The most powerful serverless platform for building AI agents. Build. Deploy. Scale.
Searching memory "langbase-docs", "command-code-docs"
Agentic re-ranking the most relevant context
Citation: Used "command-code-docs" content "Taste"
Commad Code is a frontier coding agent with Meta Neuro-Symbolic AI model `taste-1` that can continuously learns your coding taste.
Learns the taste of your code (explicit & implicit feedback).
taste-1 enforces the invisible logic of your choices and taste.
Share your taste to build consistent code using npx taste push/pull.
Learns the taste of your code (explicit & implicit feedback).
taste-1 enforces the invisible logic of your choices and taste.
Share your taste to build consistent code using npx taste push/pull.
Searching memory "langbase-docs"
Agentic re-ranking the most relevant context
Citation: Used "langbase-docs" content "AI Memory"
Memory turns RAG into a simple, agentic, & serverless API for developers.
Comes with vector store, file storage, and best in class retrieval engine.
Scalable RAG that just works. Chat with your docs, repos, or any data.
Engineered with accuracy. Get context-aware answers you can trust.
Comes with vector store, file storage, and best in class retrieval engine.
Scalable RAG that just works. Chat with your docs, repos, or any data.
Engineered with accuracy. Get context-aware answers you can trust.
Searching user memory "orders"
Tool: Checking delivery status with UPS…
Citation: Found data for "Order #786-8337390"
1import {Langbase} from 'langbase';
2import 'dotenv/config';
3
4const langbase = new Langbase({ apiKey: process.env.LANGBASE_API_KEY!});
5
6async function main() {
7 const response = await langbase.pipe.run({
8 name: 'support-assistant',
9 memory: [{ name: 'orders' }],
10 messages: [{
11 role: 'user',
12 content: 'When will my last order be delivered?'
13 }]
14 });
15
16 console.log(response);
17}
18
19main();Searching memory "langbase-docs"
Agentic re-ranking the most relevant context
Citation: Used "langbase-docs" content "AI Agents Primitive"
AI Agents as a serverless API. Pick any LLM. Connect tools. Deploy at scale.
One API. 600+ LLMs. Switch Providers & LLMs with one line of code.
Extend what your agent can do with search, crawl & MCP tools.
Deploy agents and apps in one click. Serverless & infinitely scalable API.
One API. 600+ LLMs. Switch Providers & LLMs with one line of code.
Extend what your agent can do with search, crawl & MCP tools.
Deploy agents and apps in one click. Serverless & infinitely scalable API.
Loading your pipe agent "release-assistant"
Tool call: Fetched your latest GitHub PRs
Sturctured data extraction from PRs for changelog items
Trace: Used "release-assistant" pipe agent and "GitHub" tool
1import 'dotenv/config';
2import {Langbase} from 'langbase';
3
4const langbase = new Langbase({
5 apiKey: process.env.LANGBASE_API_KEY!,
6});
7
8async function main() {
9 const response = await langbase.pipe.run({
10 name: 'release-assistant',
11 messages: [{
12 role: 'user',
13 content: 'Summarize my GitHub PRs and create a changelog draft?'
14 }]
15 });
16
17 console.log(response);
18}
19main();Searching memory "langbase-docs"
Agentic re-ranking the most relevant context
Citation: Used "langbase-docs" content "Workflow Primitive"
Powerful multi-step & multi-agent AI Workflows with timeouts and retries.
Vibe code prod-ready AI agent workflows using Command.
Steps are durable with retries, backoff, and scheduling
Step-by-step traceability. Pinpoint issues and optimize easily.
Vibe code prod-ready AI agent workflows using Command.
Steps are durable with retries, backoff, and scheduling
Step-by-step traceability. Pinpoint issues and optimize easily.
Workflow: Fetch, Analyze, Respond.
