Airtable Integration

[Oleg]

We are happy to announce Airtable integration. Find below steps on how you can sync your tables with Fibery. We will use a base created from the Product Roadmap template as an example.

Add your Airtable account

1. Click Integrate Type in the app desired to be extended with Airtable
2. Select Airtable App
3. Fill the form of Airtable Account and click “Connect”

Configure record links and change field types

The source configuration screen is shown after adding an Airtable account. Field types are detected automatically, but record links (references to records from others tables) are not detected. So you can provide links and make field types corrections on this step or it can be done later by editing integration settings.

Click Sync Now to start synchronisation.
Check the data after sync and make corrections by editing the source configuration.

Configure Name field

You may need to configure the “Name” field in case if synced Airtable’s table doesn’t have it.
By default, the record id is used as a name, but it can be changed. Find below.

NOTE: A new account should be added if you would like to sync one more table or your Airtable’s table set is changed. We know that it is not very user-friendly, but we still are waiting for Airtable API to be opened for us to simplify the whole flow.

Please let us know about questions or comments.

[Oshyan]

I’m very excited about this! Unfortunately first result is “Could not load app setting. Please try again later.” I’d try to fix it but I have no idea what the actual problem is from the error message.

Update: OK so I was trying it with a fairly big production database, 25,000+ records, initially. Not sure if that’s the issue or what, but I got it to work with a brand new test database with simple test data (CRM, ~50 records). It does create new Types if you want it to (answering my crossed out question below), so that’s great. Can also link to some existing data, though I found that process a little unclear so far.

Main problem is that, in the simplest case where I just created all new Types for Airtable data, the fields that are Relationships just have a record ID instead of an actual connection to the correct other Type/Entity. Something like “recDsd41H4RZRQTpC”. And in some cases there are multi-relationship, and then you just get a string of comma separated IDs: “reciDkG5eB8SZemYH,recwv81yfeykiCLQU,recLk9CQYlFlgcX4k”
Compare to the Hubspot integration where this kind of inter-type linking seems to work fine.

Another thing is that the field types are not maintained in some cases (but are in others). Checkboxes come in, for example, but single-selects don’t. Dates and websites do as well. Mostly it’s single/multi-selects that appear not to. From looking at the Hubspot integration, this seems like generally how it is with integrations to Fibery. I’m not sure whether that’s a necessary limitation long-term nor - more importantly - whether it will cause any issues for the data being referenced/used in other places (e.g. in a calculation). But it does at least make it much less visually distinguishable vs. Airtable in the case of selects/tags.

I’m also wondering whether the sync is change-aware or otherwise incremental, or if it has to fetch all the data from the remote Base to compare against what it has and determine any changes. I would hope the Airtable API would have some kind of date-based change detection that it could use to limit the transferred data. Otherwise trying to keep 25k records in sync is not likely to be feasible, I would think.

All that said, it’s promising, and probably usable in its current form for some (perhaps many) purposes. But without the correct links between entities/types, it has some notable limitations for me. Thanks for the work so far!

Also unclear whether all fields need to already exist in Fibery and just be matched with Airtable data, or if it will actually create new fields that don’t already exist (would be nice).

[Oleg]

Hi, @Oshyan

Thanks for the feedback. Please find my comments below.

@Oshyan

Main problem is that, in the simplest case where I just created all new Types for Airtable data, the fields that are Relationships just have a record ID instead of an actual c

The main issue with Airtable there is no meta data about base since they stopped to issue developer tokens which allow to get that information. That’s why we added the possibility to make corrections to synced schema. It is possible to select field which contains record id and select corresponding link in type selector.

@Oshyan

Another thing is that the field types are not maintained in some cases (but are in others). Checkboxes come in, for example, but single-selects don’t. Dates and websites do as well. Mostly it’s single/multi-selects that appear not to

The same reason for selects. Airtable API doesn’t allow to retrieve information about field types and values list. We can’t do anything with that until they will give us special token which allows to retrieve meta. We implemented automatic field types detection for Airtable, but we can’t detect selects or record links due missing meta.

@Oshyan

I’m also wondering whether the sync is change-aware or otherwise incremental, or if it has to fetch all the data from the remote Base to compare against what it has and determine any changes

We fetch all data from Airtable. There is no information about the changes in Airtable rows. So I am afraid that it can be a problem to download 25K records, but we will test huge tables additionally and will try to fix things.

@Oshyan

But without the correct links between entities/types, it has some notable limitations for me.

Please try to provide link corrections by editing integrations settings as it is done on screen above.

Thanks again for your feedback,
Oleg

[Oshyan]

Ah, thank you! This explains a lot, especially that the linking process can be quite manual and I basically skipped that step hoping Fibery would just figure it out.

