Published August 4, 2026

July 2026 product update

August 4, 2026

In July, Unified.to won the 2026 API Award for Best Business Software API, added 108 integrations, launched a Customer Data Platform API, added SAML SSO login, shipped Core Unified API mode for managing a Unified workspace through MCP, and added support for the stateless 2026-07-28 Model Context Protocol revision on the Unified MCP server.

2026 API Award for Best Business Software API

On July 10, 2026, Unified.to won the 2026 API Award for Best Business Software API, presented by DevNetwork and API World. The award recognizes design, engineering, and innovation in API and integration technology.

Breadth and depth in the same month

Unified added 108 integrations in July, after 57 in June, bringing the catalogue to 618 integrations across 32 categories as of July 31, 2026. Depth work grew faster than the catalogue over the same period, and both are visible in the same changelog.

Three notable additions shipped in July:

A Customer Data Platform API category with a six-object model. Profile, Segment, Event, Source, Destination, and Activation, with Profile returning segment membership, identity keys, and consent state on one record.
payment_terms across all nine accounting integrations in a single pass on July 29: QuickBooks, QuickBooks Desktop, Xero, NetSuite, Sage Intacct, Microsoft Dynamics 365 Business Central, Odoo, Zoho Books, and Freshbooks. category_ids, payments, is_elimination, organization_id, and attachments landed across subsets of the nine in the same pass.
A normalized document type across seven HR platforms. hris_document returns type on a fourteen-value enum, so an application filters for signed contracts or expiring visas on one value rather than a different label per platform.

We outlined why breadth and depth stop competing for the same resource in How Unified.to supports 500+ integrations without sacrificing depth.

MCP v2: the Unified MCP server supports the stateless 2026-07-28 revision

The Unified MCP server supports the stateless protocol core of the 2026-07-28 Model Context Protocol revision, in production as of July 28, 2026, the day that revision was finalized.

The revision removes the initialize handshake and the protocol-level session. Every request carries its own protocol version, client identity, and capabilities. Routing information travels in HTTP headers, so a gateway routes an MCP request without reading the JSON-RPC body.

	Before 2026-07-28	2026-07-28
Session	Established by an initialize handshake, torn down at close	None
Protocol version and capabilities	Negotiated once, then remembered	Sent with every request
Server state between calls	Required	None
Load balancing	Session affinity or a shared session store	Round-robin, no affinity
Gateway routing	Request body inspection	HTTP headers
What that changes in deployment: round-robin load balancing, no sticky routing, no shared session store. MCP requests cross corporate gateways and firewalls on the same terms as any other HTTP request.

MCP clients built against earlier revisions continue to work against the Unified MCP server.

Unified implements the stateless protocol core. The revision's extensions framework, MCP Apps, the Tasks extension, and the OAuth and OpenID Connect authorization changes are not implemented, because none of them are required for authorized agent access to an end-customer's integration data.

Endpoints: https://mcp-api.unified.to/mcp for the US, https://mcp-api-eu.unified.to/mcp for the EU.

The Unified MCP server carried 43,456 tools as of July 22, 2026, including 8,242 unified tools, up from 27,825 and 7,609 on June 30, 2026. The 2026-07-28 revision is what makes that catalogue reachable from where agents run. Inside a company network, reachability was the binding constraint rather than integration coverage.

Reachability and selection are separate problems. Current LLM APIs accept a small fraction of a catalogue this size in a single call, so permissions, tools, and defer_tools are what make it usable. The 2026-07-28 revision solves reachability.

→ MCP v2 is here: Unified.to now supports the stateless 2026-07-28 revision in production

Core Unified API: managing a Unified workspace through MCP

The Unified MCP server now runs in two modes. Connection mode points at a single end-customer connection and provides that connection's integration tools: CRM contacts, accounting invoices, HRIS employees. Core Unified API mode authenticates with a workspace API key and no connection ID, and provides management tools for your own workspace configuration.

