Okum is an operations layer for agencies and freelancers who run n8n automation workflows on behalf of clients. It deploys workflows, monitors every run, and — when something breaks — tells you whose fault it is before the client asks.
An agency running automations for ten clients is running ten things they do not fully control. Credentials expire on the client's side. Clients rename a column in a spreadsheet without saying anything. Third-party APIs go down. And sometimes the workflow itself has a bug.
All four arrive looking identical: "workflow failed." Two of them are a message to the client. One is a code fix. One is waiting it out. Working out which, every time, by hand, is where the margin on a retainer goes.
The cost is not really the debugging time. It is the thirty seconds where the client asks what happened and the agency does not have an answer. "Your OAuth token expired on the 4th, here's the trail" keeps the relationship. "I'm not sure yet" turns a technical failure into doubt about whether the agency is in control — and that doubt outlives the fix.
It is not aimed at people running n8n only for themselves. The whole design assumes there is a client on the other side who has to be told something.
Okum reads each failed execution from the n8n API, digs into the error object — message, description, code, HTTP code, and the nested cause that usually holds the real API response — and sorts the failure into one of five categories:
| Category | Meaning | Whose side |
|---|---|---|
| Credential Expired | A connected account needs re-authorising | Not your bug |
| Client Data Changed | A sheet, column or file on their side moved | Not your bug |
| Service Issue | Third-party API outage or rate limit | Not your bug |
| Workflow Bug | The failure is in the workflow logic | Yours to fix |
| Needs Review | Okum could not classify it confidently | Unclear |
Classification is deliberately conservative. It returns "Needs Review" rather than guessing, because a confidently wrong blame label destroys trust in every correct label that follows it.
An n8n execution can report success while a node inside it
died — this happens whenever a node is set to continue after an error.
Okum walks the per-node run data rather than reading the execution status
alone, so these surface instead of staying invisible.
Detailed write-up here.
Most tools in this space wait for something to fail. Okum also inspects the workflow definition itself and reports risk before anything breaks:
Failure emails carry the verdict in the subject line — for example, "Okum: 3 new failures — 2 yours, 1 not yours" — so you know whether to open the laptop before you open the email.
Okum pushes a workflow from your library to a client's setup, injecting your own API keys from an encrypted vault, then walks you through authorising each remaining credential with direct links into n8n.
Clients get a plain status page — connected, needs attention, live — instead of an email thread full of n8n terminology.
There are several observability and alerting tools for n8n. They answer the question "did it break?" Okum is built around a second question: "whose side did it break on?"
Two other differences follow from that:
| Plan | Price | Clients |
|---|---|---|
| Free | $0 | 1 |
| Starter | $29 / month | 5 |
| Pro | $99 / month | 25 |
| Agency | $159 / month | Unlimited |
All plans include unlimited workflows and one n8n server. These are launch prices. Payments are handled by Dodo Payments as merchant of record; card details never reach Okum. There is a 14-day refund policy.
Your n8n API key is encrypted with AES-256-GCM before storage and decrypted only inside server-side functions — it is never returned to the browser. Data is hosted on Supabase in Singapore. Okum holds no client OAuth tokens and no card details.
The security page covers this in full, including what is not in place yet.
Okum is built and operated by Ashim Doley, a solo founder based in Assam, India. The name is the Mising word for home — the idea being an operations home for an automation agency.
It is early-stage software built by one person. That is worth knowing before you connect anything to it, which is why it is written here rather than buried.
No. Monitoring and classification read the n8n API and require no changes to a workflow. Deploy and assisted connect apply to workflows you push through Okum.
Yes. Anything reachable with an n8n API key works the same way.
Workflows already deployed to your n8n server keep running. Credentials in n8n stay in n8n. You can revoke Okum's access at any time by deleting the API key inside n8n, without contacting anyone.
Not directly, and this is worth being clear about. Because Okum does not hold client OAuth tokens, it cannot test one for liveness. It infers a dead credential from what the execution error reveals — which means it finds out at the same time the workflow does, not before.
Because being told "not your bug" when it is in fact your bug is worse than being told nothing. Every wrong verdict costs credibility on the correct ones, so unclear cases are labelled unclear.