Step 1: Analyzing sentiment…
Step 3: Determining if response is required…
Step 4: Generating reply (if needed)…
1import 'dotenv/config';
2import { Langbase } from 'langbase';
3
4async function processEmail({ emailContent }: { emailContent: string }) {
5 const langbase = new Langbase({
6 apiKey: process.env.LANGBASE_API_KEY!,
7 });
8
9 const workflow = langbase.workflow();
10 const { step } = workflow;
11
12 try {
13 const [summary, sentiment] = await Promise.all([
14 step({
15 id: 'summarize_email',
16 run: async () => {
17 const { output } = await langbase.agent.run({
18 model: 'openai:gpt-5-mini',
19 instructions: 'Create a concise summary of this email.',
20 apiKey: process.env.LLM_API_KEY!,
21 input: [{ role: 'user', content: emailContent }],
22 stream: false,
23 });
24 return output;
25 },
26 }),
27 step({
28 id: 'analyze_sentiment',
29 run: async () => {
30 const { output } = await langbase.agent.run({
31 model: 'openai:gpt-5-mini',
32 instructions: 'Analyze the sentiment of this email.',
33 apiKey: process.env.LLM_API_KEY!,
34 input: [{ role: 'user', content: emailContent }],
35 stream: false,
36 });
37 return output;
38 },
39 }),
40 ]);
41
42 const responseNeeded = await step({
43 id: 'determine_response_needed',
44 run: async () => {
45 const { output } = await langbase.agent.run({
46 model: 'openai:gpt-5-mini',
47 instructions: 'Determine if a response is needed.',
48 apiKey: process.env.LLM_API_KEY!,
49 input: [{
50 role: 'user',
51 content: `Email: ${emailContent}\nSummary: ${summary}\nSentiment: ${sentiment}\nDoes this require a response?`
52 }],
53 stream: false,
54 });
55 return output.toLowerCase().includes('yes');
56 },
57 });
58
59 let response = null;
60 if (responseNeeded) {
61 response = await step({
62 id: 'generate_response',
63 run: async () => {
64 const { output } = await langbase.agent.run({
65 model: 'openai:gpt-5-mini',
66 instructions: 'Generate a professional email response.',
67 apiKey: process.env.LLM_API_KEY!,
68 input: [{
69 role: 'user',
70 content: `Email: ${emailContent}\nSummary: ${summary}\nSentiment: ${sentiment}\nDraft a response.`
71 }],
72 stream: false,
73 });
74 return output;
75 },
76 });
77 }
78
79 return { summary, sentiment, responseNeeded, response };
80 } finally {
81 await workflow.end();
82 }
83}
84
85async function main() {
86 const sampleEmail = `Subject: Pricing Information and Demo Request
87
88Hello,
89
90I came across your platform and I'm interested in learning more about your product for our growing company. Could you please send me some information on your pricing tiers?
91
92We're particularly interested in the enterprise tier as we now have a team of about 50 people who would need access. Would it be possible to schedule a demo sometime next week?
93
94Thanks in advance for your help!
95
96Best regards,
97Jamie`;
98 const results = await processEmail({ emailContent: sampleEmail });
99 console.log(JSON.stringify(results, null, 2));
100}
101
102main();Trace: Tracing all steps.
Agent: "Chat with PDF"
Step 1: `retrieve_pdf_content` from memory
Evaluate and collaborate to see exactly what your agents are doing at every step.
Create, version, collaborate, and monitor agents in our AI Studio.
Trace everything your agents do. Predict cost, usage, and run evals.
World-class developer experience for AI engineering collab like GitHub.
Create, version, collaborate, and monitor agents in our AI Studio.
Trace everything your agents do. Predict cost, usage, and run evals.
World-class developer experience for AI engineering collab like GitHub.
"Langbase is transforming the AI market. Easy to use, handy integrations, and serverless AI agents infra. What else could we ask for."
What developers and founders are saying about Langbase.
“The real breakthrough here is how easily we can test RAG via Langbase memory agents—actually seeing which chunks get retrieved for specific queries. That kind of visibility just isn't available with other providers. Typically, you'd need specialized vendor knowledge, framework-centric coding, or custom actions to get the same result. But with Langbase, there's next to no overhead or prior expertise needed. The straightforward setup, developer experience, collab features, and built-in version control instantly give it a huge plus for anyone building AI-driven apps. Langbase basically turned me into an Al engineer overnight.”
"Ahmad is uniquely positioned to dramatically improve the AI developer experience.He has done exactly that with Langbase, building on his deep expertise creating products for developers"
"LLM's are redefining the meaning of an application andLangbase is the Vercel of AI that supercharges every developer & company's efforts in building for this new wave."
"Langbase AI serverless dev experience is powerful and unique, truly designed to meet the needs of developers building and operating LLM apps."
"Langbase lets us manage all our LLM-related infrastructure in one place, quick-iteration, actionable analytics, version controlled prompts, and rapid testing of different LLM models. Read more in FQ's customer story."
"Really impressed with Langbase (we use at Ignition) - it's one of the most "need to have" tools i've seen in the past decade. AI is moving so quickly so a serverless composable infra to mix / match / test / deploy new models as they are released is the fastest way for an org to stay on the bleeding edge without vendor lock-in. Seeinghow excited my team has been using Langbase.Langbase is simplifying the complexity of it all."
"Langbase is unique for its composable serverless AI cloud that just works . Makes AI dev dead simple for everyone, not just the ML experts. I believe an AI Pipe is the easiest way to build AI features you can actually use.Build, ship, and innovate with zero-config, making ship happen."
"Langbase is transforming AI development with serverless AI agents infrastructure, making it easy for any developer to build, collaborate, and deploy AI apps. Think Docker containers, but for AI agents! Proud to have supported them from the beginning."
Build, test, & deploy in minutes. Scale your agents instantly, with built-in
memory and tooling.