const res = await claude.messages.create({ model, tools, max_tokens: 4096 })embedding[1536] · cosine_sim = 0.9124agent.run(task) → tool_call: query_ledger()
const res = await claude.messages.create({ model, tools, max_tokens: 4096 })embedding[1536] · cosine_sim = 0.9124agent.run(task) → tool_call: query_ledger()
public function postJournalEntry(Ledger $ledger): voidSELECT * FROM invoices WHERE zatca_status = 'cleared'await api.get('/erp/gl/trial-balance')
public function postJournalEntry(Ledger $ledger): voidSELECT * FROM invoices WHERE zatca_status = 'cleared'await api.get('/erp/gl/trial-balance')
GEN AIAI AGENTSAGENTIC AILLM · RAG · MCPTOOL USEVECTOR SEARCH
GEN AIAI AGENTSAGENTIC AILLM · RAG · MCPTOOL USEVECTOR SEARCH
if (!$invoice->isCleared()) throw new ZatcaException($uuid)GET /api/v2/stock/movements?warehouse=RUH-01 → 20011001010 · 01101101
if (!$invoice->isCleared()) throw new ZatcaException($uuid)GET /api/v2/stock/movements?warehouse=RUH-01 → 20011001010 · 01101101
softmax(qkᵀ/√d)·vtokens_in 2,418 · tokens_out 512 · latency 840mspgvector <-> ORDER BY embedding <=> $1 LIMIT 8RAG: 6 chunks retrieved
softmax(qkᵀ/√d)·vtokens_in 2,418 · tokens_out 512 · latency 840mspgvector <-> ORDER BY embedding <=> $1 LIMIT 8RAG: 6 chunks retrieved
$ pm2 reload erp-api --update-envRedis::pipeline(fn ($pipe) => $pipe->hset('stock', $sku, $qty))0x1F · CRC OK · 200
$ pm2 reload erp-api --update-envRedis::pipeline(fn ($pipe) => $pipe->hset('stock', $sku, $qty))0x1F · CRC OK · 200
DB::transaction(fn () => $this->postToLedger($batch))flutter build apk --release --flavor prodCREATE INDEX idx_gl_period ON journal_entries (period_id)
DB::transaction(fn () => $this->postToLedger($batch))flutter build apk --release --flavor prodCREATE INDEX idx_gl_period ON journal_entries (period_id)
const [state, dispatch] = useReducer(erpReducer, initial)docker compose up -d --buildmigrate:fresh --seed · 128 tables
const [state, dispatch] = useReducer(erpReducer, initial)docker compose up -d --buildmigrate:fresh --seed · 128 tables
fine_tune(dataset="erp-support-tickets", epochs=3)0.0412 loss · 0.981 val_accinference: batch=32 · gpu:0
fine_tune(dataset="erp-support-tickets", epochs=3)0.0412 loss · 0.981 val_accinference: batch=32 · gpu:0
live
