🤖 Feature Request: AI Generation for Scripts

Similar to the ability to use AI for generating formulas, it would be great to see this for Scripts within automation.

e.g. Create a script that retrieves {{field 1}} & {{field 2}} and updates {{field 4}} with XYZ, then make a GET request and update {{field 5}}.

Reference:

This is not only a fearure request, but a prediction

I like your idea, and it is in line with what every tech company is currently facing: the rush to make AI innovate their app or product. Fibery especially already has a lot in place to suppor that evolution, however:

Need for a co-creative, AI empowered user community active on all levels of the product development and strategy

And regardless of how much I love and see potential in Fibery and its team, I need to say this: In my view the Fibery team is not yet making optimal use of its own user community, developer community and does not focus on building such community. This is what I see as the main reason why there is a lack of collective intelligence and slow growth over the years. This has to change through a radicak co-creative development strategy that builds an egaged community.
Note: the co-creation needs to be intelligent in the sense that it not only keeps adding and fixing small and cosmetic features that users want, but the product development insight strategy and key decisions need to be opened up to advanced users to co-create, in combination with AI on all levels.

1 Like

To me community is self-organization around great product. Notion did nothing to grow the community, they even did not communicate much. It did not stop Notion from explosive growth.

3 Likes

Now now

Lets do some research if that is true, that Notion did nothing to grow the community:

Several articles and resources discuss Notion’s growth strategy, particularly its community-led growth approach. Here’s a summary of the findings:

  1. Notion has experienced significant growth since its re-launch in 2018, with over 20 million active users by 2022, up from just one million reported in 2019. Source
  2. Community-led growth is highlighted as a key strategy for improving customer retention and scaling businesses. Notion has focused on building a loyal user base by providing an exceptional user experience and actively listening to customer feedback. Source
  3. Notion has leveraged its user community to drive growth, foster innovation, and build a brand that resonates with millions of users worldwide. The company has made community building and engagement a central focus of its growth strategy. Source
  4. Notion’s former Head of Marketing, Camille Ricketts, has shared insights on the company’s early community-led growth strategy. Source
  5. Olivia Nottebohm, the chief revenue officer at Notion, has discussed the role of community in organic growth at scaleups. Source

From the information gathered, it’s evident that Notion has placed a strong emphasis on community-led growth, and this approach has played a significant role in its rapid expansion and success.

I know that ChatGPT can answer some questions, but not such questions so far. They started to focus on community when they already had explosive growth.

Anyway, it will be cool to hear top things we can do to enable that.

2 Likes

Thank you. Notion and others are interesting case studies, and obviously there are a combination of factors influencing success. I’m interested to research this further in some dedicated forum topics.

Hope to see this added, as it will give users the power to get much more creative with Fibery and create opportunities where current “native” Automation feature sets currently don’t accommodate (outside of scripting). This is also related to:

Surrounding the ability to sculpt Fibery operations using AI. This especially should help with user adoption of Fibery, since learning curve is steep. @mdubakov

Anyway, continuing the brainstorm for the feature request of this topic (AI Generation for Scripts)
I see roughtly 3 approaches:

  1. Simpel: AI Retrieves from Repository:
  • Workflow:
    1. User inputs function.
    2. AI searches repository.
    3. AI presents script.
  • Technical Requirements:
    • Repository access.
    • Search functionality.
  1. Moderate: AI Combines Scripts:
  • Workflow:
    1. User inputs functions.
    2. AI searches repository.
    3. AI combines and presents script.
  • Technical Requirements:
    • Repository access.
    • Search functionality.
    • Logic for conflict-free combination.
  1. Advanced: AI Develops New Script with User Help:
  • Workflow:
    1. User inputs function.
    2. AI asks questions.
    3. AI generates and presents script.
    4. User reviews and provides feedback.
    5. AI adjusts script.
  • Technical Requirements:
    • Natural language processing.
    • Resource and documentation access.
    • Testing environment.

