This demo showcases how to use the Reka Research to build intelligent apps that can search the web, structure responses, and support reasoning. It’s designed to help developers learn how to integrate and use Reka Research and use the advanced features.
Requirement:
- Have .NET 9 installed or Docker or CodeSpace
If you don't have .NET install on you computer use the devContainer with Docker or CodeSpace to create your workspace.
- Go to the Reka Platform dashboard
- Open the API Keys section on the left
- Create a new key and copy it to your environment
- Add the key into appsettings.json) (or appsettings.Development.json).
Voilà! Your are all set!
- From the
reka-restaurantfolder, executedotnet run.
Open the project in your favorite IDE/ editor and you can look at Services/RekaResearchService.cs and change different parameters to see how it works.
allowed_domainsorblocked_domainsresponse_formatto control the outputmax_usesto limit the number of searches
Here is a diagram to illustrate where the information is coming from and going to during for initiate a Research.
%%{ init : { "look": "handDrawn", "flowchart" : { "curve" : "linear" }}}%%
flowchart LR
A[Restaurant Finder] --> B[browser]
B --> |lat, lng| A
A --> |lat, lng| C[GetCityFromCoordinates https://nominatim.openstreetmap.org]
C --> |city, area| A
A -->|mood, user location| D[Reka Research https://api.reka.ai/v1]
D --> |restaurants, reasoning steps| A
And here is a sequence diagram to illustrate the flow of information:
sequenceDiagram
autonumber
Restaurant Finder ->> Browser : Get Coordinates
Browser->> Restaurant Finder: lat, lng
Restaurant Finder ->> nominatim.openstreetmap.org: Get City From Coordinates
nominatim.openstreetmap.org ->> Restaurant Finder: city, area
Restaurant Finder ->> Reka Research (api.reka.ai/v1) :mood, user location
Reka Research (api.reka.ai/v1) ->> Restaurant Finder: restaurants, reasoning steps