Mode	Credential	Scope
Connection	Connection ID, or a connection-scoped signed token	One end-customer connection's integration objects
Core Unified API	Workspace API key	Workspace configuration: connections, webhooks, integrations, API-call logs, issues, environments
Both modes run on the same endpoints: `https://mcp-api.unified.to/mcp` for the US and `https://mcp-api-eu.unified.to/mcp` for the EU. The credential selects the mode.

Core Unified API mode operates on workspace configuration only. It does not read or write end-customer data inside a connection.

Tools available in Core Unified API mode:

Group	Tools
Documentation	`search_docs`, which answers a question against docs.unified.to
Connections	`list`, `get`, `create`, `update`, `remove`, plus `get_unified_connection_mcp_url`
Webhooks	`list`, `get`, `create`, `update`, `remove`, `update_unified_webhook_trigger`
Integrations	`list_unified_integrations`, returning available integrations and their supported capabilities
Issues	`list`, `get`
API calls	`list`, `get`, against recent API-call logs
Environments	`list`, `create`, `remove`

Generating a connection-scoped URL

get_unified_connection_mcp_url is the bridge between the two modes. An agent manages connections with the workspace API key, then generates a ready-to-use MCP URL for any individual connection.

The mode parameter decides what the generated URL carries. api, the default, embeds the workspace API key, which grants full account access and should not be shared. user returns a connection-scoped signed token that grants access to that one connection and never carries the workspace key, so it is safe to hand out.

The generated URL can be narrowed at the point of creation: permissions limits it to specific scopes such as crm_contact_read,crm_company_read, tools limits it to named tool IDs or wildcard patterns, hide_sensitive strips PII from responses before they reach the model, defer_tools reduces context usage, dc selects the data region, and include_external_tools adds the integration's raw pass-through endpoints as callable tools.

The result: an agent can create a connection, generate a read-only, PII-stripped URL scoped to a single connection and a single category, and hand that URL to whoever needs it. The workspace API key never leaves the workspace.

A workspace API key grants access to every connection and the entire Unified.to account. Core Unified API mode is for your own infrastructure, not for distribution to end-customers.

→ Core Unified API reference

Customer Data Platform API

We launched a new Customer Data Platform API category for reading customer profiles, segments, events, sources, destinations, and activations across CDP integrations through one Customer Data Platform API.

Supported integrations: Adobe Experience Platform, Amperity, Bloomreach Engagement (Exponea), BlueConic, Lytics, mParticle, Salesforce Data Cloud, Tealium, Treasure Data, and Twilio Segment.

The API provides consistent objects across these integrations:

Object	Purpose
Profile	A customer record, identified or anonymous, carrying identity resolution keys, consent state, and the segments the profile belongs to
Segment	An audience definition, with its computation mode, membership size, and active state
Event	A behavioural event attached to a profile, with occurrence time and identity keys
Source	An inbound data connection feeding the platform
Destination	An outbound connection the platform sends audiences to
Activation	The link between a segment and a destination, with schedule, status, and last run time
Profile embeds segment membership. One call to `GET /cdp/{connection_id}/profile` returns the customer record with its segments, identity keys, and consent state on the same object, rather than a profile read followed by a membership read per segment. Every object also returns `raw`, so integration-specific fields stay reachable without a second call.

Activation is the object that describes what a CDP does with an audience: which segment goes to which destination, on what schedule, with what status, and when it last ran. Reading activation status is how an application confirms an audience reached its destination rather than assuming it did.

Each request is executed directly against the source API: no caching, no sync jobs, no stored records. Traffic is regionalized across US, EU, and AU for data-residency requirements. The connection_id changes. The API calls don't.

Coverage as shipped, by readable field count:

Integration	Profile	Segment	Event	Source	Destination	Activation
Twilio Segment	6/18	11/11		6/8	5/8	5/11
BlueConic	13/18	6/11	7/7	7/8	7/8
Lytics	10/18	8/11		6/8	6/8	7/11
Adobe Experience Platform		7/11		7/8	7/8	10/11
Salesforce Data Cloud	11/18			5/8		3/11
mParticle	16/18			5/8
Treasure Data		6/11		4/8	4/8
Bloomreach Engagement (Exponea)	9/18		7/7
Amperity		2/11
Tealium		2/11
Twilio Segment returns Segment complete at 11 of 11. mParticle is deepest on Profile at 16 of 18. BlueConic and Bloomreach Engagement return Event complete at 7 of 7.