So I have now redone it and most of the links do seem to work, which is a good step forward! However some of them still have just the ID name still, as entity name, even though they have correct links to other entities. This makes them less useful for sure. Perhaps I still need to improve configuration? Try replicating this with the simple CRM base and its demo data, that’s what I’m using.

A question about Type linking, before I move on, is whether there is any possibility of auto-detecting this at all. From your description I guess no, at least for now. So it will always be manual, unless the developer token issue is resolved, I am guessing.

@Oleg

The main issue with Airtable there is no meta data about base since they stopped to issue developer tokens which allow to get that information.

@Oleg

We can’t do anything with that until they will give us special token which allows to retrieve meta.

So is this something you applied for and are waiting for them to provide? Or are they *no longer issuing developer tokens" (as your quote above suggests)? If they are no longer doing so, is this a permanent thing?? Any explanation or public article/post I can read about this?

@Oleg

We fetch all data from Airtable. There is no information about the changes in Airtable rows. So I am afraid that it can be a problem to download 25K records, but we will test huge tables additionally and will try to fix things.

Would this be solved by the “developer token”, or just has no solution? Do all integrations work this way?

[Oleg]

@Oshyan

Regarding developer tokens Airtable says:

We’ve received an overwhelming interest in access to this metadata API! We are pausing creating new tokens for the time being as we iron out our infrastructure, monitoring, and policies here.

So we registered for token and wait for response from Airtable.

[Oshyan]

Ah, I see. Thank you for the quick response. I hope they figure it out soon!

[Oshyan]

@Oleg

So we registered for token and wait for response from Airtable.

Still no response on this? It’s been more than 6 months. Maybe email support@airtable.com? Apparently Enterprise customers don’t need an API token either.

The only way to get the fields setup correctly right now seems to be importing first via CSV, where you can define field types more flexibly. The problem seems to be that relation fields need to somehow be setup in advance to be linked to (since you can’t change field type later), but can’t be since they’re based on imported data that itself would be dependent on other data. I suppose an alternative would be to allow more flexible options for field types when setting up Airtable sync, in fact I’m not sure why this is not the case.

Am I missing some obvious solution here?

[Oleg]

Hi, @Oshyan

That’s right. There is no simple way to setup columns and relations for now. Still waiting for ticket. Will try to reach Airtable support. Sorry for inconvenience.

Oleg

[Oshyan]

I understand, but am quite disappointed in Airtable. How rude of them not to let me pull all my data out of their servers and go use some other tool.

More seriously, what I don’t understand is why with CSV import - which has no column type definitions - there is somehow more ability to accurately guess the column type, but also more allowance to change the column type. Why can’t we just do with Airtable integration/sync the same thing we can do with CSV import? That would at least solve much of the issue for me. It wouldn’t be as convenient as it being automatically pulled from column definitions Airtable provides, but it would work just fine once setup.

[Lukas_Hroch]

I have tried integration with AirTable and tried change Name filed to different, but it does not work. There is still record id from airtable. I delete whole integration and try a new one, but with same result.

Could you help, please?

cc @Oleg

[helloitse]

I’m testing out the Airtable integration as an importer and have run into errors as well. The current integration doesn’t allow for any field configuration at all. To my shock/surprise, the tables imported with no reference to any of the linked fields at all.
Is this a bug? Did the oAuth changes break the importer?

[mdubakov]

We changed Airtable sync in December, did you check the guide?

[helloitse]

Thanks for sharing the link here! I don’t believe the link to the guide is not available through any of the modals that appear when setting up the integration so I had not seen it.

That said, " Step 2. Configure record links and change field types" didn’t happen in my workspace. Did I encounter a bug? I initially attempted to import multiple tables at once. Does the field mapping only work if setting up a single table?

[helloitse]

I did some more troubleshooting. When setting up a new Airtable sync/integration from the Integrations screen, I am not prompted to select mapping. This prompt fails for a single table or multiple tables.

However, when you select “Integrate” from the app configuration screen, you are able to select Airtable as an App, choose any number of tables, and configure the mapping.

Beyond that, I don’t think it’s fair to say that the Airtable integration can serve as an integration or importer of any sort if we cannot import the values of certain fields at all. I think I’ll want to create a dedicated post for requested improvements. Generally, the ability to import fields as plain text is better than their complete exclusion. The actual UX trying to move from Airtable to Fibery is smoother if we forgo the “integration” and do a csv export from Airtable.

[kalman]

You are right, installing it from the template behaves differently, which is not intuitive.

Thank you for putting together a post for the requested improvements. Please add me to that post so I can investigate it further.

[helloitse]

Just finished the post: Airtable Integration Improvements