30+ LLMs hot loaded to run instantly
Built in prompt optimization for every LLM
No virtual rate limits, no tokens, no GPUs
Store and manage dynamic variables in prompts
Get guaranteed consistent JSON response
Prompt Engine Preview (Alpha)
Try all advanced features ->
{
"success": true,
"result": [
{
"poem": "In New York's dark alleys, shadows play,\n Where the city's secrets whisper through the day.\n A chill runs down your spine, as you walk alone,\n In the city that never sleeps, where terror's home.",
"sentiment": "Fear"
},
{
"poem": "The subway tunnels stretch, a labyrinth deep,\n Where the lost souls wander, in eternal sleep.\n The trains rumble by, a haunting, mournful sigh,\n As the city's darkness swallows, the living and the dead.",
"sentiment": "Despair"
},
{
"poem": "In Times Square's bright lights, a monster's grin,\n A reflection of the city's, darkest within.\n The crowds rush by, a chaotic, maddening throng,\n As the city's madness consumes, and the sane are wrong.",
"sentiment": "Anxiety"
},
{
"poem": "The skyscrapers loom, a towering, Gothic spire,\n Where the city's ghosts reside, in a haunted, endless fire.\n The wind howls through the streets, a mournful, lonely cry,\n As the city's darkness whispers, a haunting, final goodbye.",
"sentiment": "Sorrow"
}
],
"message": "It is recommended to store the prompt engine and execute it using the ID rather than running it directly for the best performance and reliability. Learn more: https://docs.jigsawstack.com/api-reference/prompt-engine/create"
}
Built in dynamic caching with parallel processing for the fastest response times
All data is secured in privatized serverless containers that don't store any data
Log and monitor every prompt and response for debugging and performance monitoring
Automatically get access to the latest models without breaking changes
Store, manage and update prompts with dynamic variables with built in prompt optimizations
Get consistent JSON structure for every prompt based on your defined structure
Easy to use REST APIs that work out of the box in every language and framework with fully managed caching, logging and authentication
import { JigsawStack } from "jigsawstack";
const jigsaw = JigsawStack({
apiKey: "sk39wo393.....32ncsmw9339RNj3"
});
const response = await jigsaw.prompt_engine.create({"prompt":"Generate a short story about a {topic} in {language} based in {location}","inputs":[{"key":"topic"},{"key":"language"},{"key":"location"}],"return_prompt":{"story":"The story","sentiment":"Sentiment of the story"}})
$
npm i jigsawstack
3 ways our customers use JigsawStack to build Prompt Engine powered applications
Run any application specific prompt without having to compare the quality of output or the model to use
Chain prompts together to build complex automatons with dynamic variables and responses
Extract relevant application specific data by running prompts on the fly
Striking the right balance between code and dashboard
Access real-time logs and analytics on all your APIs. Debug errors, track users, location maps, sessions, countries, IPs and 30+ data points
Fine grained control over API keys. Whitelist domains with flexible wildcard support, set expiration date and limit access to specific APIs with unlimited keys
The best docs are the kind that you don't need. Fully typed SDKs with auto-completion and self explanatory params
Manage multiple projects and teams with access control. Invite unlimited team members and assign roles
JigsawStack APIs are built from the ground up on the edge network
99.5% uptime with APIs latency reaching as low as 200ms globally
Scale up and down as you need without worrying about abused cost with usage based pricing
Consistent request and response structure across all API services for predictable use
Consistent training for all JigsawStack models to ensure the latest technology is always available without breaking changes