July 2026 Summaries

The Unified MCP server has adopted the 2026-07-28 stateless revision of the Model Context Protocol (MCP), marking a significant shift in how MCP operates by eliminating protocol-level sessions and enabling it to run on standard HTTP infrastructure. This change allows for more seamless integration with existing internet infrastructure, as it no longer requires special session-aware handling, thus enhancing reachability within corporate networks. The revision introduces a stateless protocol core, routable headers, an extensions framework, and a formal deprecation policy, while maintaining backward compatibility with earlier MCP client versions. Despite not implementing MCP Apps or the Tasks extension, the Unified MCP server focuses on providing authorized agent access to integration data with features like data region selection and PII filtering. As of July 22, 2026, it boasts a large tool catalogue of 43,456 tools, underscoring its extensive integration capability across numerous categories.

On July 28, 2026, the Unified Model Context Protocol (MCP) server became one of the first to support the stateless 2026-07-28 MCP revision, a significant update that removes protocol-level sessions, allowing MCP to operate on standard HTTP infrastructure. This change enhances compatibility with corporate networks and simplifies load balancing by eliminating the need for sticky sessions or shared session stores. The revision introduces self-contained requests, carrying protocol version, client identity, and capabilities, and facilitates routing through HTTP headers, making it more efficient and scalable. Although the Unified MCP server does not implement features like MCP Apps or the Tasks extension, it supports data security enhancements such as data region selection and PII filtering. This update has expanded the server's tool catalogue to 43,456, improving its reachability within corporate environments and maintaining backward compatibility with clients built on earlier protocol versions.

To integrate BambooHR with Unified using OAuth 2, you must first register an application in the BambooHR Developer Portal to obtain OAuth credentials such as the Client ID and Client Secret. You then configure these credentials within Unified, set up the necessary redirect URI, and choose the required scopes for accessing different BambooHR functionalities. Once the setup is complete, end users can authorize their BambooHR accounts through an embedded Unified connection flow. This process involves users entering their BambooHR domain, approving access permissions, and confirming the connection in Unified, which should display a "Healthy" status once the authorization is successful.

Despite the low initial cost of building AI-driven SaaS integrations, the ongoing maintenance and unforeseen expenses post-launch remain significant challenges. These hidden costs include dealing with third-party API changes, stale data issues, compliance complexities due to copied records, and increased support time due to weak observability. Additionally, integration sprawl across teams and pricing models that scale unfavorably with usage further complicate the situation. AI features also inherit underlying weaknesses, impacting their reliability and accuracy. The opportunity cost of maintaining integrations detracts from core product development, highlighting the importance of evaluating integration architecture over integration count. Unified.to offers solutions to mitigate these costs by managing authentication, rate limits, schema changes, and providing strong observability, although it introduces a dependency on their architecture and service.

Connecting Sage 100 to Unified involves setting up Sage 100's eBusiness Web Services module on the Sage 100 server and ensuring server access through Internet Information Services (IIS). The process requires a Sage 100 system administrator to install and configure eBusiness Web Services, confirm the endpoint URL's accessibility, enable web services for the company in Sage 100, and create a dedicated web-services user with appropriate permissions. This user and endpoint information are then entered into Unified to establish the connection, allowing Unified to perform operations like managing customers and sales orders but excluding functionalities such as invoices and inventory items. Troubleshooting common issues involves ensuring the correct server name matches the TLS certificate, enabling web services for the system or company, and verifying the user's permissions and account status.

AI agents face challenges when using unified APIs due to limitations in handling vendor-specific query languages, permission models, and internal hierarchies, which are better addressed by native APIs. Unified APIs often struggle in scenarios such as needing specific vendor languages like JQL or SOQL, where the unified model's generic structure does not suffice, and in scenarios involving complex permissions or hierarchical data access. The issue of intersection-only schemas, where unified APIs only expose common fields among all integrations, further complicates their utility, potentially reducing the richness of data available to the AI agent. However, solutions like Unified.to offer a Passthrough API that allows agents to access native endpoints without leaving the connection, thus mitigating some of the limitations by maintaining the original object ID and enabling field-level access. The debate on whether to use unified or native APIs depends on the specific task requirements and whether both can operate under the same authorized connection, with considerations for context window efficiency and alignment with an AI model's training data.

