❯ Question and answer
What happens after an automation goes live, and who maintains it?
At Notkode, the first week after go-live is closely watched: the team monitors the automation, fine-tunes it while it runs for real, documents the flows and trains your team. After that, someone needs to watch errors and changes in the connected systems, because AI models and APIs have dates when they change.
Updated on
What are the numbers, with sources?
- Anthropic provides at least 60 days’ notice before retiring a publicly released model. (Source: Anthropic, model deprecations)
- OpenAI gives at least 6 months’ notice before shutting down a generally available model, and at least 3 months for specialized variants. (Source: OpenAI, deprecations)
- Each version of Meta’s Graph API, which the WhatsApp API is built on, is maintained for at least 2 years after release. (Source: Meta, Graph API versioning)
- In n8n, an error workflow runs whenever an execution fails and starts with the Error Trigger node. (Source: n8n, error handling)
What happens in the first week?
The automation runs with people watching. That is when the cases testing missed show up: a message phrased in an unexpected way, an empty field in the CRM, a customer replying at odd hours. Each one becomes an adjustment.
What changes in an automation over time?
Even if nobody touches it, the things around it change:
- AI models are retired, with 60 days’ notice at Anthropic and 6 months at OpenAI.
- APIs get new versions; at Meta, each version lasts at least 2 years.
- Your process changes: a new product, a new price, a new stage in the funnel.
Who should look after the automation day to day?
Someone on your team who knows the process, with the flow documentation at hand and an automatic alert when something fails. Process or system changes go back to whoever built it.
Common mistakes
- Having no error alert, so the automation stops working and nobody notices for days.
- Leaving the flow undocumented, dependent on one person’s memory.
- Ignoring model or API deprecation emails until the day it shuts down.
- Changing the process (a new CRM field, a new stage) without adjusting the automation with it.
What does Notkode deliver after go-live?
Every Notkode automation and agent comes with 1 week of monitoring after go-live, fine adjustments after seeing the real operation, documented flows and training for your team. An automation project is ready in 2 to 3 weeks on a normal timeline.
Frequently asked questions
Do I need a developer to maintain the automation?
Not day to day: someone who knows the process, with the documentation and error alerts, is enough. Changes to flows or integrations need someone technical.
What happens if the AI model is deprecated?
The flow moves to a newer model and is tested again. Anthropic and OpenAI give advance notice: at least 60 days and 6 months, respectively.
How do I know the automation has stopped?
With an error alert. In n8n, for example, an error workflow runs whenever an execution fails and notifies the team.
Does Notkode make adjustments after delivery?
Yes. The first week after go-live is for monitoring and fine-tuning, with the real operation running.
Last verified on