Shensi Ding · Launch Video Breakdown: Hook, Pacing & Motion Design

Scene-by-scene timeline & spoken transcript

1. 0:00 – 0:10

   The Hook

   The AI Integration Challenge

   “You added AI features to your product. Now you need to allow your customers to set their own routing preferences and pick their own models. Routing policies, provider integrations, all on your team to build. Allow your customers to set up their own LLM key management, and pass through costs. All on your team to build. Or you can embed the Merge Embedded Routing Stack. Route across LLMs. Optimize for cost, performance,”

   On screen

   AI Configuration Providers and models available in this workspace Search providers and models Select all providers and models OpenAI (5) > 5 MODELS 3 selected Anthropic (4) > GPT-5.6 Sol 128K vision Google (4) > GPT-5.5 Meta (3) > GPT-5.2 Mistral (4) > GPT-5.3 xAI (2) > GPT-5.4 1M context DeepSeek (2) Cohere (2) Amazon (3) ROUTING PREFERENCE Cost Performance Capability Claude Sonnet 5 800 ms Applies to this workspace or Cancel Save VANTLE Hi! CORDIA Team Daniel Osei Program Manager NUMERA Priya Nair Project director R CORDIA AI Configuration Providers and models available in this workspace Search providers and models Select all providers and models Anthropic (5) > 4 MODELS 3 selected Claude Fable 5 200K Google (4) > Claude Opus 4.8 200K Meta (3) > Claude Haiku 4.5 200K Mistral (4) > Claude Sonnet 5 200K xAI (2) DeepSeek (2) Cohere (2) Amazon (3) ROUTING PREFERENCE Cost Performance Capability Claude Sonnet 5 800 ms Applies to this workspace or Cancel Save Route to the most cost-efficient model for customer A. Do no allow for any traffic in China today. Always select the most efficient model. Keep US traffic on US-hosted endpoints. Send Mistral Large for everything else. Routing policies Provider integrations // customer-setup.ts - runs when a customer connects their AI set async function setupCustomerAI(customer) { // 1. Create customer const created = await merge.customers.create({ name: customer acme co, origin Id: customer acme co, }); const customerId = created.Id, "yT8b/c6d 5e4t 4a3b 8c2d 1e0t9a8b/c6d" } "Acme_co_customer_1" "id": "7ed02f04-3916-4c48-a4e6-2b7225090d0", "name": "Acme_co_customer_1", "routing_policy_strategy": "PRIORITY", "models": ["Claude Opus 4.8", "GPT-5.2"], "usage": 184320, "Acme_co_customer_2" "id": "0e8d6884-bc11-4f15-b72d-818a3c5c7422", "name": "Acme_co_customer_2", "routing_policy_strategy": "INTELLIGENT", "models": ["Claude Sonnet 5", "GPT-5.4 Mini"], "usage": 521170, "Acme_co_customer_3" "id": "ef17e1c5-26ab-4eeb-9e9f-59f18f54e6c", "name": "Acme_co_customer_3", "routing_policy_strategy": "PRIORITY", "models": ["Gemini 2.5 Flash", "Claude Haiku"], "usage": 9803, "Acme_co_customer_4" "id": "aadcce3e-592f-4448-a36a-58f75a23f166", "name": "Acme_co_customer_4", "routing_policy_strategy": "INTELLIGENT", "models": ["Claude Opus 4.6", "GPT-5.5"], "usage": 271644, 1 Anthropic Deepseek v4 Pro GPT-5.4 Grok 4.3 Best fit Gemini 3 Muse Spark 1.1 Claude Sonnet 4.6 GLM-5.2 Optimize model routing Cost Performance Capability Maximize output quality and logic Customer A Allowed providers Choose which providers and models are eligible. Provider Models Anthropic Claude Opus 4.8, Claude Sonnet 4.6 OpenAI GPT-5.6, GPT-5.5, GPT-5.2

2. 0:10 – 0:20

   Product Reveal

   Merge Embedded Routing Stack

   “capability. Your customers pick the strategy, every customer runs on their own preferences, own policies, and can use their own keys. Pull routing performance, usage, and cost. Pass it straight through to their invoice. They configure it inside your product. Every customer gets the control they need. Merge is the connective infrastructure for AI.”

   On screen

   Optimize model routing Cost Performance Capability Maximize output quality and logic Customer A Allowed providers Choose which providers and models are eligible. Provider Models Anthropic Claude Opus 4.8, Claude Sonnet 4.6 OpenAI GPT-5.6, GPT-5.5, GPT-5.2 OpenAI GPT-4.1 mini, GPT-4o mini Anthropic Claude Haiku Google Gemini Flash / Flash-Lite 1 PREFERENCES Routing strategy Priority Intelligent Build Your Own Router Model priority 1 Gemini 2.5 Flash Fast 2 GPT-4.1 mini Balanced 3 Claude 3.5 Haiku Low latency Enable fallback if primary fails Fallback model Lumen 3 Max latency target 800 ms 2 POLICIES Data residency EU-hosted only PII protection Rate limit 0 Use your own API keys Customer Usage API $ curl https://api $ curl https://api-gateway.merge.dev/v1/customers/ customer_id/usage { "customer_byok_spend": 124, "organization_byok_spend": 13, "merge_spend": 879, "request_count": 12M, "per_model": [ { "provider": "openai", "spend": 7.42 }, { "per_routing_policy": [ { "routing_policy": "2a1b3c4d-5e6f-478b-790c-1d2e3f4a5b6c", "spend": 7.42 } ] } ] } INVOICE Customer B July 2025 AI request (48.21B) Premium model usage 9.2M tokens Total $1,284.06 Passed through from your backend 2025-07-07T21:18:49Z VANTLE I want to only use Anthropic models I don't want to route through China I want to prioritize quality over latency I want to test prompts before production I want every team to use the same policies Handle failures gracefully Block jailbreak prompts automatically Notify me when latency spikes. We need audit logs for every request. I want detailed logs for every API request I want alerts on failed requests I want to set a budget limit per team I want to control max output tokens I want data to stay in the EU I want SSO for my whole team connective infrast