That’s true, but that doesn’t mean it works for everyone… In fact we can generally see that is not the case, i.e. Notion got lucky.

But I agree 100% with the fact that Notion embraced this community strength after it already happened organically. I watched it happen from near the beginning with Notion, and saw with great frustration at the time how little they fostered and communicated with the community. Then saw them start to talk about their “community led growth” and was, again, quite frustrated. :roll_eyes:

All that said I also agree that Fibery could be doing a lot more to capitalize on/benefit from and foster its community. The Fibery Partners Program was a good step (are you continuing to promote and onboard people actively for this?), but an in-app “Solutions Directory” of some kind, or at least a better way for people to share templates, scripts, etc. would be - I think - quite valuable, both for new users, and for giving existing community a sense of appreciation and reward. Most other competitors already have this in one form or another. The #show-your-space Category doesn’t really do it, though maybe could be leveraged better if it were renamed/adjusted in purpose a bit. Right now it seems more like “Look at this pretty/cool workspace I made”, kind of like the endless Notion “dashboard” shares, except Fibery is less visually flashy than Notion and also not PKM but business focused, so there is a lot less interest in doing that kind of sharing.

Isn’t ChatGPT already pretty good at doing coding though? It’s not like Fibery’s code is proprietary, GPT probably already understands most of it. And Fibery team have already figured out how to give GPT understanding of Fibery primitives for e.g. formula generation. So I think actually this is the way to do it, and the other two would probably be either/both A: less effective, B: possibly actually harder to implement (because ChatGPT already does full-on, from scratch coding).

2 Likes

The partner program is still operational, but a significant issue is that Fibery isn’t widely known, so there apparently isn’t a great deal of bottom-up demand for partner support. Which makes it hard for partners to justify being specialised in Fibery, and therefore less likely to evangelise.
Maybe it was naive to think there would be significant top-down promotion by partners.

Interestingly, I think these are related.
We have previously invited users to share ideas for templates, and encouraged users to show off workspaces that they have configured, but tbh, there was not a great deal of interest.
At the moment, it’s not obvious what the motivation would be for existing users to share a template. In the Notion world, template sharing was partly driven by ‘exhibitionism’, which is not likely to be effective for Fibery given that, as you correctly say, it is focussed on function rather than flashiness.

What ideas do you have in this regard?

1 Like

I never really thought of it as “specialized” in Fibery, per se. Rather one of a set of apps/tools someone would be familiar with and recommend based on suitability for a given client. Most people I know of who do use Fibery regularly are pretty into and enthusiastic about it though, so I would think there would be at least some evangelizing. But you have relatively few partners, an the scope/scale of them seems small as well, so I don’t think there is a lot you can reasonably expect here. It is helpful, but not sufficient.

Yes, that’s true. But I do think there is potential for sharing of functional solutions to have its own appeal, if it is framed and supported in the right way.

To my mind there are two main issues with the “Show your space” Category:

The first is the emphasis on “show”, which orients it toward a different market than Fibery’s current core target market. A focus on sharing of functional components and examples might be a better framing for the current customer focus. E.g. a category simply named “Template Sharing”, or perhaps more creatively (but not necessarily intuitively) “Building Blocks” or something. Keep in mind that the description of the category can help clarify its intent, the most important thing is to get people sharing stuff in there, and the forum Search will help surface it for new people looking for related stuff.

The second issue is visibility, or lack thereof, which negatively impacts both the incentive to share from experts/existing users, and the potential for new users to benefit from that sharing. This may need more help than simply renaming the category here in the forum. Promoting it in e.g. social media, etc., or even linking to it on the Fibery website somewhere, both could help. The ideal solution, IMHO, might be a dedicated “solutions directory” of some kind within Fibery itself, especially during onboarding, but also within the existing Templates gallery. There should either be an entirely separate and also searchable “From the community” section there, or perhaps mix-in the community stuff in each appropriate category with obvious indication of which is official and which is community, and some quick and easy filtering to only show official solutions. Many other companies with similar tools show models of how this could work, e.g.