Event delivery:

Event	Integrations
`segment_created`	Adobe Experience Platform, BlueConic, Lytics, Twilio Segment, Tealium, Treasure Data
`segment_updated`	Adobe Experience Platform, BlueConic, Lytics, Twilio Segment, Treasure Data
`segment_deleted`	Tealium
`profile_created`	BlueConic, Lytics, mParticle, Salesforce Data Cloud, Twilio Segment
`profile_updated`	BlueConic, Lytics, mParticle, Salesforce Data Cloud
`profile_deleted`	Twilio Segment
`event_created`	BlueConic
What this enables:

Read customer profiles across CDP integrations through one Customer Data Platform API
Read segment definitions, membership size, and computation mode without per-vendor query logic
Read activation status to confirm an audience reached its destination
Receive created and updated events on profiles and segments
Feed profile and segment data to AI features with consent state on the same record

Field support, webhooks, and list parameters per integration are on the Supported Integrations page.

→ Unified Customer Data Platform API

You can now offer SAML SSO to your enterprise customers, so their employees sign in to your product through their own identity provider.

Supported at launch: Google Workspace Directory, JumpCloud, Microsoft Entra ID, and Okta.

Access is managed in your customer's directory. Their IT team controls who can sign in, and revokes access without touching your product. When someone leaves, the offboarding that already happens in the directory removes their access to your application as well, with no request to your support team and no manual step on your side.

This is the enterprise-managed alternative to OAuth2-based Business Login, which covers 166 identity providers. Business Login suits self-serve sign-up, where each user authorizes individually. SAML SSO suits the enterprise buyer whose security review requires that account provisioning and deprovisioning stay inside their directory.

New integrations

108 integrations added in July.

Enhancements

Improved connection health reporting

Connections are now marked unhealthy only when an integration no longer accepts their credentials or access. Unhealthy connection notifications include the integration's HTTP status information where it is available. Validation errors, missing records, rate limits, and temporary integration outages no longer affect connection health.

Accounting field normalization

On July 29 we added the same field names against the same objects across nine accounting integrations in one pass: QuickBooks, QuickBooks Desktop, Xero, NetSuite, Sage Intacct, Microsoft Dynamics 365 Business Central, Odoo, Zoho Books, and Freshbooks.

A normalized field only removes per-vendor logic when it lands everywhere at once. A field present on three of nine still forces a conditional.

Readable fields added, by object:

Field	Integrations and objects
`payment_terms`	Bill, Contact, and Invoice on QuickBooks, QuickBooks Desktop, Xero, and Sage Intacct; Bill and Invoice on Odoo, Zoho Books, and Freshbooks; Contact on NetSuite and Microsoft Dynamics 365 Business Central
`category_ids`	Bill, Expense, Invoice, Journal, and Purchaseorder on QuickBooks and Xero; Expense, Journal, Purchaseorder, Salesorder, and Transaction on NetSuite; Bill, Invoice, Purchaseorder, Salesorder, and Transaction on Microsoft Dynamics 365 Business Central; Invoice, Purchaseorder, and Salesorder on QuickBooks Desktop
`payments`	Bill and Invoice on QuickBooks, QuickBooks Desktop, Xero, and NetSuite
`type` and `is_elimination`	Organization on QuickBooks, QuickBooks Desktop, Xero, NetSuite, and Microsoft Dynamics 365 Business Central
`organization_id`	Category, Expense, and Order on Sage Intacct; Bill, Contact, Creditmemo, and Order on NetSuite; Order and Salesorder on Microsoft Dynamics 365 Business Central
`attachments`	Bill and Journal on Microsoft Dynamics 365 Business Central; Journal on Xero and Sage Intacct
`applications`	Creditmemo on QuickBooks and QuickBooks Desktop; Vendorcredit on QuickBooks
`lineitems`	Bill and Expense on QuickBooks Desktop
Writable fields added:

