Your custom GPTs retire on December 11, and the actions don't carry over.
We inventory your workspace's GPTs, migrate each one to a plugin, rebuild custom actions as authenticated connections, and test every migrated GPT against saved prompts before people switch over.
You have this problem if…
- Your team runs real work through custom GPTs: an HR policy assistant, a sales helper, a project assistant.
- A migrated GPT came out private, sharing is disabled, or someone migrated a shared GPT without telling its owner.
- A GPT calls an outside system through a custom action, and the migration dropped it.
What you get
- An inventory of the workspace's custom GPTs: owner, users, instructions, knowledge files, connected apps and custom actions
- Each GPT migrated to a plugin, with its instructions as a skill and its knowledge files as reference files
- Custom actions rebuilt as authenticated connections: an MCP server or a webhook, with credentials stored outside the prompt
- A parity test for each GPT: saved prompts run before and after, with any differences written up
- Sharing and publishing permissions set for the workspace, and a one-page note for staff
How it is done
- 01
Inventory
We list every GPT, what it does, who depends on it and what it connects to.
- 02
Decide
Each GPT is migrated, merged with another, or retired. Not every GPT is worth moving.
- 03
Save the baseline
Before anything moves, we run a set of real prompts through each GPT and keep the answers.
- 04
Migrate and rebuild actions
Instructions and files move with the built-in migration. Custom actions are rebuilt as authenticated MCP servers or webhooks.
- 05
Parity test
The saved prompts run again on the migrated plugin. Differences are fixed or written down, because the GPT's selected model doesn't carry over.
- 06
Permissions and handover
We set sharing for the workspace, write a short note for staff, and remove our access.
What usually goes wrong, and how we handle it
Custom actions do not transfer
OpenAI’s migration FAQ says rebuilding them "may require a custom MCP server and technical setup". This is most of the work in a migration, and the part a click-through migration leaves broken.
The selected model does not carry over
A GPT that depended on a particular model can answer differently after migration. The parity test is how you find out before your staff do.
Sharing settings do not transfer
Migrated plugins can come out private to whoever migrated them. Permissions have to be set again, deliberately.
Enterprise has a different date
Enterprise workspaces with an approved deferral have until February 11, 2027. Everyone else has until December 11, 2026.
Why you can trust us with this
Custom GPT retirement was announced in September 2026, so nobody has a long record of migrating them. What matters is the work underneath: authenticated connections between assistants and business systems, and testing behavior before and after a change.
Our founder owned CRM, order-management and payment workflows at BYJU’S and Freshworks, where an integration defect is a failed payment, and built the quality gate that checks AI-generated content before it reaches users.
What we don't do on this package
- No migrating GPTs that handle health information (PHI).
- No changes to what a GPT does beyond parity, unless you ask for them.
- No shared passwords. You add us as a temporary workspace member, or we work on a screen share with your admin.
Questions
Can’t we just click "Migrate to plugin"?
For a GPT with instructions and files only, often yes, and the inventory will say so. The work is in GPTs with custom actions, GPTs many people share, and GPTs whose answers people rely on.
What happens to the knowledge files?
They become reference files on the plugin. We check that answers which depended on them still cite the same content.
Who does the work?
Suresh Rajendran, our founder, does the engineering on every engagement himself. Nothing is handed to a junior engineer or a subcontractor.
Where does our data go?
We work inside your accounts: your cloud, your repository, your model-vendor keys. Where a vendor offers zero data retention we turn it on, and your data is never used to train a model. Our access is removed and credentials are rotated at handover.
How do we pay?
The first step is paid up front. Builds are 50% to start and 50% on acceptance. We invoice from Destello AI Incorporated in US dollars, payable by ACH or card. If you go ahead with a build within 30 days of the first step, its fee is credited.
What if the result doesn't meet the bar?
Acceptance criteria are agreed in writing before work starts, and the second payment is due only when they're met.
Will you sign our NDA?
Yes, usually within one business day.
Start with the inventory and migration plan.
$400, credited to the build if you go ahead within 30 days. Book a 20-minute call, and you'll have a written scope with acceptance criteria within one business day.
Or email hello@destello.us or call (628) 202-5322. We reply within one business day.