but you almost never need to.

@avikshit_lp everything is automated. Just turn on Spellar on your calls/meetings - and the app will take the rest 🤙

Thanks @vaibhavi_suvarna  — that "conflicting information across long client relationships" question is exactly the messy case most cross-meeting tools wave at.

Honest answer on how it works today:

Cross-meeting search surfaces all the relevant decisions, not just the latest. If you ask "what did we agree about pricing for Acme?", Spellar pulls every mention from the folder and lets you see the evolution — March said $50k, May said $60k, here's both with links to the source meetings. We deliberately don't collapse to "latest wins" because the change is usually the interesting signal.

The pattern that makes this really sharp is Folders + custom-context Templates per client:

- A folder per client (e.g. Acme Corp) scopes the memory.

- A custom Template with client-specific context — e.g. "Track decisions on pricing, scope, deadlines, and owners. If a current decision contradicts prior one in this folder, call it out explicitly and cite the earlier meeting." — instructs the AI to flag contradictions on each new summary, not just at query time.

What we haven't shipped yet: proactive cross-meeting conflict alerts (e.g. notify me the moment a new meeting introduces a contradiction with an earlier one). The data layer makes it possible — we just don't auto-trigger it today. High on the list.

Quick one back: would alert-on-conflict be more valuable inside the meeting (a heads-up while you're still talking) or as a digest after the call?