How it works
Adealia is being designed around a simple habit: tell your AI what changed, keep the useful details together, and let the next person continue from there.
Here's the experience we're building, using an example client onboarding.
Walk through an onboardingThe planned setup starts with a guided connection from Claude or ChatGPT. You'll sign in and choose what to authorize. Your AI will then help you set up the team, invite the right people and create your first client record.
Start with the facts and procedures that matter for that work. Add more as you need them.
Add an existing written procedure or describe how your team handles onboarding. Your AI helps organize it for review, and the team can share the agreed version.
The Library also holds reusable AI instructions—often called skills—so a useful way of doing something can be shared with the next teammate.
Tell your AI what changed or give it the relevant notes. It helps identify the update, asks about unresolved details and shows the saved result.
Information still needs to reach the record. Calls, emails and work in other tools are not automatically visible; you or your AI must bring in the relevant details.
A teammate can ask their AI what's next or open the web view. The record brings together current work, responsibilities, sources and the applicable procedure. Missing answers stay visible, so the person taking over knows what needs checking.
ONE CLIENT. A SHARED RECORD.
Example client onboarding · Planned experience. Sample people, client and records illustrate how it will work.
Maya manages client onboarding at a small design studio. Leo covers for her when she's away. They add their agreed procedure to the Library:
Before a client is ready for delivery, confirm the agreement is signed, accept the brief and brand files, and test website access. Files or access details arriving does not mean they are ready to use.
Their shared AI instruction is simple:
When preparing a handover, include the next action, its owner, what remains unchecked and the source for each important decision. Ask about missing information.
For Willow Studio, the signed agreement is recorded, and Maya has already accepted the brief and brand files. Website access is the remaining requirement. Those earlier checks remain in the history.
Update Willow's onboarding. The website access details arrived in the client's email this afternoon. I haven't tested them. Leo will test access by Tuesday at noon. Here's the relevant email excerpt.
Illustrative saved result
Saved to Willow Studio's onboarding:
The working date and team time zone are already set. If the client, date or owner were unclear, the AI would ask before saving that part. No passwords or credentials are included in this example or stored in the status record.
I'm covering Willow's onboarding. What's next, and what needs checking?
Illustrative response
Your next action is to test website access by Tuesday at noon. Maya recorded that the details arrived, but they haven't been checked yet.
The agreement, brief and brand files are already recorded as complete. The onboarding procedure requires usable access before the client is ready for delivery.
Access status was last confirmed by Maya on Monday at 3 pm. Check the linked update if anything has changed since then.
After testing, Leo reports the result:
Access works. I completed the check. Mark the access requirement accepted.
The saved result records Leo's check. With all required conditions now met, the onboarding can be shown as ready for delivery. If access had failed, the record would retain that blocker and need a next action.
The procedure, current state, source, owner and remaining check. The handover continues in a different AI without treating “received” as “accepted.”
LEARN FROM THE WORK
When a handover or shared instruction misses something, record the correction. The team can use that example to review a better version of the procedure or skill before making it the agreed approach.
GOOD TO KNOW
Supporting both Claude and ChatGPT is part of the first product plan. The connection and handover experience will be tested in each before it is offered.
The goal is to make routine work available through your AI, with clear web pages for viewing and correcting the record. Signing in and granting access remain your actions.
You and your team remain responsible for bringing in changes. We're designing guided updates and shared review to make that easier. A status will show when it was confirmed; that is not a guarantee that nothing has changed since.
Not yet. These pages describe what we're building. Signup and connection instructions will be added when the first usable version is ready.