Local AI Chat
MeetNote lets you choose from multiple on-device AI chat models, so you can generate summaries, action lists, and MoM-ready drafts with the model that fits your device best.
One summary workflow
In fast-moving teams, the biggest bottleneck is often interpretation, not access.
People ask the same questions repeatedly:
- What did we decide last week?
- What changed between meetings?
- Who owns this follow-up?
- What is still unresolved?
Without an intelligent layer, teams manually scan long records, and decision latency increases. Local AI Chat reduces this by making meeting context queryable in natural language, so users can extract value quickly and keep momentum.
MeetNote doesn’t force a one-size-fits-all model. It offers a lineup:
- SmolLM2 (135M) — ultra-light and fast for broad compatibility
- Qwen 2.5 (0.5B) — balanced speed + quality
- Qwen 2.5 (1.5B) — stronger reasoning with larger footprint
- Phi-3.5-mini (3.8B) — supermodel tier for deeper logic
- Gemma-2-9B — supermodel tier for richer analysis
- Mistral NeMo (12.2B) — supermodel tier for high-capacity multilingual/long-context work
So your “Summarize this meeting” can run on lightweight hardware or scale up on high-memory devices.
How it works across device classes (“any device” done right)
MeetNote’s local AI path is adaptive:
- Download model files to device storage (GGUF-based model flow)
- Load and run with native runtime acceleration preferences (Vulkan on Android, Metal on iOS)
- Apply model-specific context/window and inference settings
- If one runtime path is unavailable, use fallback runtime behavior when supported
- If a chosen model is too heavy, fallback logic can move to a compatible option
In short: the app is designed to meet the device where it is—light models for broader reach, heavier models where hardware allows.
Built for summaries, recaps, and minutes composition
Inside meeting chat, users can trigger fast summary-style workflows such as:
- Summarize
- TO-DO List
- Budget Overview
Under the hood, responses are grounded in:
- Current meeting transcript content
- Meeting metadata (title/date/status/owner/length/team/project)
- Transcript-focused context selection and compression for long sessions
This is what makes outputs practical for a Minutes Composer flow: cleaner recap drafts, better action tracking, faster MoM preparation.
What you get
1) Faster context retrieval for real decisions
Local AI Chat helps users pull relevant insights from meeting history quickly.
Instead of digging through long records, teams can ask direct questions and move from search to action faster.
2) Better support for high-frequency collaboration
Teams running many weekly meetings often struggle to maintain context continuity.
AI-assisted querying makes it easier to reconnect with prior discussions and avoid repetitive alignment cycles.
3) More useful meeting outputs for busy stakeholders
Not everyone can review full records in detail.
Local AI Chat supports practical synthesis so leaders and contributors can absorb key outcomes, risks, and next steps with less cognitive overhead.
4) Stronger follow-up quality
When teams can quickly clarify what was decided and who owns what, follow-up quality improves.
AI-assisted interpretation reduces ambiguity and helps keep execution aligned with actual meeting outcomes.
5) Lower friction between “recorded” and “usable”
Many organizations capture meetings but underuse the content afterward.
Local AI Chat helps bridge that gap by making historical context actively useful during planning, reviews, and decision cycles.
6) Better support for iterative planning environments
In iterative work, decisions evolve and priorities shift.
Local AI Chat makes it easier to compare reasoning, identify unresolved issues, and carry forward important context into the next planning step.
Use cases
Supermodels are powerful, but not always practical on every setup.
- Product teams: Ask what changed in scope decisions across multiple planning meetings.
- Operations teams: Extract recurring blockers and owner assignments rapidly.
- Client services: Summarize stakeholder concerns and confirm agreed next steps.
- Leadership: Get concise context before strategic review calls.
Result: fewer crashes, fewer dead-end loads, better day-to-day reliability.