Breakthrough Jobs, a candidate analysis platform for hiring teams, switched to a unified API with Unified.to to enhance scalability and efficiency by reading and writing candidate and job data from ATS integrations. Initially considering competitors like Merge, Breakthrough Jobs found Unified.to's usage-based pricing model more suitable than the per-linked-account pricing, which becomes costly as integration numbers grow. Unified.to's extensive ATS coverage, responsive support, and strong security posture were key factors in their decision, enabling Breakthrough Jobs to expedite their market entry and improve their product through faster integration and feedback loops. This strategic shift not only streamlined operations for Breakthrough Jobs but also benefited their clients, such as Candid, a dental technology company, by significantly reducing resume review times. As Breakthrough Jobs expands its services from candidate analysis to sourcing, the unified API supports their evolving business model efficiently.

Greenhouse Recruiting candidate webhooks must be manually configured in the Greenhouse Dev Center to send events through Unified, which normalizes them into the `ats_candidate` format before forwarding them to an application’s public HTTPS endpoint. Setup requires an active Greenhouse connection, its connection ID, a Unified API key, appropriate Greenhouse webhook permissions, and separate native Unified subscriptions for created, updated, and deleted candidate events. Each resulting Unified subscription ID is used to construct a region-specific Unified endpoint that is entered into Greenhouse rather than the application’s own webhook URL. Greenhouse requires individual webhooks for each trigger, with creation triggers such as application or prospect creation mapped to the created subscription, candidate changes including hiring, rejection, stage changes, and updates mapped to the updated subscription, and candidate deletion or merging mapped to the deleted subscription. Greenhouse activates webhooks only after a successful test ping, and testing should verify receipt of normalized payloads containing the expected native webhook type, candidate object type, event, and candidate data, while deleted events provide the deleted candidate ID. Removal requires disabling or deleting both the Greenhouse webhooks and the associated Unified subscriptions, and troubleshooting focuses on endpoint accuracy, regional hostnames, permissions, active webhook status, trigger mapping, and idempotent handling of possible duplicate deliveries.

The Unified API employs custom HTTP headers to provide detailed insights into the time distribution of API request processing, helping identify latency sources either from third-party integrations or Unified's processing layer. The key headers include X-Response-Time, which represents the total time taken for a request from receipt to response; X-EndApi-Response-Time, detailing the time spent waiting on third-party APIs like Salesforce or Slack; and X-Unified-Response-Time, indicating the time consumed within Unified's processes such as authentication and data mapping. While X-Response-Time is included in every response, the other two headers only appear if the request involves a third-party API call. These headers are valuable for debugging, monitoring, and providing support by pinpointing whether latency issues lie with third-party providers or within Unified, thus allowing for more targeted troubleshooting and performance discussions.

The guide explores the landscape of unified APIs for CRM systems as of 2026, emphasizing that key considerations for CRM buyers should focus on the depth of write capabilities and object design rather than integration quantity or caching architecture. Unified.to, a prominent player in this field, offers extensive read, write, and change-event support across 57 CRM integrations, with a notable focus on embedding related records within a single object, thus simplifying data retrieval. In contrast, competitors like Apideck and Merge present different models, with Apideck offering a more granular object model suited for embedded marketplace UIs, and Merge catering to enterprise needs with a sync-and-store approach. The document also highlights the pricing strategies of these platforms, noting that usage-based pricing, such as Unified.to's, becomes more cost-effective as the number of connected customers increases. It further discusses the implications of data synchronization methods, OAuth app ownership, and change detection capabilities, providing a comprehensive overview for potential buyers to make informed decisions based on their specific CRM needs and scale.

A "unified payments API" is a term used to describe three distinct types of products, each serving different roles in the payment processing ecosystem: processors, orchestrators, and data layers. Processors like Stripe handle the direct processing and movement of money, while orchestrators such as Primer and Spreedly route transactions across multiple processors, offering capabilities like retries and card vaulting without handling the funds themselves. Data layers, exemplified by Unified.to, focus on reading and normalizing payment data across existing processors for reconciliation and reporting, without moving the money. The confusion surrounding the term arises because vendors across these categories often market themselves under the same umbrella, despite their fundamentally different functions. Selecting the right type of unified payments API depends on the specific needs of the user—whether it is the direct processing of payments, orchestration across multiple systems, or the aggregation and analysis of payment data across platforms.

Unified.to offers a streamlined solution for promoting job openings from any Applicant Tracking System (ATS) through a unified Advertising API, enabling businesses to sync job listings with advertising campaigns on platforms such as LinkedIn, Meta, Google, and TikTok. By utilizing the Unified.to TypeScript SDK, recruiters can manage job advertisements as projections of the original job postings, ensuring alignment as job details change without storing any candidate or job data at rest. The platform bridges the gap between different systems by supporting a unified model that encompasses both the ATS and advertising categories, simplifying the integration process through a single API connection. Unified.to provides full CRUD support across multiple advertising integrations, allowing for the creation or alignment of campaigns directly from ATS job data, while maintaining compliance with employment ad restrictions that prohibit demographic targeting. Additionally, the system ensures that ad campaigns automatically reflect changes in job status by utilizing ATS job webhooks, thereby reducing the need for manual updates and integration maintenance.

