Celigo Ora is generally available: an AI assistant that asks before it acts
Celigo made its AI assistant Ora generally available in September 2026. The interesting part is not the chat, but that every change is staged for approval.
On 2 September 2026, Celigo made Ora generally available to all customers. Ora is an AI assistant that works across the whole Celigo platform: it can build integrations, write field mappings and JavaScript, diagnose failed flows, manage users and environments, and answer questions about the account. Celigo’s CTO Scott Henderson puts it simply: “Anything you can do in the Celigo UI, you can hand to Ora.”
Approval before action
Many vendors now offer a chat assistant for integration work. What makes Ora worth a closer look is how it handles changes. Ora does not change a live integration directly. Every change is staged for the user to approve first, which Celigo describes as a pull request review for integration work. Dry runs, evaluation of output and a full audit trail are part of the same model.
That is a sensible answer to the biggest objection to AI in integration: nobody wants an assistant that quietly changes an order flow to the ERP system on a Friday afternoon. With staged changes, AI becomes a way to produce proposals faster, while responsibility stays with a person.
Built by agents, tested in beta
Celigo says Ora was built almost entirely by AI agents over 14 months, with people reviewing and approving the output, and that the engineering team reached nearly 700% of its starting pace by spring 2026. Before general availability, Ora ran in a six-month beta with more than 16,800 real conversations.
Vendor figures on productivity should be read with care, but the point is still relevant: Celigo uses the same governance model internally that it now offers customers.
Who it is for
Celigo’s customers are often mid-sized companies and business teams in finance, e-commerce and supply chain, typically around NetSuite, Shopify and similar systems. In these teams there is rarely a dedicated integration developer. An assistant that can explain why a flow failed and propose a fix may be worth more to them than to a large enterprise with a full integration competence centre.
Celigo also launched Agent Builder earlier this year, which lets business users build AI-driven workflows with governance built in.
How it compares
Most major iPaaS vendors now have an AI assistant: Boomi has AI agents in its platform and Companion skills, Workato has AI copilots, MuleSoft has AI-assisted development in Anypoint, and SnapLogic has SnapGPT. The differences lie in how much the assistant may do on its own and how changes are controlled. Celigo has chosen a clear line: the assistant may do everything, but nothing goes live without approval.
What to check
- Roles. Who is allowed to approve changes proposed by Ora, and can it follow your existing roles?
- Logging. Is the conversation and the approved change stored together, so you can see why a change was made?
- Data. Which data from integration runs is sent to the language model, and where is it processed?
For Celigo customers, Ora is probably most useful first for error handling and troubleshooting, where the gain is immediate and the risk is low.
Sources
This post was written with AI assistance and reviewed by the editor before publishing.