10x the intranet with
Meet the intranet that builds itself, helps you prioritize what matters, and acts as your own personal assistant – without the coffee breaks.
Features
Happeo uses a mix of three AI model categories: Google’s LLM, open-source models, and our own proprietary models.
Page builder
Focus on sharing content instead of building it. Happeo’s AI-powered Page Builder let’s you prompt what you want – building best-in-class intranet Pages from a model that’s trained to know what good looks like.
Notification Dashboard
Eliminate notification overload. Happeo's AI-powered dashboard digests notifications, boxes them into actions or FYIs, and prioritizes them based on who you are and what you care about in your job.
Search
Stop Searching, start asking. With the power of AI, we’ve combined today’s keyword-driven search with the power of semantic search to provide instant, precise answers in plain language – English, or anything else. Ask a question, get a direct answer and instantly be referred to the source if you need more.
The Tech
Google’s LLM is the first thing users will see, but there’s much more under the hood. Find out how Happeo is powered by AI, right here.
The models
Happeo uses a mix of three types of AI models: Google’s LLM, open-source models, and our own proprietary models. Google’s LLM serves as the language-input.
Large Language Model
This is the part of AI that most people are now familiar with. We’ve chosen to work with Google’s LLM as it’s an advanced AI model that’s also extremely cost-efficient. By comparison – OpenAI’s ChatGPT costs more than twice as much. You could look at it as the “front-end” of our AI model.
Open Source
We employ a combination of open-source models that are used for specific purposes. One lets you classify text. Another anonymizes the vast database of text available. The third is trained to create accurate and fast embeddings.
Proprietary models
Happeo is a specific product – an intranet, to be precise. If you’re thinking of incorporating Generative AI into your own product, expect to eventually run into limitations. That’s why we’ve developed our own proprietary models enabling us create prompt-based output, specifically for an intranet.
Our advice on time
Training an AI model is the most time-consuming part of incorporating Generative AI. If you can stick to language-based output, you can probably get by with Google’s LLM and some open source models – there’s a lower chance you would need to train and create your own proprietary model. Start with one and see how far you can take it until you need to add an additional model.
Cost considerations
AI models are expensive to use. Google’s LLM is less than 50% the cost of OpenAI’s, and just as good but it’s still a costly endeavour. With Generative AI, every prompt and response has a cost implication that can quickly add up, so it's worth thinking about your monetisation strategy.
Decreasing costs
One of the first things to ensure is that your product does not use AI in every instance.
-
Optimize for efficiency
Identify the highest-value use cases in your product and start there. Don't run it everywhere.
-
Decide a cut-off point
Realize that there might come a point where it's more cost-efficient (and faster!) to train and create your own model.
-
Design your own infrastructure
Run AI-systems on your own infrastructure with clusters that can handle smaller workloads. This way you don’t have one giant machine crunching prompts – smaller tasks can be delegated to smaller systems, saving costs.
FREE WEBINAR
OCT 26 @ 4PM CEST | 10AM EDT
Sign up for our free webinar and learn from the creators
Here's your chance to learn from the creators. We'll host an exclusive webinar where they'll walk you through the why, what and how of getting started with AI building. Sign-up now and dive deeper than our Google talk did!
Hosted by:
Antero Hanhirova
Chief R&D Officer, Happeo
Lorenzo Berti
Customer Success Manager, Happeo