Field	Integrations and objects
`category_ids`	Bill, Invoice, Purchaseorder, and Salesorder on Microsoft Dynamics 365 Business Central; Invoice and Purchaseorder on QuickBooks Desktop
`organization_id`	Contact on NetSuite; Order on Microsoft Dynamics 365 Business Central
Accounting field work landed earlier in the month too. `term` became readable on Bill and Invoice across QuickBooks, Xero, Sage Intacct, Odoo, Zoho Books, and Freshbooks on July 6, and writable on QuickBooks. `paid_at` was added on July 23 to QuickBooks Bill, NetSuite Invoice, and Freshbooks Bill. `attachments` was added to Xero Expense, NetSuite Expense, and Freshbooks Invoice and Expense on July 23, and to QuickBooks Expense on July 13.

New `hris_document` object

We added an hris_document object for employee and company documents, following the same pattern as ats_document. List documents, optionally filtered by employee ID, and download an individual document.

Documents come back with a normalized type across fourteen values: CONTRACT, OFFER_LETTER, POLICY, TAX, ID, VISA, PAYSLIP, BENEFITS, CERTIFICATION, PERFORMANCE_REVIEW, ONBOARDING, TERMINATION, MEDICAL, and OTHER when a source value can't be mapped. An application asking for every employee's signed contract, or every visa document approaching expiry, filters on one value rather than on a different label per HR platform.

document_url returns a download link that expires after one hour. user_id points to the Employee object that owns the document, and is readable on all seven integrations. type is readable on six of the seven; Paychex does not return it.

Integration	Methods
BambooHR	list, get
FactorialHR	list, get
HiBob	list
Personio	list, get, remove
Paychex	list, get, remove
Workday (Legacy)	list, get
UKG HR Service Delivery	list, get
UKG HR Service Delivery also delivers created and updated virtual webhooks for Document.

New `hris_taxonomy` object

We added an hris_taxonomy object for standard skills and role reference data. A taxonomy is the canonical set an organization or a country treats as authoritative, as opposed to the free-text values that differ record to record.

type identifies the kind of item and takes six values: SKILL, KNOWLEDGE, COMPETENCE, ABILITY, CERTIFICATION, and ROLE.

Taxonomy objects shipped on three new integrations:

Field	ESCO	Lightcast	O*NET
`id`	yes	yes	yes
`type`	yes	yes	yes
`name`	yes	yes	yes
`url`	yes	yes	yes
`description`	yes	no	yes
`parent_id`	yes	no	yes
`alternative_names`	yes	no	no
`role_ids`	yes	no	no
ESCO returns the full model. `role_ids` holds the IDs of the ROLE items a skill relates to, so one call returns a skill with the roles it maps to rather than a skill read followed by a relationship read. `alternative_names` carries synonyms and alternative labels, which is what makes the same skill resolvable when it arrives written five different ways across five records. `parent_id` handles arbitrary-depth trees on ESCO and O*NET.

Event delivery expansion

Native and virtual webhook coverage expanded across accounting, ATS, HR, call center, and analytics integrations.

Accounting and payments

Exact Online: created and updated virtual webhooks for Invoice
Maxio (Chargify): created and updated virtual webhooks for Contact and Invoice
NetSuite: created and updated virtual webhooks for Transaction and Category
Sage Intacct: created and updated virtual webhooks for Creditmemo, Vendorcredit, and Category
QuickBooks: created and updated virtual webhooks for Transaction
Freshbooks: created and updated virtual webhooks for Payment, Taxrate, and Collection
Xero: created and updated virtual webhooks for Timeoff and Employee

