One API key. 300+ models. Theo routes every prompt to the perfect model, automatically.
You're building with AI. You shouldn't need a PhD in prompt engineering and 7 vendor relationships to ship.
Every provider has its own SDK, auth, billing dashboard, and rate limits. You're managing 7 relationships just to cover your use cases.
You're hardcoding model selection. One for code, another for chat, another for images - all manual, all fragile, all wrong when a model goes down.
4 billing dashboards. 4 invoices. No unified view of what you're spending or which requests cost the most.
Built your whole app on one provider's SDK. Now they raise prices 40% and you're rewriting everything.
THEO SOLVES ALL FOUR ↓
Every request flows through five stages. Your prompt in, the perfect model out, automatically.
CLICK ANY STEP TO LEARN MORE
Intent classification in <1ms. The right model for every request, every time.
CLICK ANY MODEL TO EXPLORE
Regex fast-path classifies obvious intents in <1ms. Ambiguous prompts fall through to an LLM classifier that understands 'build me a dashboard' is a code task — no keyword required.
Simple questions route to Theo Fast (~$0.25/M tokens). Complex reasoning routes to Theo Think (~$3/M tokens). You save ~90% on routine queries without any configuration.
Server-side keyword matching runs before the LLM classifier. If patterns match with high confidence, classification completes in under 1ms with zero model calls.
Redis-backed circuit breaker (5% error threshold, 30s trip) routes through OpenRouter → Anthropic Direct → Gemini Direct. Each model also has a per-mode fallback. Your users never see downtime.
Real scenarios. Real code. See what Theo makes possible.
Ship an intelligent chatbot in your SaaS product. Theo picks the right model per message - fast responses for simple questions, deep reasoning for complex ones.
Generate blog posts, images, and social media cards from a single prompt. Theo chains the right modes: Code for copy, Image for visuals, Fast for meta tags.
Build support that remembers. Theo persists user context across sessions and channels - your customer never repeats themselves.
Let your team generate dashboards, scripts, and data pipelines with natural language. Theo Code writes production-ready code with 128K output tokens.
One brain, every channel. Deploy the same AI assistant to Telegram, WhatsApp, and your web app. Same memory, same skills, same routing.
Type a prompt and watch Theo classify your intent in real-time. See which model gets selected and why.
TRY A PRESET OR TYPE YOUR OWN
import { Theo } from "@hitheo/sdk";
const theo = new Theo({ apiKey: "theo_sk_..." });
const result = await theo.complete({
prompt: "Compare term life quotes",
mode: "auto",
});
// → Theo Code (auto)npm install @hitheo/sdk - then pick your use case.
OPEN SOURCE · TYPESCRIPT · FULL DOCS
import { Theo } from "@hitheo/sdk";
const theo = new Theo({ apiKey: "theo_sk_..." });
const res = await theo.complete({
prompt: "Explain quantum computing simply",
mode: "auto", // → routes to Theo Fast (quick chat)
});
console.log(res.text);
console.log(res.model); // "theo-fast"
console.log(res.cost); // "$0.0003"Your persona. Your skills. Theo's engine under the hood.
Replace Theo's personality with your own brand. Your users never see Theo — they see your AI.
Install marketplace skills for domain expertise. Pass inline tools for your product's actions.
Multi-model routing, intent classification, agent loop, memory, billing, caching — all handled.
const model = pickModel(prompt);
const tools = resolveTools(user);
const system = buildPrompt(persona, skills);
const res = await openrouter.complete({ model, ... });
const toolResults = await executeTools(res);
const final = await openrouter.complete({ ... });
await debitCredits(user, res.usage);
await logAudit(user, res);import { Theo } from "@hitheo/sdk";
const theo = new Theo({ apiKey: process.env.THEO_KEY });
const evi = theo.evi({
persona: "You are Eva, AcmeCorp's AI...",
skills: ["insurance-quoting"],
tools: [lookupCustomer, updatePolicy],
});
const res = await evi.complete({ prompt: "..." });ONE API CALL. EVERYTHING HANDLED.
