September 2023 Product Update
New Ways to Send Notifications
Historically AidKit notifications are configured to happen when a particular participant meets a specific criteria, such as when we’ve sent funds or when we need someone to submit an alternate form of eligibility.
Sometimes, however, our partners want the ability to send a notification when they’re on the phone with someone or are otherwise communicating with them. To facilitate this, we’ve added a couple new features:
- Inline NotificationsNow, as part of a review flow we can configure a component for your staff to send a specific notification that includes information relevant to a specific applicant and a link to collect any further information that may be needed.
- Dashboard NotificationsIf you want to issue a large number of notifications manually, Dashboards can also be configured to include buttons to send many notifications without dipping in and out of a participant’s profile page.
We still recommend automated notifications where appropriate as they ensure people don’t slip through the cracks but we recognize that every program is different and needs different tools!
Improved Ways to do Translations
Language accessibility is critical to ensuring that everyone has access to help and as a result we invest a lot in ensuring our translations are high quality and accurate. Traditionally, this has involved either bootstrapping translations for a program from Google Translate and then having human editors review or using language experts from the start to establish a translated baseline.
With recent advancements in large language models, we can bootstrap translations in a far superior way to previous forms of machine translation by prompting the models to use specific terms in translation or to ensure that the translated text does not use terms above a certain grade-level.
So while expert review of language is still critical, we’re hopeful that these new models can reduce the burden to get to high quality translations that are both correct and accessible to people from a variety of different backgrounds.
As we continued to work with a multitude of great research institutions across the country, we continue to expand our capabilities to meet the needs of robust statistical studies and analysis. In addition to pre-existing research focused questions like Likert Questions, we’ve recently added the ability to randomize the order of question options to ensure that people’s tendencies to pick the first option doesn’t bias research data. This is now available for any multiple choice question across the platform.
Dropdowns with Millions of Options
For a while, AidKit has supported the ability for a participant to select an option from a (large) dropdown — often used in scenarios where you might want a participant to select their job role from a list of a few thousand jobs.
In some scenarios, the set of options may be much larger — in the millions. In this scenario, it doesn’t make sense for a user’s browser to download the list up front because it would be prohibitively large. So instead, we’ve now built the ability to do these large searches server-side in a way that is both fast and accommodating to fuzzy matching so that people don’t have to get everything right 100% of the time.