ATS, HR, and Learning

Atlas: created and updated virtual webhooks for Application, Candidate, Company (ATS), Company (CRM), Contact, and Lead
Cornerstone: created and updated virtual webhooks for Location, Application, Candidate, and Student
Oracle HCM: created and updated virtual webhooks for Candidate and Job
Stardex: created and updated virtual webhooks for Scorecard, Company (ATS), Company (CRM), and Deal
Bullhorn: created and updated virtual webhooks for Group, Event, and Interview
Pinpoint: created and updated virtual webhooks for Application, Group, and Activity
AFAS Profit: created and updated virtual webhooks for Employee, Company, and Group
Paychex: created and updated virtual webhooks for Payslip
UKG HR Service Delivery: created and updated virtual webhooks for Document
O'Reilly: created and updated virtual webhooks for Content
Lever: created native webhooks for Interview, Candidate, and Application; updated and deleted native webhooks for Interview and Candidate

Call center, messaging, and analytics

Calendly: created, updated, and deleted native webhooks for Recording and Contact; created and updated virtual webhooks for Contact
3CX: created and updated virtual webhooks for Call and Recording
8x8 Work & Contact Center: created and updated virtual webhooks for Recording
8x8 Connect: created and updated native webhooks for Call
Dubber: created and updated native webhooks for Recording
Zoom: created and updated virtual webhooks for Message; created native webhooks for Event
Telegram: created and updated virtual webhooks for Message
PostHog: created and updated virtual webhooks for Visitor
Salesforce: created and updated virtual webhooks for Space
Atlassian Jira: created and updated virtual webhooks for Project; created virtual and native webhooks for Change

Task object coverage

type, start_at, end_at, story_points, and time_spent were added as readable fields on Task across Atlassian Jira, ClickUp, Linear, Asana, Trello, Shortcut, Azure DevOps, BitBucket, Microsoft Planner, Microsoft Dynamics 365 Customer Engagement, Motion, NetSuite, Sage Intacct, Streamtime, Cascade, Connecteam, Gainsight, Lemlist, Team Software, TrackerRMS, and YOOBIC.

assigned_user_id pagination was added on Task across Atlassian Jira, Asana, ClickUp, Linear, GitHub, GitHub Enterprise Server, Motion, NetSuite, Team Software, and YOOBIC.

Writable start_at and end_at on Task: Microsoft Dynamics 365 Customer Engagement, Streamtime, TrackerRMS.

Ticket attachments

attachment_ids was added as a readable field on Ticket across Zendesk, ServiceNow, Intercom, HubSpot, freshdesk, Gorgias, Pylon, and Atlassian Service Desk. A File object was added to Zendesk, ServiceNow, Intercom, freshdesk, Gorgias, Pylon, and Atlassian Service Desk, with get or list and get method coverage.

slow_fields pagination was added on Ticket for Zendesk, ServiceNow, Intercom, freshdesk, and Atlassian Service Desk.

New objects and method coverage

Zoom: Channel, Event, Message, and Contact objects; list and get for Channel and Contact, full list, get, create, update, remove for Message
Spotify: Organization, Campaign, Group, Ad, Creative, Target, and Report objects across the Advertising API, with create and update coverage on Campaign, Group, and Ad
Telegram: Channel and Message objects, plus OAuth2 sign-in authentication
Freshbooks: Payment, Taxrate, Creditmemo, Item, Collection, File, and Profitloss objects, with full method coverage on Payment, Taxrate, Creditmemo, and Collection
Metabase: Database, Table, and Query objects through the Datastore API
Stardex: Company (ATS), Company (CRM), Deal, Event, Pipeline, Scorecard, and Employee objects
Carerix: Company, Metadata, Activity, and Employee objects
NetSuite: Transaction, Profitloss, Category, and File objects
Sage Intacct: Creditmemo, Vendorcredit, and Category objects with full method coverage on the first two; create and update on Contact with 17 writable fields
Oracle HCM: Course, Content, and Activity objects
Cornerstone: Course, Activity, Student, and Location objects
Exact Online: Invoice and Organization objects
Maxio (Chargify): Subscription, Contact, and Invoice objects
QuickBooks: Vendorcredit object with full method coverage
Bullhorn: Interview object
Atlas: Activity, Metadata, and Lead objects
Pinpoint: Activity and Group objects; create on Candidate
Google Ads and TikTok Ads: Asset object, with create coverage on both
Calendly: Recording and Contact objects
Concur and Concur (Company): Contact object with full method coverage
Workday (Legacy): Document object; get, create, update on Location, Timeoff, and Group
Paycom: Location object
Pennylane: Bill object
Wave: get method on Contact, Account, Invoice, and Item
Salesforce: create on Metadata
Atlassian Jira: Metadata object with full method coverage
Acuity Scheduling: OAuth2 sign-in authentication