Intent classification. Model routing. Skill injection. Tool execution. Memory. Streaming. Billing. Audit. Caching. All from evi.complete().
An app store for AI capabilities. Install domain skills that give Theo specialized knowledge - or build and publish your own.
BROWSE · INSTALL · PUBLISH
Compare rates across 5+ carriers in real-time. Pre-built data connectors, carrier-specific rate tables, and compliance guardrails.
Analyze PRs for security vulnerabilities, performance issues, and architecture anti-patterns. GitHub, GitLab, Bitbucket.
Multi-step web research with source synthesis. Crawl, read, and produce cited reports, not just summaries.
OCR + structured extraction from PDFs, images, and scanned documents. Returns typed JSON, not raw text.
Blog posts, emails, social copy. Adapts tone to your brand voice. SEO optimization built-in.
Define custom tools, knowledge bases, and workflows. Publish to the store or keep private to your org.
import { defineSkill } from "@hitheo/sdk";
export default defineSkill({
name: "inventory-check",
description: "Real-time inventory lookup and reorder alerts",
tools: [checkStock, reorderAlert, getSupplierPrices],
knowledge: ["./product-catalog/*.json"],
});
// Publish to the store:
// npx hitheo publish --public
// Other developers can install it instantly.EARN FROM YOUR SKILLS
Publish skills to the store. Other developers install and use them. You earn credits for every API call that activates your skill.
Install domain skills that give Theo specialized knowledge. Like an app store for AI capabilities.
CLICK ANY SKILL TO SEE MORE
Channel-agnostic core. Same brain, same memory, same skills - every channel. Open-source adapters. npm install and you're live.
@hitheo/telegram@hitheo/whatsapp@hitheo/sdkOne dashboard for everything. Keys, usage, logs, routing config. No more switching between 4 provider consoles.
You could wire up multiple AI providers yourself. Here's what that looks like vs. Theo.
import ProviderA from "provider-a-sdk";
import ProviderB from "provider-b-sdk";
import ProviderC from "provider-c-sdk";
// Init 3 SDKs
const a = new ProviderA({ apiKey: KEY_A });
const b = new ProviderB({ apiKey: KEY_B });
const c = new ProviderC({ apiKey: KEY_C });
// Manual routing logic
function pickModel(prompt: string) {
if (isCodeTask(prompt)) return "a";
if (isImageTask(prompt)) return "c";
return "b"; // default
}
// Manual failover
try {
const model = pickModel(userPrompt);
if (model === "a") {
res = await a.messages.create({...});
} else if (model === "c") {
res = await c.models.generateContent({...});
} else {
res = await b.chat.completions.create({...});
}
} catch (e) {
// Try fallback... manually
res = await b.chat.completions.create({...});
}import { Theo } from "@hitheo/sdk";
const theo = new Theo({ apiKey: "theo_sk_..." });
const res = await theo.complete({
prompt: userPrompt,
mode: "auto",
});
// Routing, failover, cost optimization,
// model selection - all handled.SELECT A COMPETITOR TO COMPARE
to start
Then pay-as-you-go per token. Passthrough model costs plus a small platform fee. No minimums. No commitments.
CREATE ACCOUNTEARLY ACCESS FEEDBACK
“We replaced 3 AI SDKs and 200 lines of routing logic with 6 lines of Theo. Our response times dropped 40% because Theo actually picks the right model.”
“The memory system is what sold us. Our support bot remembers context across Telegram and web. Customers stopped having to repeat themselves.”
“I built a custom insurance quoting skill in an afternoon. My agents now compare 5 carriers in real-time through a Telegram bot. That would have been a 3-month project.”
One API key. 300+ models. Intelligent routing. Built-in memory. Extensible skills. Open-source SDK.
Start free in 30 seconds. No credit card. Your users get better answers. You write less code.