New Features • Added Confluence Search Tool that uses a hybrid approach combining native API and semantic search for comprehensive content retrieval, supporting both content and title searches while filtering duplicates. • Added Notion Search Tool that enables users to search across Notion pages by both title and content, integrating this functionality into the workflow system. • Implemented AI-powered Notion search functionality, allowing users to generate reports with references and relevant summaries from existing Notion pages.
Enhancements Deep research workflow: • Enhanced workflow sharing model to implement personalized job status visibility and report privacy, ensuring reports are only visible to users who trigger workflows. • Extended search functionality in Deep Research Workflow to include workflow names and templates beyond just usernames and made search work across Templates and My Workflows tabs. • Increased the number of visible workflow tiles from 10 to 100 to eliminate pagination issues, allowing users to see more workflows at once. • Updated workflow examples to be more general purpose rather than JIRA-focused, making them more relevant for all users regardless of which tools they use. • Changed workflow reports to be personal instead of shared, ensuring that reports are only visible to the user who triggered the workflow for improved privacy. • Modified the workflow description field to be a shorter text box with a maximum of two lines, enhancing readability and overall UI neatness.
Slack bot related: • Improved Slack bot appearance by updating the icon and refining message text, enhancing visual consistency and user communication. • Customized Slack at-ayraa messaging to reflect currently connected apps, providing users with contextual information about their integration status.
+8 other enhancements
Bug Fixes • Fixed formatting issues in Slack DM reports for DRW by improving markdown formatting, resulting in more readable and professional-looking reports. • Fixed DRW failure issue, where reports were showing as failed and sending error messages to users due to keyword format inconsistencies. • Fixed two issues in the workflow text box: cursor jumping to the bottom when editing and formatting problems with bullet points and numbered lists. • Fixed missing references for JIRA and Slack in Deep research Workflow reports by implementing token limit handling and processing references in chunks. • Resolved hallucination issues in Deep research Workflow reports where AI was incorrectly reporting non-existent information, particularly for JIRA project details. • Fixed "Copy to clipboard" functionality for Deep Research Workflow, which was producing unreadable formatted text, now ensuring clean, readable content when pasted anywhere. • Fixed JIRA indexing in Recent Mode for opensearch, ensuring new tickets are properly indexed and searchable within 1-2 hours of creation. • Fixed issue where workflows wouldn't execute when using the 'Run' option after editing instead of saving first, ensuring consistent execution.
New Features • Add Confluence Search Tool: – Introduces an integrated Confluence search capability that uses a hybrid approach (native API plus RM semantic search) to deliver more comprehensive content retrieval. • Add Meetings Tool (searching over meeting transcripts) – Adds a new capability for searching through meeting transcripts so that users can quickly retrieve discussion details; note that this feature remains blocked by a pending dependency in the backend integration but has been flagged as a new feature.
Enhancements
Deep Research Workflow • Remove "work" from email prompts – The UI text ("Enter your work email") has been updated by removing the term "work" to ensure clarity and consistency during signup and across related screens. • Decouple Workflow Save from Execution – Workflow creation now allows users to save without triggering immediate execution, thereby aligning with user expectations and design guidelines. • Personalize Workflow Reports – Workflow report visibility has been revised so that reports are always tied to the user who triggered the workflow rather than being shared broadly.
• Revise Critical Scan Query – The critical pricing scan query has been updated to use a non-collection approach, improving automation reliability when handling pricing structure queries. • Integrate LLM-Based Filtering – Enhanced connector search functionality now uses LLM-based filtering (especially for Slack), which reduces irrelevant noise and improves overall result quality.
+4 other minor enhancements
Bug Fixes A broad range of bug fixes has been implemented to address issues across workflows, reporting, and integrations:
• Address Pagination Issue in Workflow Tiles – Increased the data limit (from 10 to 100) so that users can scroll to view additional workflow tiles without a design rework. • Resolve Multiple UI/UX Issues in Workflow Pop-Up – Improvements include matching button sizes, proper spacing (10px gaps), and reduced font size in certain areas for better consistency. • Fix Mobile and Web Email Formatting for Workflow Reports – Email presentation has been refined to align with Figma designs across devices. • Ensure Correct data Appears in Workflow Reports – Adjustments to the JQL query have resolved issues where workflows were previously returning no data or references. • Address Ad Hoc Meeting Summary Delivery – Corrected an issue where only the host was receiving meeting summaries, ensuring that all qualified participants receive notifications. • Address Missing Jira References in Salesforce Responses – Resolved a token-related issue that was causing only a partial set of Jira references to appear. • Reinforce Personalization of Workflow Reports – Adjustments have been made to ensure that reports appear only for the creator even for shared workflows. • Restore Salesforce Query Functionality for Assist – Updated query logic now prevents errors and ensures Salesforce data is returned correctly. • Automate Sanity Checks Post Deployment – Automated critical scans now follow commit deployments, ensuring that new changes are promptly verified. • Fix Google Calendar Search Chat Issue – The search+chat feature now reliably returns responses for calendar-related queries.
