new structure for chat and values

packages/ai/src/steps/extract-values-search-step.test.ts

@@ -19,17 +19,16 @@ vi.mock('braintrust', () => ({
   wrapAISDKModel: vi.fn((model) => model),
 }));
 
-// Create a ref object to hold the mock generate function
-const mockGenerateRef = { current: vi.fn() };
+// Mock the AI SDK
+vi.mock('ai', () => ({
+  generateObject: vi.fn(),
+}));
 
-// Mock the Agent class from Mastra with the generate function
+// Mock Mastra
 vi.mock('@mastra/core', async () => {
   const actual = await vi.importActual('@mastra/core');
   return {
     ...actual,
-    Agent: vi.fn().mockImplementation(() => ({
-      generate: (...args: any[]) => mockGenerateRef.current(...args),
-    })),
     createStep: actual.createStep,
   };
 });
@@ -41,18 +40,17 @@ import { extractValuesSearchStep } from './extract-values-search-step';
 
 // Import the mocked functions
 import { generateEmbedding, searchValuesByEmbedding } from '@buster/stored-values/search';
+import { generateObject } from 'ai';
 
 const mockGenerateEmbedding = generateEmbedding as ReturnType<typeof vi.fn>;
 const mockSearchValuesByEmbedding = searchValuesByEmbedding as ReturnType<typeof vi.fn>;
-
-// Access the mock generate function through the ref
-const mockGenerate = mockGenerateRef.current;
+const mockGenerateObject = generateObject as ReturnType<typeof vi.fn>;
 
 describe('extractValuesSearchStep', () => {
   beforeEach(() => {
     vi.clearAllMocks();
     // Set default mock behavior
-    mockGenerate.mockResolvedValue({
+    mockGenerateObject.mockResolvedValue({
       object: { values: [] },
     });
   });
@@ -72,7 +70,7 @@ describe('extractValuesSearchStep', () => {
       runtimeContext.set('dataSourceId', 'test-datasource-id');
 
       // Mock the LLM response for keyword extraction
-      mockGenerate.mockResolvedValue({
+      mockGenerateObject.mockResolvedValue({
         object: { values: ['Red Bull', 'California'] },
       });
 
@@ -141,7 +139,7 @@ describe('extractValuesSearchStep', () => {
       runtimeContext.set('dataSourceId', 'test-datasource-id');
 
       // Mock empty keyword extraction
-      mockGenerate.mockResolvedValue({
+      mockGenerateObject.mockResolvedValue({
         object: { values: [] },
       });
 
@@ -195,7 +193,7 @@ describe('extractValuesSearchStep', () => {
       runtimeContext.set('dataSourceId', 'test-datasource-id');
 
       // Mock successful keyword extraction
-      mockGenerate.mockResolvedValue({
+      mockGenerateObject.mockResolvedValue({
         object: { values: ['Red Bull'] },
       });
 
@@ -226,7 +224,7 @@ describe('extractValuesSearchStep', () => {
       runtimeContext.set('dataSourceId', 'test-datasource-id');
 
       // Mock LLM extraction success but embedding failure
-      mockGenerate.mockResolvedValue({
+      mockGenerateObject.mockResolvedValue({
         object: { values: ['test keyword'] },
       });
 
@@ -254,7 +252,7 @@ describe('extractValuesSearchStep', () => {
       runtimeContext.set('dataSourceId', 'test-datasource-id');
 
       // Mock successful keyword extraction
-      mockGenerate.mockResolvedValue({
+      mockGenerateObject.mockResolvedValue({
         object: { values: ['test keyword'] },
       });
 
@@ -284,7 +282,7 @@ describe('extractValuesSearchStep', () => {
       runtimeContext.set('dataSourceId', 'test-datasource-id');
 
       // Mock two keywords: one succeeds, one fails
-      mockGenerate.mockResolvedValue({
+      mockGenerateObject.mockResolvedValue({
         object: { values: ['keyword1', 'keyword2'] },
       });
 
@@ -327,7 +325,7 @@ describe('extractValuesSearchStep', () => {
       runtimeContext.set('dataSourceId', 'test-datasource-id');
 
       // Mock everything to fail
-      mockGenerate.mockRejectedValue(new Error('LLM failure'));
+      mockGenerateObject.mockRejectedValue(new Error('LLM failure'));
       mockGenerateEmbedding.mockRejectedValue(new Error('Embedding failure'));
       mockSearchValuesByEmbedding.mockRejectedValue(new Error('Database failure'));
 
@@ -378,7 +376,7 @@ describe('extractValuesSearchStep', () => {
       runtimeContext.set('dataSourceId', 'test-datasource-id');
 
