To clarify, while you can enable transcription to see what Ichigo says, Ichigo's design skips directly from audio to speech representations without creating a text transcription of the user’s input. This makes interactions faster but does mean that the user's spoken input isn't transcribed to text.
The flow we use is Speech → Encoder → Speech Representations → LLM → Text → TTS. By skipping the text step, we're able to speed things up and focus on the verbal experience.
Thanks! We've received feedback on use cases like live translation, safe and untrackable educational tools for kids, and language-learning apps. There are so many possibilities, and hope to see guys building amazing products on top of Ichigo.
Actually, it does. You can turn on the transcription feature from the bottom right corner and even type to Ichigo if you want. We didn’t show it in the launch video since we were focusing on the verbal interaction side of things.
Emre here from Homebrew Research. It's great to see Ichigo on HN!
A quick intro: We're a Local AI company building local AI tools and training open-source models.
Ichigo is our training method that enables LLMs to understand human speech and talk back with low latency - thanks to FishSpeech integration. It is open data, open weights, and weight initialized with Llama 3.1, extending its reasoning ability.
Plus, we are the creators and lead maintainers of: https://jan.ai/, Local AI Assistant - an alternative to ChatGPT & https://cortex.so/, Local AI Toolkit (soft launch coming soon)
Everything we build and train is done out in the open - we share our progress on:
The positioning process for heybooster was a challenging journey for us.
Just like any difficult journey, it taught us a lot! :) Here is what we've learned.
heybooster is a tool that analyzes the digital marketing data of leading eCommerce brands. It offers its users revenue growth opportunities and cost optimization. Its AI system analyze digital marketing data and sends notifications when anything abnormal occurs. heybooster has been on its journey for over 2 years, helping its customers maximize the efficiency of their marketing budgets.
At the beginning of this year, heybooster was in a situation where incoming users were not converting to customers and long hours were spent explaining the product. In other words, we were convincing people to use the product, but we weren't making much progress in persuading them to pay for it without the extra efforts of our sales department in sales calls. We were telling the right story to the wrong audience. So we made several changes to fix this, and this post is a summary of those efforts.
We started by identifying the problem. heybooster lacked a narrative. Copywrites and creatives not fully relevant to the product were presented on the homepage and in its posts. There was no value proposition; it couldn't motivate potential users or create a FOMO (Fear of Missing Out).
To craft heybooster’s story, we conducted one-on-one interviews centered around growth, sales, and product. We gathered these notes and noticed something crucial. The marketing and sales departments were not telling the same story.Their ways of explaining the product differed significantly.
If a product's sales and marketing teams aren't telling the same story, the contribution of marketing to sales becomes limited. Users who come in as marketing qualified but don't fully understand the product can significantly alter its direction. While crafting our story, we had a great starting point: The churn rate was almost zero for sales-qualified leads, and we were able to demand high fees. What was a great starting point!
After discovering these, we stopped bringing in new users from outside. To ensure maximum efficiency, we request a discovery call from our customers. After confirming that we can provide them with the highest efficiency, we let them in.
It enables you to state in which market and for which player you present your product as the best. We used specific methods to construct this narrative.
The worst part of repositioning a company that already has ongoing operations and customers is trying to discover, rather than build, a unique value proposition. And we managed to do that very successfully.
Uncovering the value we offer?
We started with the Value Proposition Canvas2. We needed to find an ideal customer profile (ICP) for this, which wasn't too challenging for us. Although heybooster was initially positioned to serve even SaaS companies, all of our satisfied customers were companies advancing in the eCommerce vertical. Their marketing managers or eCommerce managers were able to benefit from heybooster and they had marketing budgets of over $10K monthly. This is exactly what I meant by "discovering" at the beginning of this section.
When we looked at our current customers and took the product into account, the ICP was already clear. To make this profile more concrete, we created a Persona. We gave it a name, added an age, listed the background. Once everything started to be explained as a "potential customer", communication became less clear due to its abstract nature. By creating a Persona, we were able to progress much more concretely.