Deep Research workflow - • Implemented real-time workflow status updates through automatic polling • Enabled sharing and using teammates' workflows for better collaboration • Enhanced AI to prioritize data related to specific people mentioned in queries • Added References section to Reports tab for better source tracking • Added capability to pull entire Slack threads for matching results in Deep Research • Launched Reports Tab backend infrastructure • Integrated Semantic Search functionality for more intuitive querying • Added Copy Link feature for Workflows and Reports for easy sharing • Added relevance filtering to Slack search tool, improving search result quality • Implemented validation step before sending final responses to ensure accuracy • Introduced scheduled run behavior for workflows, enabling automated execution • Added ability to pre-select apps in workflow templates for faster setup
Enhancements • Improved timeline dropdown to display selected value when expanded • Enhanced Reference List View with unique app logos for better visual clarity • Implemented form data retention when navigating between 'Manage Integration' and main form • Enhanced UI for displaying selected apps • Optimized workflow preview, edit, and duplicate view with clearer titles • Refined UI for workflow pages • Improved initial workflow experience with polished Day 0 screens • Implemented persistence for View Report button after report generation • Improved navigation by opening 'All' tab instead of 'Templates' when workflows exist
+10 other minor enhancements
Bug Fixes • Fixed formatting issues in Salesforce reports • Corrected source attribution in External Query follow-up responses • Resolved scheduler prompt issues when accessing scheduled workflows • Fixed report generation for Gmail workflowss • Fixed report link issues where View Report opened workflow templates instead • Fixed Deep Research failures for detailed real-world queries
New Features • Ayraa Workflows- Powered by Deep Research- Pilot Project: Implemented the Prrof of concept for Agentic Framework to automate workspace analysis and document generation • Enhanced LLM Integration: Added Sonnet 3.7 support with thinking flag configuration for improved reasoning capabilities • Advanced Prompt Processing: Implemented ability to parse outputs from LLM prompt tool, enabling more sophisticated workflows • Planning Phase Framework: Conducted POC of planning phase with various simulated user inputs to improve execution plans quality • Specialized Prompt Templates: Created generate_jira_jql and launch_notes_planner prompt templates to enhance AI functionality
Enhancements • Search Optimization: Enhanced keyword search tools for more comprehensive results. • Results Filtering: Added parameter to control whether to display all results or only most relevant ones across applications. • Tools Registry Optimization: Restructured tools categorization to improve efficiency • Responsive Design Improvements: Enhanced UI for 13-inch screens and fixed Teams scrollbar and cursor issues. • Advanced SFDC Query: Improved Salesforce query functionality with better prompting and categorization. • Profile Localization: Updated profile page terminology to be more US-friendly by changing "Designation" to "Role" and removing gender preference options. • Jira Tools Enhancement: Updated Jira tools to include labels and sprint fields in the input schema.
+2 other minor enhancements
Bug Fixes • At-Ayraa Improvements: Fixed multiple issues including responses hanging due to Slack character limit, latency in request handling, and thread context confusion. • Collections Functionality: Resolved multiple collections issues including non-clickable citations, editing functionality, and rapid app switching. • Meetings Transcripts: Fixed issue where stopping recording during meetings didn't provide transcripts for recorded portions. PULSE-10828 • Performance Improvements: Investigated and fixed 99% CPU utilization on RDS, improving overall system performance. • API Responses: Fixed incomplete SF-Tool API responses and corrected issue with opportunities links. • Dashboard Analytics: Fixed search count functionality when using recent mode filter. • Error Handling: Resolved various errors including Hubspot queries with Anytime filter, Jira text search, and workflow responses. • UI Refinements: Fixed search grid view text overflow, Gmail reconnect button color, and left menu thickness issues.
New Features • Redesigned Day 0 empty screen with improved header visuals and enhanced call-to-action styling, making first-time usage more intuitive. • Introduced a new Bedrock prompt for AI summarization compatible with multiple LLM models, enhancing the quality of our summary capabilities. • Added support for rich-text formatting in release note templates, ensuring instructions and labels appear in proper context.
