For years, Artificial Intelligence has been driven, even from its own predecessors, as something more futuristic than what’s available, without too many tangible approaches to the day-to-day life of the end user. There have been and are commercial uses, but it was blurry for the ordinary citizen. This year the tables have changed.

The arrival of ChatGPT a few weeks ago started to change the story that OpenAI’s GPT-3 started redirecting from a text a while ago or DALL·E and other renderers this year. And now integrations with the applications used by the average citizen are starting to come.

Personal productivity through AI

Notion, an app that has been gaining integers on its own for years, announced the arrival of Notion AI, a phased integration a few months ago. alpha and with a waiting list that allows creating any type of text from a few simple instructions. From a summarized script to a poem, a press release, a job description, or a situation to update a social network.

Of course, it is based on GPT-3, which combines with predefined instructions to make it easier for the user to use. And while it doesn’t serve to achieve great results in one click, at least for now, it helps in gathering information at the startup or before starting work.

It’s been less than a month since Craft, a vitamin note-writing and management app specifically geared towards the Apple environment (it has released its app for Windows, but not for Android and doesn’t support the mobile web), announced the arrival of Assistant. , An integrated assistant with some default options in the same vein as Notion -of course-: translate text in several languages, expand a text that has already started, create from a few words or a headline, soften or formalize the tone of that text, etc.

It still uses GPT-3, and it was released to users very shortly before the ChatGPT announcement, so it may lose some of its appeal, especially since it limits the number of queries that can be made to the Assistant.

Despite this, it has become too important a practice to ignore. Of course, it doesn’t leave much room for those who want to take it seriously without going through the box: it has a free modality with restrictions that don’t take long to emerge.

Third, a category-leading task management app, specifically for being cross-platform and keeping generosity high to free plan users, Craft: unlike Todoist, an old acquaintance for productivity geeks that hasn’t passed for years, but around the other way: every year just over the previous year It also manages to be better than the industry average.

Although AI came to Todoist (or according to Todoist) six years ago, its approach wasn’t very revolutionary: It was limited to suggesting due dates based on our habits or other task’s date settings. It wasn’t bad, but it wasn’t anything spectacular either.

One of the more flashy or at least more functional and impressive is one of the recently released integrations, A kind of AI-based assistant that allows three options for each task we add:

1. Get hints to complete
2. Convert it to more actionable text
3. Divide into subtasks

The latter is the most useful. For example, a task called “clean house” can be split into the following subtasks with a simple click:

• empty the cabinets
• Pick up clothes from the floor
• Sweep
• To clean the furniture
• Sweep and mop the floor
• Wash the dishes

Do not let anyone take this example literally. Just “emptying the cabinets” as a form of cleaning is not questionable. Or assume it’s normal to have clothes lying on the floor. This is just one example of understanding a request for further action.

What matters is how it opens up hitherto unrelated areas to more easily complete areas without effort or time. A separate problem is that the created tasks or texts are verified and require some tweaking: compared to what we have so far, this would already be a shortcut.

More applications for the visual part are available not only from DALL E, Stable Difussion and the company, but also for the creation of drawings. they know how to re-transform themselves to make maximum use of these capacities. This is the case with Canva, which is often used for the creation of images for marketing or professional social networks, and in addition to images it also integrates GPT-3 to create: copies creatives to accompany them on every social network. It is part of the Magic Write program.

We’ve come so far with GPT-3 and GPT-3.5. If GPT-4 finally arrives in 2023, or at the latest in 2024, it is assumed that most of the current limitations will be overcome and the chances will be much greater and much better.

Some enterprise tools are also opening up to GPT-3, such as Replier, a business-focused solution that automatically responds to customer reviews from places like Uber Eats, Facebook, Shopify, Yelp, but with appropriate wording and semantic understanding. ..

And GPT-4 hasn’t arrived yet.