The hardest part of enterprise AI isn't the model. It's everything around the model: the intake pipelines, the access controls, the auditing, the apps your teams actually use every day. For the past year, we've been filling in those gaps.
With Vertesia, we've crossed an important threshold.
Let’s walk you through what's new, and what it means for your organization.
The biggest shift in this release is one we've been building toward for a while: you can now build, host, and ship production applications directly on Vertesia.
Until now, teams using Vertesia still had to manage external hosting, write glue code to wire up interfaces, and maintain their own deployment pipelines. That's over. With Vertesia , the full lifecycle (build, configure, deploy, host) happens on the platform.
And for teams that want to move even faster, we've introduced the ability to generate Vertesia applications with our new App Builder (now in preview). Describe what you want in plain language. Vertesia builds it. You review and ship. A compliance team that needs a contract review interface doesn't have to spec it out with developers. They simply describe what they need, and the app exists. An operations team that wants an intake dashboard for vendor invoices can generate one in hours, not sprints.
The combination of hosted apps and AI-generated apps means the time from idea to running product just got a lot shorter.
AI that reads documents is useful. AI that reads documents and tells you exactly where it found the answer is trustworthy enough to use in production.
Vertesia introduces Grounded Extraction for Intelligent Document Processing (IDP). This is our our most recent advance in document intelligence. When the platform extracts structured data from a document, it now returns source citations alongside every result: the exact passage, page, and location it drew from. If confidence is low or a field requires human judgment, the system flags it automatically for review, rather than silently returning a value you can't verify.
This changes the calculus for regulated industries. Financial services teams, legal departments, and insurers have all told us the same thing: they can automate document extraction, but they can't automate it blindly. Grounded Extraction gives them the audit trail they need to act on AI output with confidence.
Alongside this, we've shipped Collaborative Editing, an AI assisted document editing experience where AI proposes changes to a document, shows you a clear before/after diff, and waits for your approval before applying anything. No surprise rewrites. No silent modifications. The human stays in control; the AI does the drafting.
And Intake Policies for Content Types let you define exactly how each type of document should be processed the moment it enters the platform: what workflow it follows, what extraction it triggers, what rules apply. Your NDAs follow one path. Your invoices follow another. Your RFPs trigger a third. Each type has its own unique content processing.
As organizations scale their content operations on Vertesia, two things become critical: finding the right content instantly, and ensuring only the right people can access it.
Content Views solve the first problem. A Content View is a saved, reusable search configuration. It is a combination of filters, sort order, and display settings that always returns the freshest matching results. Think of it as a living report. A procurement team's "Contracts expiring in 90 days" view is always current without anyone running a query. A legal team's "Flagged for review" view surfaces exactly what needs attention, every time.
Attribute-Based Access Control (ABAC) is one of the most significant capabilities shipping in the new release, and one of the most consequential for enterprise deployments at scale. Most platforms stop at role-based access control. Roles are useful, but they're blunt instruments: a "Legal" role can't distinguish a public filing from a sealed settlement. ABAC can. Instead of asking "what role does this user have?", Vertesia evaluates the attributes of the user, the content, and the context (department, sensitivity label, tenant identifier, clearance level) and makes a precise, policy-driven access decision on every request.
ABAC in Vertesia governs your entire content estate. Collections are filtered at the data layer. A user without the right attributes won't see unauthorized collections at all; they're simply not surfaced. No enumeration. No information leakage. Write operations are now governed by the same attribute policies as reads, closing a gap where a caller who knew a document's ID could still write to it directly via API. Every write passes through the same attribute evaluation as every read, before a single byte changes. Reads, writes, collections, and machine callers, all governed from a single, consistent attribute-driven model.
Most enterprise software is built around requests: a user clicks something, the system responds. But modern business processes don't work that way. Documents arrive. Statuses change. Deadlines pass. Decisions get made. Your platform should respond to all of it, automatically.
Vertesia introduces a full Event-Driven Architecture (EDA) at the platform level.
The Event Bus is the core of it: an intelligent switchboard that detects meaningful events: a new document ingested, a field extracted, a review completed, a status changed, and triggers downstream actions in response. No polling. No cron jobs. No custom webhooks written one at a time.
The Event Bus delivers events for four connected capabilities:
The practical result: workflows that used to require scheduled jobs and manual triggers now happen the moment the conditions are right.
In this release, we've consolidated all intake workflows into a single, unified six-stage pipeline that handles every content type in a consistent, observable, and predictable way. We're also retiring the legacy Workflow Rules engine, replaced by the Event Bus: faster, simpler, and more reliable.
Every product release is a set of choices about what matters most. With Vertesia , the choices were clear: lower the barrier to building production AI products, make document intelligence trustworthy enough for regulated industries, give teams control over their content and who can access it, and build the event-driven foundation that makes real automation possible.
We're excited to get this into your hands. Follow us on LinkedIn and subscribe to our blog to stay up to date as we keep building.