Smart Auto-Router
Set the model to “auto” and Core99 picks the best one for every prompt, by task, complexity, speed and cost.
Automatic failover · Pin a model anytime
The AI cloud
Every model. One core.
One API for every leading model, agents fine-tuned on your business, and AI that runs your processes. In one stack.
The platform
Three layers. One stack.
Each layer feeds the next: automations run on tuned agents, which run on the gateway. Enter at any layer.
Layer 1 · AI Gateway
One API. Every leading model.
One OpenAI-compatible endpoint and one key for the top providers, billed in pesos with CFDI.
- Who it's for
- Teams already using LLM APIs
- How you start
- Change the endpoint and API key
Layer 2 · Fine-Tuned Agents
Models that speak your company's language.
Fine-tuned agents for every department, trained on your terms, rules and tools.
- Who it's for
- Companies that need customized models
- How you start
- Public calculator, then a managed service
Layer 3 · Agentic Automation
Put AI to work in your business.
Layer 2 agents applied to your processes, connected to your data, tools and channels.
- Who it's for
- Businesses not using AI yet
- How you start
- Start on your own, or with our team by your side
Layer 1 · AI Gateway
Every model.
One API.
One OpenAI-compatible endpoint and one key. Switch models by changing the model name; migrating takes a base URL and a key.
app.py
from openai import OpenAI
client = OpenAI(
base_url="https://api.core99.io/v1",
api_key=os.environ["CORE99_API_KEY"],
)
resp = client.chat.completions.create(
model="auto", # or pin any model
messages=[{"role": "user", ...}],
)One endpoint. Every leading model.
- OpenAI
- Anthropic
- Google Gemini
- Meta Llama
- DeepSeek
- xAI Grok
- Mistral
- Qwen
- Kimi
- Cohere
- Perplexity
- GLM
- ElevenLabs
- Amazon Nova
- NVIDIA
- Azure
Fast edge architecture
Global serverless network
Close to every user and every model. Built for speed & always on.
- No servers to manage
- No single region to fail
- Traffic shifts to another provider within the same request
List price. Or less.
Token prices match each provider's own platform. Some models come with a discount.
One bill for every provider
Control and billing
CFDI-compliant invoicing (Mexican VAT), prepaid balance, spend limits and quotas per project or team.
Usage reports · Keys with roles and audit logs
Layer 2 · Fine-Tuned Agents
Models that speak your company's language.
- 1
Train
Upload curated examples: conversations that show the answers you want, tool calls included.
- 2
Test
Training runs serverless. No GPUs or clusters to manage. Compare the new model right away.
- 3
Deploy
The model ships through the same Core99 API. Keep improving it with reinforcement learning.
An agent for every department
Fine-tuned on your terms, rules and tools. Agents work together as one multi-agent system, and are more accurate in their domain than a general model.
Start simple
Prompting, then RAG. Fine-tune only when the numbers justify it. Our public calculator tells you which one pays off.
Managed service
For teams without AI staff
Self-serve platform
For technical teams
Start here
Prompting
Recommended
RAG
When the numbers justify it
Fine-tuning
Layer 3 · Agentic Automation
Put AI to work in your business.
Agents connected to your data, tools and channels.
- Layer 2 agents applied to your business processes
- They act on their own, by event or schedule
- Any input in. Finished work out.
Layer 3 · Agentic Automation
Two ways in
Start on your own, or with our team by your side.
Sample project · HR
Problem: A company wants to automate how it selects candidates.
Do it yourself
One agent to start:
- A CV screener that compares CVs with the job description
Anyone on the team can set it up.
Done with you
Many more HR processes:
- Interview scheduling
- Onboarding paperwork
- Employee questions
- Time-off reports
We design, connect and launch it.
Local support in English and Spanish
The platform
Simple by design
Models, projects, agents, channels and billing in one place. No code needed.
Every model. One core.
Tell us what you are building. We will help you pick the right layer to start.