Enhancements • Enhanced 13-inch responsive layouts for the App Integration and Profile pages, providing better user experience on smaller screens. • Updated sharing features for Meetings and Collections with improved Teams sharing and better Slack citation navigation. • Improved latency on Production for Collections by prefetching APIs and reducing unnecessary overhead, resulting in faster load times. • Streamlined load times for Recent Mode search and assist responses by introducing parallel prompt processing and optimizing queries. +2 other improvements
Bug Fixes • Fixed critical issue where Slack automation no longer displayed email IDs, restoring proper functionality. • Resolved Gmail.com-based signup/login failures that were preventing users from accessing the platform. • Addressed persistent "[sign-up] Invite needed to access Ayraa" error that appeared as users switched pages. • Eliminated duplicate search results for collections links, providing cleaner search results. • Fixed broken clickable cards and missing attachment previews in Collections. +9 other miscellaneous fixes
• Implemented 'Recent Mode' for search and assist with turbocharged results from the last 90 days • Fixed signup process for deleted tenant/account scenarios
+13 other minor enhancements
Bug Fixes
• Fixed issue where meeting transcripts were not appearing due to null pointer exception • Resolved problem where certain queries were hanging in Search via web app • Corrected scoring issues in Recent Mode that were affecting search result relevance • Fixed issue where JIRA-related At-Ayraa queries were not providing expected responses
New Features • Integration of Microsoft Teams with native API and Recent Mode support: Now you can index and search MS Teams channels, chats, and messages directly within Ayraa. • 30-day indexing and periodic crawling for Gmail: We've implemented smart filtering that automatically removes promotional and social emails to reduce noise in your search results.
Enhancements • Sleeker onboarding experience: We've redesigned the onboarding flow to avoid double-integration and double-popup issues, with an improved "You are all set" page. • Recent Mode implementation: Access your most recent workplace knowledge faster with our new Recent Mode for search and assist interfaces, complete with time filter updates and an intuitive speed icon. • Performance optimization for search: We've significantly reduced the delay in Google Drive PDF search and optimized indexing for recently accessed files. • Promotional email filtering: Gmail results now filter out promotional and social emails to reduce noise in search results and improve the quality of responses. • Improved UI consistency: We've standardized fonts and improved global-level day 0 screen consistency across different screen sizes. • Enhanced calendar experience: Added a clear "No scheduled meetings found" message when your calendar is empty.
Bug Fixes • We've fixed an issue where meeting bots were attending meetings even when all users had selected the "I'll invite myself" option. • We've also resolved several Recent Mode issues, including incorrect scoring of search results and improved filtering accuracy when selecting specific sources. +53 other improvements and fixes across the platform
We are excited to introduce our first agentic knowledge assistant - your 24/7 Salesforce analyst!
0:00
/1:07
In this write-up, we share details of how we built this agent, along with our framework for extending it to other connectors such as JIRA, Zendesk, and Gmail. We are also introducing "Text to SQL" structured query language capabilities to our Salesforce connector.
Problem Statement
When we first built our Salesforce connector, we implemented advanced fuzzy keyword and semantic search capabilities. While powerful, we quickly discovered our customers needed more. They asked complex, parametric questions that our search couldn't handle effectively.
Our existing system handled basic Salesforce searches effectively, particularly those focused on finding specific opportunities using keywords within our indexed data (title, content, metadata). However, it struggled with more complex user queries that:
Involved relationships between Salesforce objects (Accounts, Opportunities, Contacts, and Tasks)
Examples:
"Show me all opportunities for the United Oil account."
"What tasks are associated with the United Oil account and are due this week?"
Required filtering on specific field values
Examples:
"Show me all opportunities closed won in the last week."
"What are my overdue tasks?"
Needed information not in our search index
Examples:
"What are the details on John Doe?" (Contact details)
"Who is the account owner for Acme Corp?"
Needed a broad search across multiple objects/fields, then filtering
Examples:
"Find all opportunities related to 'renewable energy' that are in 'Negotiation' stage."
"Show me opportunities related to companies or people matching 'cloud solutions'"
These limitations prevented users from asking natural, intuitive questions about their Salesforce data. Additionally, in some cases, as there was no keyword or semantic meaning to search for in the first place, the search would lead to irrelevant retrieval noise in our RAG pipeline.
Enter Text-to-SQL: Precision meets natural language
To solve this, we're introducing Text-to-SQL capabilities in our Salesforce connector. This feature allows our agent to translate natural language questions into precise SQL queries, giving you more control than ever on your search scope.
Instead of using our own SQL database, we picked Salesforce's powerful SOSL (Salesforce Object Search Language) and SOQL (Salesforce Object Query Language) APIs. This allows us to avoid crawling years' worth of Salesforce records & still provide powerful queries over all of the data in seconds.
We will now get into the details of these APIs and how to use them.
Why SOSL and SOQL?
Here's a short crash course on these powerful APIs from Salesforce.
SOSL (Salesforce Object Search Language)
Best for: Broad text searches. Imagine a Google search within Salesforce.
Use it when: You don't know exactly where the information is (which object or field) or you need to search across many objects at once.
Example:
FIND {renewable energy} IN ALL FIELDS (finds "renewable energy" anywhere).
