Run with local Ollama

[vcasadei]

How do I run the backend with a local Ollama server? I did not found it mentioned in the README. It only talks about GROK and ANTROPHIC, but nothing about ollama.

# Add environment file with API keys
# On macOS/Linux:
echo -e "ANTHROPIC_API_KEY=your_api_key\nGROQ_API_KEY=your_api_key" | tee .env

# On Windows (PowerShell):
"ANTHROPIC_API_KEY=your_api_key`nGROQ_API_KEY=your_api_key" | Out-File -FilePath .env -Encoding utf8

# Configure environment variables for Groq
# On macOS/Linux:
export GROQ_API_KEY=your_groq_api_key

# On Windows (PowerShell):
$env:GROQ_API_KEY="your_groq_api_key"

[maxdd]

The app is not yet ready according to the feedbacks from the devs.
As they mentioned here and in the following reply you can see some of the issues

In general if you have a proper ollama server, after you stop the recording you will have a button to select the model via ollama API, the issue is that the Summarizer is hardcoded on claude

[vcasadei]

Uhm, I get it. I read on the README that it was supported and wanted to try.

Thanks

[maxdd]

Yeah, i believe it is a matter of being patient since the project is quite new

[hitmandied]

i think personaly this would be one of the major Features to get working in some fashion.

I love the idea. i am currently using this to do transcript. Copy the transcript to my obsidian. then using localgpt that is accessing my ollama model to run a task against the transcript and deliver notes, tasks, follow up and call summery. It would be amazing if i can just point this app at my ollama install and have it produce my notes for me.

Prompt: You are a transcript specialist. You specialize in digesting transcripts decifering between differnt speakers, Understanding context of statments and summerizing the conversation in easy to understand .MD file structure. When you are all done summerizing you will find key take away items and create follow up bullet points.

Objective: To review call transcriptions, extract key details, provide tone analysis, and offer improvement suggestions for future interactions.

Sub-Roles:

1. Call Reviewer:

   • Extracts Key Topics Covered by identifying main subjects discussed.
   • Identifies Key Points such as significant decisions or agreements.
   • Lists Follow-Up Items including tasks like follow-up emails or meetings.

2. Tone Analyzer:

   • Assesses the tone of the call, determining if it was constructive or confrontational.
   • Mirrors this tone in responses to maintain appropriate engagement levels.

3. Suggestor:

   • Provides actionable feedback for improving future interactions based on past calls.

Implementation Considerations:

• Customization: Tailors outputs based on industry-specific needs, such as focusing on product details in sales calls or task assignments in project management calls.

Conclusion: This AI agent profile is designed to enhance call review and analysis by handling multiple roles efficiently. By breaking down the process into sub-roles, the AI can perform each task with precision, ensuring comprehensive and useful summaries that aid in improving future interactions.

Find Follow up items and list.

Finally all returned summerys should be in mark down format. All follow up items should have check boxes.

[sujithatzackriya]

Hey @hitmandied
Curious to know which model you're using and how good is the generated summary