Creation Process
Create your own agents, from knowledge graph and first steps, to global deployment.
The entire process will take 10-20 minutes, including training of language models of even millions of custom entities!
Project Launch
Launch your project by signing up with Brighten AI.
Your account provides access to our deployment tooling, device registration, and metered billing.
Once you have a project connection in the Brighten Ai agent geosphere (takes just a few seconds, typically), you can begin creating agents.
Our support team will be standing by to assist.
Build Knowledge Graph (2 minutes)
(2 minutes)
Take a couple of minutes building your first knowledge graph using our tooling.
We designed our knowledge graphs to allow sample entities to be placed right into the graph specification so you can start speaking to the agent in the next step, without any training data.
Run the agent in our tooling (10 seconds)
Seconds after you finish your first edits on your knowledge graph, you can begin speaking to the agent in our tooling. Record a few interactions if you like, for future playback. Take some notes about how things are looking, and then proceed to the next step, or do some more knowledge graph edits.
Add real data (A few seconds training time)
(We’ve done this step in 185 seconds for over a million custom entities- and you can skip it entirely for now and use your sample data embedded in the knowledge graph)
Run our corporate data sync tooling, or work with your application team to get some real corporate data pumping into your agents secure cloud mind. That will kick off the cloud learning systems, and start the process of Brighten intelligently adapting to your corporate data flows.
You should have real data in your agent’s mind a few seconds after the first sync.
Deploy to mobile device(s) (2 mins to setup, 2 mins for initial sync)
(2 minutes)
Download our Brighten binary to your mobile device
(2 minutes)
Activate Brighten with your seed and the agent is live.
Speak to the agent on device (always instant)
Begin speaking to the agent !
During early deployment phase, each interaction will automatically be saved to your cloud instance.
Don’t worry, this won’t be any private audio, but instead, a statistical representation of what you are saying.
This can be used for playback, and further investigations by your Brighten support personel.
Later, when you deploy globally, you can turn off the interaction debugging stream.
View analytics (best with a cappuccino)
View your agent population lifecycle reports to see how your population is growing, and their relative health.
As your agent population interacts with people and each other, they will send progress reports and other feeds to the global agent life cycle stream, which you can view.
This is very valuable: in additon to the expected life cycle reports around agent health, and numbers and types of interactions and their success rates, you can dive deeper and see what people are talking to your agent population about, including any emerging trends, by geo, time of day…
You can work with our analytics team for custom reporting.
Optionally: Add a custom voice!
With our partner CereProc, your agent can have his/her own voice! They are a truly professional team, and are great to work with.
Optionally: Add custom hardware!
With our partner XMOS, you can design custom microphone arrays to create even better listening from our agents. We recommend this for navigational systems in automobiles, living room installations, etc.