The enterprise platform for agentic AI

With Cenna Apex, enterprises can build, deploy, and govern custom agentic apps faster — with full control.

Complete, working applications for Legal, IT, HR, and Support. Your developers customize the source code in the tools they already use. You deploy into your own AWS or Google Cloud account — and see every model call and every dollar from one console.

Your cloud. Your code. Your control.

acme-prod — cenna deployaws
acme-prod
terraform apply · 18 resources · us-east-1
aws_vpc.main+ 3 private subnets2.1s
aws_iam_role.contract-hub+ kms secrets0.9s
aws_ecs_service.contract-hub3 tasks · us-east-13.2s
okta_app.ssoidentity wired0.8s
✔ Apply complete — 18 added · 0 changed · 0 destroyed
live https://contract-hub.internal.acme.com

Example data shown for illustration.

Built on

Available through AWS Marketplace, so Cenna draws down your existing AWS commitment.

The problem

Every team wants an agentic AI application. Legal, IT, HR, Support — all of them, right now.

Today there are two ways to get one. Build it in-house, and each team spends months on the same foundation before the first agent runs: login, integrations, deployment, audit trails. Or buy a vertical AI product, and pay per seat forever while your contracts, your tickets, and your employee records sit in someone else's cloud.

Cenna is the third way.

S1The product

A complete application on day one

Not a blank canvas, and not a chatbot bolted onto your data. Every Cenna application arrives with the interface, agents, workflows, and permissions the work actually requires — then you make it yours.

Cenna App Store catalog of enterprise applications

S2Why Cenna

Speed of buying. Control of building.

Buying a vertical AI product gets you there fast and costs you control. Building in-house keeps control and costs you a year. Cenna Apex is the accelerator that gives you both.

ComparisonVertical AI productBuild it in-houseCenna Apex
Speed to production
Time to a working applicationWeeksQuartersDays
Infrastructure to get thereThe vendor's problemVPC, IAM, secrets, CI/CD, and monitoring, all from scratchOne standard path, already configured
Who has to build itNobody — you rent itA platform team you have to hire: DevOps, SRE, securityThe developers you already have
Who runs it at 3amThe vendorYour on-call rotationCenna operates the platform, you own the application
Keeping it currentWait for the vendor's roadmapYour team owns every upgrade, foreverCenna ships platform updates, you keep the code
Cost control
What you can seeOne line on an invoiceWhatever you built the metering forSpend by app, task, and user on day one
Stopping tokenmaxxingNot your callBudgets, caps, and alerts are yours to buildBudgets, alerts, and hard caps included
The second and third applicationAnother contract, another seat countMost of the build starts overSame platform, new application
Control & security
Who owns the source codeThe vendorYouYou
Where your data livesThe vendor's cloudYour cloudYour cloud
Which models teams can callWhatever the vendor pickedYou build the policy layerOne approved list, enforced everywhere
Connecting your systemsOnly the vendor's connectorsEvery integration hand-wired and scoped by youDeclared with explicit scopes in config
Audit trailThe vendor's logs, on requestYou build itEvery agent run traced and priced

Cenna Apex is an accelerator, not another dependency. The foundation every agentic application needs — identity, integrations, deployment, spend controls, audit — is already built and already running, so reaching production does not start with staffing a platform team. You get there in days instead of quarters, and the application itself stays entirely yours.

S3How it works

From catalog to production in four steps

The same path for every application, whether you start from one of ours or bring code you already have.

  1. 01

    Choose

    Pick an application from the Cenna store, or point Cenna at a repository your team already owns.

  2. 02

    Customize

    Your developers change the agents, workflows, and interface in your own repo. Integrations are declared with explicit scopes, not hand-wired.

  3. 03

    Deploy

    Cenna provisions networking, identity, secrets, and monitoring inside your AWS or Google Cloud account.

  4. 04

    Operate

    Watch spend, model calls, agent traces, and application health for the whole fleet from one dashboard.

S4Applications

Start with the application your business needs

Each application ships with the interface, the agents, the workflows, and the role-based permissions the job actually needs. Launch the foundation in days, and spend your engineering time only on what makes your business different.

Explore Applications →

S5Developers

Build in Cursor or Claude. Ship on Cenna.

The Cenna App SDK carries the parts every enterprise application needs — authentication, interface, agent orchestration, integrations, workflows, deployment. Your team writes only the logic that is specific to your business.

Customize the business logic. Reuse everything else.

Talk to us about the SDK →
contract-hub / app.config.tsCenna App SDK
// Your team writes only this.
import { defineApp, sso } from '@cenna/app-sdk';
import { intake, review, obligations } from './agents';

export default defineApp({
  auth: sso('okta'),
  agents: [intake, review, obligations],
  integrations: ['salesforce', 'docusign', 'slack'],
  deploy: { cloud: 'aws', region: 'us-east-1' },
});

S6Deployment

Your cloud, without the three-month infrastructure project

One standard path to production on AWS or Google Cloud. Networking, identity, secrets, CI/CD, and monitoring come configured. You approve the account — Cenna handles the rest.

AWS · Google Cloud · Azure (coming soon)

Cenna deployment progress into the customer cloud account

S7Model spend

See every AI dollar before the invoice arrives

Tokenmaxxing turned AI spend into a number nobody could predict. Cenna makes it a line item: token spend by application, task, user, and model, budgets and alerts on every app, and an approved-model list enforced everywhere.