SOQL (Salesforce Object Query Language)
Best for: Structured queries with precise filtering and relationships. Like a particular database query.
Use it when: You know exactly which object and fields you need, and you need to filter based on specific values or relationships.
Example:
SELECT Id, Name FROM Opportunity WHERE StageName = 'Closed Won' AND CloseDate = THIS_YEAR (finds opportunities that are closed won this year).
Sometimes you need both!
Broad Search + Narrowing Down: Use SOSL to find a set of possible records, then use SOQL to filter those records further based on criteria that SOSL can't handle.
Example:
Query: "Find all opportunities related to renewable energy in the negotiation stage."
SOSL: FIND {renewable energy} IN ALL FIELDS RETURNING Opportunity(Id, Name, StageName) (broad search).
SOQL: SELECT Id, Name, StageName FROM Opportunity WHERE StageName = 'Negotiation' AND Id IN (<ids from SOSL>) (filter by stage).
Thankfully, modern LLMs are trained on the language powering these APIs, allowing us to generatively create them on the fly based on user queries.
Enter 24/7 Agent for your Salesforce data
We now describe our agentic search framework, which allows users to speak to their business data. By implementing an AI agent as an intermediary between user queries and various search tools, we've created a system that intelligently determines the most efficient strategy to retrieve relevant information. The agent can also easily be extended to other CRMs and apps like Zendesk, Gmail, Calendar, etc.
Agent Architecture
The agent sits at the forefront of the architecture, acting as an intelligent & flexible middleware between user queries and the backend infrastructure. This design allows for efficient query routing and seamlessly extending the agent's capabilities in the future.
Core Tools
The framework consists of four primary tools available to the agent:
Salesforce Object Search Language (SOSL) API from Salesforce
Salesforce Object Query Language (SOQL) API from Salesforce
Proprietary indexed and embedded vector data for workspace content
Native text-search API for historical text search from Salesforce
We document these tools in detail so that the agent can understand when to use them and, if needed, how to use them effectively.
Agentic Workflow
We then give the Agent a high-level workflow to follow for every user query, but with flexibility driven by its reasoning on how to navigate the workflow. The sequence is:
1. User Intent Analysis
The agent initiates by decoding the user's intent through multi-layered analysis:
Domain Classification: Determines whether the query relates to Salesforce objects (e.g., accounts, opportunities) or requires external tools.
Tool Requirement Assessment: Evaluates which Salesforce search tool or mix of tools are necessary based on the user's intent and search scope.
This phase resolves ambiguities early, ensuring downstream processes align with the user's goals.
2. Tool Selection
Leveraging insights from intent analysis, the framework selects optimal tools and strategies:
Requirement-Tool Matching: Maps query parameters (e.g., date filters, relationship mapping) to available tools.
Strategy Optimization: Prioritizes execution order—for example, running parametric date filters before semantic keyword searches to narrow the dataset.
Finalization: Confirms tool sequence and prepares for execution.
This phase eliminates redundant tool usage and ensures resource efficiency.
3. Tool Execution
Once the intent is translated into exact tool or mixture of tools to use, the agent moves towards orchestrating the execution of these tools. The agent has to command our backend code to do this via precise instructions. The agent does this by executing the following steps:
Entity & relationship extraction: Based on the user query, the agent selects Salesforce objects (accounts, opportunities, tasks & contacts) & the relationships between them ("opportunities linked to Account X")
Parametric search scope extraction: Likewise, the agent constructs the exact set of criteria the user has specified in terms of date-ranges, personnel, keywords, amounts or other fields.
Query Construction & Submission: The agent then speaks the language of each of the tools it needs to use. It translates the above entities, relationships, and search filters into the exact parameters, configurations, and/or queries to send to the backend to execute these tools (Salesforce SOSL and SOQL API, Elasticsearch). We use a simple JSON-formatted payload to explain this to the backend.
Response Handling: The agent then waits for the backend to respond. Once the responses start coming from the various tools, the agent collates responses, validates results, resolves errors & prepares an appropriate response for the user based on the combined response from the tools used.
The key innovation in any agentic framework is that the backend tools are decoupled from the workflow, and the agent is given some level of flexibility to use its reasoning in orchestrating the worfklow and responding to the user accordingly.
Key Benefits
There are several benefits from this release of our Salesforce connector.
Text to SQL querying via SoSL and SoQL APIs enable extremely powerful analysis, but now possible conversationally
Analytics become accessible - 24/7 across the sales and management organizations
Unlimited access to historical data going back years
Low cost and operational footprint
Foundation for Agentic search, actions & workflows for various workplace apps
Conclusion
We are excited to introduce this next-generation Salesforce connector & hope this write-up helps explain the technical details behind how to create it in-house.
For more information about implementing this connector or to schedule a demonstration, please get in touch with our sales team via our website - www.ayraa.io