       // Mock successful keyword extraction
-      mockGenerate.mockResolvedValue({
+      mockGenerateObject.mockResolvedValue({
         object: { values: ['Red Bull'] },
       });
 
@@ -437,7 +435,7 @@ describe('extractValuesSearchStep', () => {
       runtimeContext.set('dataSourceId', 'test-datasource-id');
 
       // Mock successful keyword extraction
-      mockGenerate.mockResolvedValue({
+      mockGenerateObject.mockResolvedValue({
         object: { values: ['test'] },
       });
 

packages/ai/src/steps/extract-values-search-step.ts

@@ -1,7 +1,8 @@
 import type { StoredValueResult } from '@buster/stored-values';
 import { generateEmbedding, searchValuesByEmbedding } from '@buster/stored-values/search';
-import { Agent, createStep } from '@mastra/core';
+import { createStep } from '@mastra/core';
 import type { RuntimeContext } from '@mastra/core/runtime-context';
+import { generateObject } from 'ai';
 import type { CoreMessage } from 'ai';
 import { wrapTraced } from 'braintrust';
 import { z } from 'zod';
@@ -12,6 +13,11 @@ import type { AnalystRuntimeContext } from '../workflows/analyst-workflow';
 
 const inputSchema = thinkAndPrepWorkflowInputSchema;
 
+// Schema for what the LLM returns
+const llmOutputSchema = z.object({
+  values: z.array(z.string()).describe('The values that the agent will search for.'),
+});
+
 // Step output schema - what the step returns after performing the search
 export const extractValuesSearchOutputSchema = z.object({
   values: z.array(z.string()).describe('The values that the agent will search for.'),
@@ -231,12 +237,6 @@ async function searchStoredValues(
   }
 }
 
-const valuesAgent = new Agent({
-  name: 'Extract Values',
-  instructions: extractValuesInstructions,
-  model: Haiku35,
-});
-
 const extractValuesSearchStepExecution = async ({
   inputData,
   runtimeContext,
@@ -264,12 +264,19 @@ const extractValuesSearchStepExecution = async ({
     try {
       const tracedValuesExtraction = wrapTraced(
         async () => {
-          const response = await valuesAgent.generate(messages, {
-            maxSteps: 0,
-            output: extractValuesSearchOutputSchema,
+          const { object } = await generateObject({
+            model: Haiku35,
+            schema: llmOutputSchema,
+            messages: [
+              {
+                role: 'system',
+                content: extractValuesInstructions,
+              },
+              ...messages,
+            ],
           });
 
-          return response.object;
+          return object;
         },
         {
           name: 'Extract Values',

packages/ai/src/steps/generate-chat-title-step.ts

@@ -1,6 +1,7 @@
 import { updateChat, updateMessage } from '@buster/database';
-import { Agent, createStep } from '@mastra/core';
+import { createStep } from '@mastra/core';
 import type { RuntimeContext } from '@mastra/core/runtime-context';
+import { generateObject } from 'ai';
 import type { CoreMessage } from 'ai';
 import { wrapTraced } from 'braintrust';
 import { z } from 'zod';
@@ -11,6 +12,12 @@ import type { AnalystRuntimeContext } from '../workflows/analyst-workflow';
 
 const inputSchema = thinkAndPrepWorkflowInputSchema;
 
+// Schema for what the LLM returns
+const llmOutputSchema = z.object({
+  title: z.string().describe('The title for the chat.'),
+});
+
+// Schema for what the step returns (includes pass-through data)
 export const generateChatTitleOutputSchema = z.object({
   title: z.string().describe('The title for the chat.'),
   // Pass through dashboard context
@@ -28,13 +35,9 @@ export const generateChatTitleOutputSchema = z.object({
 
 const generateChatTitleInstructions = `
 I am a chat title generator that is responsible for generating a title for the chat.
-`;
 
-const todosAgent = new Agent({
-  name: 'Extract Values',
-  instructions: generateChatTitleInstructions,
-  model: Haiku35,
-});
+The title should be 3-8 words, capturing the main topic or intent of the conversation.
+`;
 
 const generateChatTitleExecution = async ({
   inputData,
@@ -63,12 +66,19 @@ const generateChatTitleExecution = async ({
     try {
       const tracedChatTitle = wrapTraced(
         async () => {
-          const response = await todosAgent.generate(messages, {
-            maxSteps: 0,
-            output: generateChatTitleOutputSchema,
+          const { object } = await generateObject({
+            model: Haiku35,
+            schema: llmOutputSchema,
+            messages: [
+              {
+                role: 'system',
+                content: generateChatTitleInstructions,
+              },
+              ...messages,
+            ],
           });
 
-          return response.object;
+          return object;
         },
         {
           name: 'Generate Chat Title',