Everything for AI app frontends and backends

import { ai } from "@/gencow/ai";

const reply = await ai.chat({
  model: "gpt-4o",
  system: "You are a helpful assistant.",
  messages: chatHistory,
});

// Agent Loop (auto tool calling)
const result = await ai.agent({
  messages, tools, maxSteps: 5,
});

import { rag } from "@/gencow/rag";

// Document indexing (auto chunk + embed)
await rag.ingest(ctx, "manual.pdf", docText);

// Semantic search
const results = await rag.search(ctx,
  "What is the refund policy?",
  { filter: { source: "manual.pdf" } }
);

// Q&A — search + answer in one call
const { answer } = await rag.ask(ctx, query);

import { defineTools } from "@/gencow/tools";

const tools = defineTools(ctx, {
  getOrder: {
    description: "Look up order status",
    parameters: z.object({
      orderId: z.string()
    }),
    handler: async (ctx, args) => {
      return ctx.db.select().from(orders)
        .where(eq(orders.id, args.orderId));
    }
  }
});

await ai.chat({ messages, tools });

import { memory } from "@/gencow/memory";

// Auto-compose memory context
const memCtx = await memory.buildContext(
  ctx, userId, sessionId, query
);

const reply = await ai.chat({
  system: memCtx.toSystemPrompt(),
  messages: [
    ...memCtx.recentMessages,
    { role: "user", content: query },
  ],
});

import { guardrails } from "@/gencow/guardrails";

// PII masking + topic blocking
const safe = await guardrails.validateInput(
  userMsg,
  { maskPII: true, blockTopics: ["politics"] }
);
if (!safe.allowed) return safe.blocked;

// Wrapping utility
const result = await guardrails.wrap(
  (text) => ai.chat({ messages: [{ role: "user", content: text }] }),
  userMsg,
  { maskPII: true },
  { maxLength: 2000 }
);

// Auto-collected (no extra code needed)
// View in Dashboard:
//
// ◆ Agent response time
// ◆ RAG search hit rate
// ◆ Token usage & cost
// ◆ Failed conversation logs