Unified API for call and meeting transcripts in 2026 offers a solution for retrieving existing transcripts from various platforms like Zoom, Teams, and RingCentral without the need for live recording bots or speech-to-text engines. It contrasts with other methods such as bot capture, which involves a participant joining calls to create transcripts, and speech-to-text engines that transcribe raw audio. Unified API retrieves completed meeting and call recordings through an OAuth method, providing normalized transcript formats across video meetings and phone calls, making it ideal for applications that require post-call data processing without storing data. This API also integrates with Generative AI tools, allowing users to summarize, classify, or embed transcripts for further analysis, effectively bridging the gap between retrieval and reasoning without additional transcription or LLM vendors.

This guide provides a comprehensive walkthrough for integrating Zoom and Zoom Phone using an OAuth app, detailing the necessary steps to create and configure the app, set permissions, and connect Zoom accounts. It emphasizes the need for a Zoom Marketplace developer account, admin access, and a Zoom Phone license, and guides users through creating an OAuth app in the Zoom App Marketplace, configuring app credentials, and setting up OAuth redirect URLs based on regional requirements. The document outlines the process of adding specific scopes for various Zoom features like meetings, webinars, and phone services, stressing the importance of matching credentials between Zoom and Unified for successful integration. It also covers testing versus published app modes, addressing potential issues during testing such as adding authorized testers, and transitioning to production credentials after publication. Troubleshooting advice is provided for common errors, and the guide notes the necessity of updating Unified with production credentials post-publication to maintain connectivity, while highlighting the option to integrate other Zoom products like Zoom Calendar separately if needed.

The "Unified API for Advertising" offers a comprehensive solution for managing advertising campaigns across multiple platforms by providing a single API that reads and writes data across 19 integrations, including Google Ads, Meta, TikTok, and others. It simplifies the complex task of integrating with various ad platforms, each with different data models, by normalizing campaign structures into nine objects such as Campaign, Group, and Ad, allowing for consistent management and reporting. This API supports real-time, pass-through data requests without storing ad data, ensuring security and compliance with regulations like GDPR and CCPA. It is particularly useful for businesses that require cross-platform ad management and need the capability to pause campaigns, adjust budgets, and update targeting with explicit, provider-specific write capabilities, distinguishing it from competitors like Truto and social-ads tools that offer limited coverage and functionality.

In 2026, Unified offers a robust unified API for call centers and dialer integrations, addressing the need to connect and read data across multiple existing platforms such as RingCentral, Dialpad, Aircall, and others, without storing sensitive call data at rest. This API aggregates and normalizes call data, including calls, contacts, comments, and recordings, and provides speaker-attributed transcripts and real-time recording events across its 23 integrations. Unlike CPaaS, which builds raw telephony infrastructure, or CCaaS, which encompasses contact center solutions, Unified focuses on reading and normalizing call data for analytics, QA, and AI post-call features. It supports a variety of systems, including UCaaS, contact-center, and sales-engagement platforms, but does not offer live call control, which remains specific to each platform's API. Unified's model ensures data freshness and compliance by routing requests directly to data sources in real-time, making it a suitable choice for companies seeking to integrate diversified call-center functionalities without managing separate integrations for each provider.

In the evolving landscape of calendar and meeting integrations, the choice of a unified API in 2026 hinges on whether the API is designed to route requests in real-time or to sync and cache data, impacting data freshness, customer data management, and AI feature accuracy. Unified API is highlighted as a robust solution for products that integrate calendar and meeting data with CRM, ATS, or AI applications, emphasizing its pass-through architecture that ensures live data retrieval without storing customer data. It stands out with extensive coverage across scheduling, video conferencing, meeting-assistant products, and call-center systems, offering speaker-attributed transcripts and recordings from existing tools. Competitors like Merge, Truto, Nylas, and Cronofy vary in their approaches, with some employing sync-and-cache models that store data and others, like Truto, offering a hybrid option. Unified's approach eliminates data storage complications, provides real-time updates, and supports a wide range of integrations, making it particularly suitable for AI-driven applications and those requiring comprehensive meeting data without the complexities of bot capture or data caching.