Field-level expansions

ATS candidate detail: skills, experiences, and education added as readable fields on Candidate across Cornerstone, iCIMS, Recruiterflow, Lever, Crelate, Ashby, and JobDiva; writable on Phenom and Workday (Legacy)
ATS ownership: user_ids added as readable and writable on Candidate across TrackerRMS, JobAdder, Crelate, and Stardex
Application status: job_id added as a readable field on Applicationstatus across Greenhouse, Workable, Teamtailor, Trakstar Hire, JobScore, Stardex, onlyfy, Occupop, and FactorialHR
Expense detail: status added as readable on Xero, Expensify, Brex, Concur, and Concur (Company); users, posted_at, payment_method, attachments, and approver_users added on both Concur integrations
Advertising status: effective_status added as readable on Campaign and Group across Google Ads, Meta Ads, and LinkedIn
Repository detail: pullrequest_ids on Commit and title, source_branch_id, target_branch_id, notes on Pullrequest across GitHub, GitHub Enterprise Server, GitLab, GitLab Dedicated, BitBucket, and Azure DevOps; lines_added, lines_changed, lines_deleted on Commit
Learning content: tags, published_at, difficulty, and subjects added as readable on Content across O'Reilly, Skillsoft (Percipio), Degreed, Coursera, D2L Brightspace, LearnUpon, and Oracle HCM
Workday (Legacy) writes: writable fields added across Group, Timeoff, Location, Deduction, Candidate, and Document

Pagination improvements

slow_fields support added across Zendesk, ServiceNow, Intercom, freshdesk, Atlassian Service Desk, Lever, Atlassian Jira, Carerix, Cornerstone, Workday (Legacy), Paychex, Sage Intacct, Xero, Workable, Pinpoint, and the six repository integrations
Atlas: updated_gte, offset, query, and virtual webhook pagination variants across Application, Candidate, Company (ATS), Company (CRM), and Contact
Cornerstone: query, updated_gte, and virtual webhook variants across Employee, Application, and Candidate
3CX: updated_gte and virtual webhook variants on Call and Recording
8x8 Work & Contact Center: updated_gte, start_gte, end_lt, and virtual webhook variants on Recording
Expense date and status filtering: start_gte, end_lt, status, and virtual webhook variants across Xero, QuickBooks, Brex, Concur, Concur (Company), and Expensify
Atlassian Jira: project_id, task_id, updated_gte, and native webhook variants on Change; order and virtual_webhook_limit on Project
Xero: updated_gte, offset, and virtual webhook variants on Employee and Timeoff
PostHog: limit, offset, and virtual webhook variants on Visitor; updated_gte on Session, Visitor, and Report
AFAS Profit, Bullhorn, Oracle HCM, Ashby, Salesforce, Workday (Legacy), and Paychex: updated_gte and virtual webhook pagination across newly covered objects

Final takeaway

July added 108 integrations, a Customer Data Platform API with a six-object model, SAML SSO login, Core Unified API mode, and support for the stateless 2026-07-28 MCP revision, while depth work grew faster than the catalogue did.

The direction is consistent: more integrations behaving predictably, more objects covered completely, and less custom logic required per vendor.

→ Start your 30-day free trial → Book a demo

Author

Written for Unified.to by Mallory Greene

About the author: Mallory Greene is a writer specializing in generative engine optimization (GEO) and content, through her practice Search Everywhere. She covers integration infrastructure and technical content across Unified.to's technical content library. Based in Toronto.

All articles