Token spend by app, task, user, and model · Budgets and alerts · Approved model list

app.cenna.io/platform/costslive · 4 apps
$0+12% vs JuneAI spend · MTD — cloud $992, in your account
Fleet budget0%
$3,400 of $5,000 cap
Token spend by
Contract Hub244.6M tok$1,842
Vendor RiskOver pace88.9M tok$964
HR Onboarding104.7M tok$486
IT Helpdesk36.1M tok$118
Approved models
  • Claude Sonnet 5Allowed
  • Claude Haiku 4.5Allowed
  • GPT-5.2Allowed
  • GPT-5.2 miniAllowed
  • gpt-4oBlocked
AlertBlocked: Vendor Risk attempted gpt-4o — not on the approved model list.

Example data shown for illustration.

Every agent run is traced and priced on its own — a clause analyzer at 63 spans and $0.92, a redline drafter at 38 spans and $0.64. When spend moves, you can name the task that moved it.

S8Cloud spend

The other half of the bill, in the same console

Model tokens are only part of what an agentic application costs. Cenna reports cloud spend by application, target, and service — Fargate, Postgres, load balancers, cache, storage — right next to the AI spend it belongs with. The true cost of an application is one number, not two dashboards and a spreadsheet.

Cloud spend by app, target, and service · Budget pacing per application · AI and cloud cost side by side

app.cenna.io/platform/costs/cloudin your account
$0Within budgetCloud spend · MTD — billed to your own cloud account
Cloud spend by
Contract Hubaws · us-east-1$474
HR Onboardingaws · us-east-1$198
Vendor Riskaws · eu-west-1$186
IT Helpdeskgcp · us-central1$134
True cost per application
  • Contract Hub$2,316$1,842 AI · $474 cloud
  • Vendor Risk$1,150$964 AI · $186 cloud
  • HR Onboarding$684$486 AI · $198 cloud
  • IT Helpdesk$252$118 AI · $134 cloud
AI Cloud$4,402 total · MTD

Example data shown for illustration.

S9Data, security & governance

Controls your reviewers can verify

Cenna never becomes a new place your data lives. Access follows the identity provider you already run, every integration declares what it is allowed to touch, and every agent run leaves a trace you own.

Cenna security and access view showing allowed models, integration scopes, findings, and recent agent traces with per-trace cost
  • Data residency

    Applications and data stay in your cloud account and region. Nothing is copied to Cenna.

  • Enterprise identity

    Access follows your existing SSO and IAM, with role-based permissions per application.

  • Model governance

    Each application carries its own allowed-model list, and a call to anything off it is refused.

  • Integration scopes

    Every connection declares exactly what it may do — SharePoint read, Slack write, DocuSign signature — and nothing runs outside that grant.

  • Agent traces

    Every agent run is recorded as a trace with its spans and its cost, so any action can be replayed and any dollar explained.

  • Emergency controls

    Pause an agent, revoke an integration, or stop an application without waiting on a support ticket.

  • SOC 2 Type II
  • Google Cloud for Startups

SOC 2 Type II attested. Reports and policies live in our Trust Center. Because applications run in your account, Cenna stays inside your existing compliance boundary.

View Trust Center

S10FAQ

Frequently asked questions

What exactly do we have at the end of the first week?

A working application running in your own cloud account, its source code in a repository you own, and the console tracking its health, spend, and agent activity. Not a sandbox or a pilot environment — the same deployment you keep.

Who has to be in the room to deploy it?

Someone who can approve a cloud account or project, someone who administers your identity provider, and a developer. Cenna provisions the networking, IAM, secrets, CI/CD, and monitoring on a standard path, so there is no platform team to staff before you start.

What can the agents actually reach in our systems?

Only what you grant. Every integration declares an explicit scope — SharePoint Sites.Read.All, Slack chat:write, DocuSign signature — and those scopes are listed per application in the console. An agent cannot act outside its grant, and revoking one takes effect immediately.

What happens when an agent gets something wrong?

Every run is recorded as a trace you can open: the spans it executed, the calls it made, and what it cost. You can replay the run to see what the agent decided, pause the agent, or revoke its integrations yourself — none of it requires a support ticket.

How do we stop an agent from running up an enormous bill?

Budgets and hard caps per application, plus a fleet cap. Spend is metered by application, task, user, and model as it happens, and pace alerts fire before month end. Tokenmaxxing is a governance failure, and the controls for it ship with the platform.

Which models can our teams use?

The ones you approve, chosen per application across providers. The approved list is enforced centrally, so a team cannot quietly reach for something off it — a blocked call surfaces as an alert instead of a line on next month's invoice.

Does this replace ServiceNow or Salesforce?

No. Cenna applications work alongside your systems of record, reading and writing through scoped integrations. Your data stays where the business already keeps it.

What does customization actually take?

The application arrives complete, so your developers change business logic and configuration instead of building a foundation. The work is TypeScript in your own repository, in Cursor, Claude, or any IDE — no proprietary editor and no new skill set to hire for.

You are an early-stage company. What happens to us if Cenna does not make it?

Your applications keep running. The source code sits in your repository and the deployment sits in your cloud account, so neither depends on Cenna being reachable. You are buying an accelerator, not a runtime you can be evicted from.

And if we simply want an application gone?

Uninstalling it tears down the cloud resources it created and snapshots the data for thirty days before deletion. Removing an application is a normal operation in the console, not a support escalation.

Start with a working application. Not a blank page.

See a complete Cenna application customized and deployed into a cloud account, end to end, in a thirty-minute walkthrough.

Get a Demo

Built for speed. Engineered for control.

Cenna control plane showing AI spend by app, provider, and model with token counts