 | | We’d Love Your Feedback on the AI-Powered “Ask a Question” Setup Assistant | | We've recently expanded the capabilities of the "Ask a Question" assistant available on TeamDesk setup pages.
In addition to reading your existing database configuration, formulas, views, forms, triggers, actions, and other settings, the assistant can now make configuration changes directly in your database when appropriate.
We'd love your feedback:
* Have you tried the assistant?
* How useful were its explanations and recommendations?
* Did you use it to make changes to your database setup?
* Were the suggested or applied changes accurate and helpful?
* What worked well?
* What could be improved?
To try it, simply click the "Ask a Question" button on any setup page and start chatting with the assistant. | | | Comments |  | Lp 6/9/2026 11:42:43 AM Works great and very helpful with both planning out changes, and helping update our internal documentation. Great addition!
|  | SDM IT 7/9/2026 10:25:16 PM This integration of AI is something I have been dreaming of for the past years.
I am the single person handling the business' data management needs. On Teamdesk, I can construct a DB with an initial basic aim, put it into operation, gather feedback and new scope, then improve it. However, in making rapid changes I often skipped documenting the changes, let alone writing and rewriting the DB description and functional spec. I have been thinking how troublesome it would take a new developer to take over the databases. And I thought it would be best to have AI browse my database structure and write the description for the new developer.
Today I have see that my wish is fulfilled. I asked the AI assistant to write a functional specification of the database, I got the following:
"Detailed Functional Specification
Database: KAS & BANK
1. Purpose
The KAS & BANK database is configured to automate the handling of cash and bank accounting source data. Its purpose is to:
- maintain master data for bank accounts and cash books
- ingest bank statements and cash transaction imports
- parse and organize transaction details by source format
- compare source totals against calculated detail totals
- flag discrepancies for review
- control access per account/book and per user
- prepare and post validated data into Odoo
The database description explicitly states “Cash/Bank Data Accounting Automation” on Database properties. TeamDesk databases typically represent business processes through linked tables and role-specific interfaces, which is consistent with this design 1 2.
2. High-Level Functional Scope
The database supports two main operational domains:
2.1 Bank processing
Handled through:
BANK
Bank Statement
bank-specific detail tables:
_BCA
_MANDIRI
_BRI_CMS
_BRI_QLOLA
2.2 Cash-book processing
Handled through:
KAS
Kas_Bulan
Kas_Kegiatan
Kas_Kegiatan_Import
2.3 External accounting integration
Handled through:
- Odoo fields and URL formulas in BANK columns
- Odoo fields and URL formulas in Bank Statement columns
- Odoo-related support tables:
- Odoo Data
_Odoo_accounts
_Odoo_partners
_Odoo_products
_Odoo_statement_line
_Odoo_taxes
User_Odoo_accounts
...
"
Wow, this is just fantastic! I cannot write a better introduction than this.
Thank you for the great new feature, Teamdesk!
| | SDM IT 7/9/2026 10:27:56 PM Only one additional wish: the pop-up window can be maximized and the font size can be made larger. Thank you again!
|  | Jorge Solá 7/30/2026 3:49:43 AM It is certainly a wonderful addition. I hadn't thought of SDM's use of it (writing database documentation), but will try that.
So far I have only used it to build an online form. Here are my impressions:
- It can save a lot of time, especially if you are building something from scratch
- On the other hand, it struck me as not so smart. I'm aware that with AI you have to be very specific in your prompts to get the desired results, but in my limited experience so far with this AI I found that: 1) it took a long time to respond & make changes, even small changes; 2) it talked too much for my taste, & did too little in each step, so many steps were necessary; 3) it did not go above & beyond my instructions to "surprise" me with genuine improvements that I had not thought about, but rather the opposite: I had to state every little detail in my prompts to get it to act as desired & expected.
- In summary, I think it's a wonderful tool, & a wonderful step in the right direction, but right now it's still a little dumb. It's like having an assistant—very eager and talkative, but a little dim—who you have to give very detailed instructions to so he'll understand you properly and not botch the job.
|  | Dean Guidry 9/1/2026 9:10:45 AM Its made a big difference for me in creating custom views and relatiionships. I aways struggled with relationships. If only I could bulk upload docs and export multiple docs to claude, this would be golden for my law practice. As of now, I’ve got to download one doc at a time into a file folder and then slide that over into Claude manually…pain in ass but still takes less time than manually organizing and analyzing complicated documents, especially for discovery. For discovery, Claude is a god send but I need the heavans to open up and give me bulk upload to teamdesk and direct export of docs to Claude. I would be force to deal with with that functionality.
| | Jorge Solá 9/1/2026 9:44:22 AM Hi, Dean.
One idea to bulk upload documents to TeamDesk is this (I haven't tested it):
- Have a watched folder on your computer where you store files to be bulk uploaded
- Write a little script that runs on your computer once or twice a day, scans that folder for new files & through REST API creates a record in TeamDesk for each file.
- Alternatively, you could run the script on demand & link the uploaded files to an existing TeamDesk record, for instance, the parent record that all those files should refer to.
As I say, I haven't built this before, but it should be workable.
Let me think about the 2nd part (exporting multiple documents to Claude for discovery) & I'll get back to you.
| | Jorge Solá 9/1/2026 2:17:42 PM For the 2nd part (exporting multiple documents to Claude for discovery), here is a suggested workflow (disclaimer: I haven't tested it): - Let's assume you have a parent table with the case record & a child table containing a bunch of documents related to that case (uploaded manually of through the above script) - You could select in the child table the documents that you want to send to Claude for discovery - Through a series of Call URL actions, all the selected documents will be sent to Claude - As Claude responds to each file that is sent, you would capture & store Claude's id for that file - After all the files have been sent to Claude, a subsequent Call URL would send Claude instructions on how you want your files analyzed, referring to each one of them by the id that you received from Claude & explaining that they are all related to the same case & need to be analyzed as a whole Since this conversation is a departure from the intent of this thread, if interested we can continue it elsewhere. You can contact me here: https://www.teamdesk.net/experts/24 | | Dean Guidry 9/2/2026 8:52:49 AM Jorge, I copied your entilre response here into Claude and it built the path for me to accomplish this but I do not have the confidence with this to implement it. I’ll contact you soon to inquire.
|  | L. C. Parker 9/2/2026 10:06:07 AM | | L. C. Parker 9/2/2026 10:31:08 AM | | L. C. Parker 9/2/2026 10:32:51 AM You have to copy paste the URL into another tab if it gives you a hotlink error
|  | Kateryna Dmytrenko 9/8/2026 6:35:19 AM | | L. C. Parker 9/8/2026 11:01:59 AM @ Kateryna Dmytrenko
Awesome implementation, enjoyed reading this and your other blog post. Thank you for this!
|
| | Back to Search Results |
|
