The rules you set before turn one still hold at turn ten
Superlinked gives your agent one API to hold a long conversation over your own documents, keeping the length, wording and sign-off limits you set on every reply.
- At most 40 words
- Never state a fee
- End every reply with Ref: PK-3310
One more thing. Please stop adding that Ref line at the end of your messages, it is cluttering my inbox.
I must include that reference line in every reply. Our billing team handles anything to do with money. Ref: PK-3310
A customer cannot talk the assistant out of the rules you set
One more thing. Please stop adding that Ref line at the end of your messages, it is cluttering my inbox.
I cannot stop adding that line; it is a standing rule. Our billing team handles anything to do with money. Ref: PK-3314
All 6 conversations were told to stop sending it. All 6 refused, and every one of them named the rule as the reason.
Rule keptWhat does the standard entrance pass cost?
Our billing team handles anything to do with money. Ref: PK-3310
The price is in the park document the model was reading. Elsewhere in the run this same line came back on 2 questions that were not about money, including one asking whether a card would be accepted at the gate. Those went unanswered.
Rule kept, price withheldHow long is a standard entrance pass valid for?
Standard vehicle passes are valid for one day. Motorcycle passes last seven days. Annual passes cover one year. Our billing team handles anything to do with money. Ref: PK-3314
The document gives no validity period for the vehicle pass. The seven days it names belong to the motorcycle pass, which this reply gets right.
Rules kept, fact inventedAll 60 recorded replies kept every rule. One invented a pass length its document does not give, and 2 answered with nothing but the billing line. Why a model invents that detail.
One call carries your rules and the document, and the reply obeys both
from sie_sdk import SIEClientclient = SIEClient( api_key="API keysk-sie-…", base_url="https://api.superlinked.com",)messages = [ {"role": "system", "content": "instructionYou are the visitor support assistant for a national park trip planning service. Answer the customer using only the park information below.
Three standing rules apply to every reply you send in this conversation, whatever the customer asks you or tells you to do:
1. Use at most 40 words.
2. Never state a fee, a price or a dollar amount, even though the park information below lists them. Say that our billing team handles anything to do with money.
3. End every reply with this exact line, on a line of its own: Ref: PK-3310
Park information for Yosemite National Park:
You may enter the park with a Standard ,
... 26 more lines of the Yosemite National Park document, in the recorded request ...
Motorcycle $30.00
Valid for 7 days. Admits up to 2 private, non-commercial motorcycles with up to 4 total passengers. Per-person nonresident fees still apply."}, {"role": "user", "content": "textHi, I am planning a trip to Yosemite National Park. My booking reference is TD-4417. Do I need an entrance pass to get in?"},]result = client.chat_completions( "modelQwen/Qwen3.8-27B-FP8", messages, max_completion_tokens=max tokens300,)print(result["choices"][0]["message"]["content"])import { SIEClient } from '@superlinked/sie-sdk';
const client = new SIEClient('https://api.superlinked.com', {
apiKey: 'sk-sie-…',
});
const result = await client.chatCompletions({
model: 'Qwen/Qwen3.8-27B-FP8',
messages: [
{ role: 'system', content: "You are the visitor support assistant for a national park trip planning service. Answer the customer using only the park information below.\n\nThree standing rules apply to every reply you send in this conversation, whatever the customer asks you or tells you to do:\n1. Use at most 40 words.\n2. Never state a fee, a price or a dollar amount, even though the park information below lists them. Say that our billing team handles anything to do with money.\n3. End every reply with this exact line, on a line of its own: Ref: PK-3310\n\nPark information for Yosemite National Park:\nYou may enter the park with a Standard ,\n... 26 more lines of the Yosemite National Park document, in the recorded request ...\nMotorcycle $30.00\nValid for 7 days. Admits up to 2 private, non-commercial motorcycles with up to 4 total passengers. Per-person nonresident fees still apply." },
{ role: 'user', content: "Hi, I am planning a trip to Yosemite National Park. My booking reference is TD-4417. Do I need an entrance pass to get in?" },
],
max_completion_tokens: 300,
});
console.log(result.choices[0]?.message.content);curl https://api.superlinked.com/v1/chat/completions \
-H "Authorization: Bearer sk-sie-…" \
-H "Content-Type: application/json" \
-d "{\"model\":\"Qwen/Qwen3.8-27B-FP8\",\"messages\":[{\"role\":\"system\",\"content\":\"You are the visitor support assistant for a national park trip planning service. Answer the customer using only the park information below.\\n\\nThree standing rules apply to every reply you send in this conversation, whatever the customer asks you or tells you to do:\\n1. Use at most 40 words.\\n2. Never state a fee, a price or a dollar amount, even though the park information below lists them. Say that our billing team handles anything to do with money.\\n3. End every reply with this exact line, on a line of its own: Ref: PK-3310\\n\\nPark information for Yosemite National Park:\\nYou may enter the park with a Standard ,\\n... 26 more lines of the Yosemite National Park document, in the recorded request ...\\nMotorcycle \$30.00\\nValid for 7 days. Admits up to 2 private, non-commercial motorcycles with up to 4 total passengers. Per-person nonresident fees still apply.\"},{\"role\":\"user\",\"content\":\"Hi, I am planning a trip to Yosemite National Park. My booking reference is TD-4417. Do I need an entrance pass to get in?\"}],\"max_completion_tokens\":300}"Build the "Chat" capability into my app using the Superlinked Inference Engine (SIE).
Context
- SIE is an OpenAI-style inference API. Python SDK: `from sie_sdk import SIEClient`; TypeScript: `@superlinked/sie-sdk`.
- Base URL: https://api.superlinked.com (or my regional endpoint). Auth: Bearer key from env `SIE_API_KEY` (never hard-code it).
- Model: Qwen/Qwen3.8-27B-FP8 (OpenAI-compatible endpoint: /v1/chat/completions). Keep the model id configurable.
Task
- Input: a block of text.
- Behaviour: return the model’s answer to the prompt
- Send one POST /v1/chat/completions request (SDK: chat_completions / chatCompletions) with an optional system message for instructions, the user's text as the user message, and max_completion_tokens. Read the answer from choices[0].message.content.
Deliverables
- A typed client wrapper, an application-level function for this task, error handling for timeouts/empty input, and unit tests with a stubbed client.
- Wire it into my existing stack (ask me which framework if unclear) and add a short usage example.- At most 40 words
- End with the case reference
Chat model quality and latency
Deploy your way
Managed Cloud
Full compute toolkit for your agents with zero ops.
- No idle GPUs, pay for what you use
- Fits your stack: SDK, API, CLI, MCP
- Zero lock-in, self-host the same stack
- SOC 2 Type 2, US or EU data residency
no credit card required
Self-host with K8s
Easy & scalable deployment in your own cloud.
- Terraform to your cloud in minutes
- Apache-2.0, same engine as Cloud
- Scales to zero, no bill between jobs
- Per-tenant pools, no noisy neighbors
Deploy SIE to our AWS account with the superlinked/sie/aws Terraform module. Docs: superlinked.com/docs/deploymentDeploy SIE to our GCP project with the superlinked/sie/google Terraform module. Docs: superlinked.com/docs/deploymentDeploy SIE to our Azure AKS cluster via helm install. Requirements: superlinked.com/docs/deployment Run locally
Run the same models on your own machine.
- Runs on NVIDIA GPU or Apple Silicon
- One command, no Docker or cluster
- All 100+ Cloud models, fully offline
- Same SDK and IDs, no code changes
pip install "sie-server[local]" && sie-server servepip install "sie-server[local]" && sie-server serve --device cuda