Notice the link at the bottom of the category list on the left back to Templates, and on the Templates page a link back to this “Universe”. As well as the individual Profiles pages with photo, etc. It’s a very “full” solution, maybe overkill for Fibery, but quite nice…

Or Coda’s approach:

Notice that they simply have an “Author” on each one, often Coda itself, but could be others, they don’t have a major distinction between “official” and community stuff, . They also let users charge for their work, which is another incentive to sharing. Again a very comprehensive solution that may be overkill for Fibery at this stage, but interesting as a reference point.

Also note that both of those were found easily just by searching “Airtable templates”/“Coda templates”. Searching for “Fibery templates” just pulls up the official ones (not surprisingly).

Now as to how things get into that library/gallery, if there is some kind of approval process, etc., that I can’t say, and I recognize it would take some employee time to manage. It may not be worth it. But I do think it’s a useful part of a customer-led growth strategy, which may be Fibery’s best hope for now, given that it is hard to differentiate and get visibility in this crowded market, especially when what you are offering people is a cool Lego box and not a super fancy out-of-the-box, turnkey solution that is easy to demo.

1 Like

I guess my point was that there has not yet been any evidence yet that partners are generating new leads at a significantly greater rate than other Fibery users.

As for a ‘solutions gallery’ including community-provided templates, I see that there might be two reasons for users to create solutions to share: 1) exhibitionism 2) reward from the ‘consumers’.
My point was that Notion did well initially with the former, and I imagine Airtable does well with the latter, given their massive user base, which means you only need a fraction of a percent of AT users to buy your template for it to be a reasonable proposition.
I don’t know if we would get the same traction, and it would require some fairly serious effort to establish.

FWIW, I have created a load of templates (some from before I started working for Fibery and some after) and the number of people who have chosen to install a template I created is depressingly small :rofl:

1 Like

Fair point. Have you actually looked at this metric? Do you have a clear way to measure it?

Yes, the effort for such an in-depth solution admittedly does not seem worth it at this point. However I do think there are lower effort ways of getting at least some of these benefits, starting to grow the Fibery sharing community organically. One thing that would help, I think, is renaming and changing the description of that forum category, as I said. Another would be simply adding a link to that category within the Fibery Templates browser, e.g. “Browse Community Templates and Workflows” or something. Then Tweet about it, all that.

Sure, but Fibery’s overall userbase is also… sorry, but depressingly small. (at least as compared to many other players in the industry and given Fibery’s comparative feature set) :confused: So it’s a bit of chicken and egg here. Having more templates and more access to user-created templates in place won’t necessarily give you tons of new users, but having those things in place will make it easier and more interesting for any new users you do get. The effects may be small, all but invisible, especially at first, but if the steps to be taken are simple and low effort, I think they’re worth it. At least as an experiment. :man_shrugging:

1 Like

I don’t know if I can concur with that assumption.
I feel like we have a good number of templates already, and I am not convinced that it is the lack of templates that is discouraging new users.

And maybe the quality of user-generated templates will actually be below average, to the point where they don’t generate a positive experience for new users.

I didn’t say it was. I said lack of visibility is a key issue. Quality might be too, sure. I’m just saying that visibility and messaging around Template Sharing and accessing Community Templates could be improved, which may help both existing and new users. I think the concerns you raise are legitimate but solvable fairly easily with appropriate language, framing, etc.

I feel like I’ve argued my case as much as I can at this point, so I’ll leave it at that. :smile:

1 Like

Could be my own lack of experience, but I recently tried to generate a script with chatGPT for use in an automation rule (unsuccessful so far so if anyone has ideas - I’m all ears). The idea was to look at a LinkedIn URL entered into an entity field and use that to access the LinkedIn API to get the First Name and Last Name from the profile and populate two corresponding fields in the same entity. The script that chatGPT generated was not particularly specific to Fibery as far as I can tell.

async function getLinkedInLastName(profileId) {
    const accessToken = 'YOUR_ACCESS_TOKEN'; // You need to obtain this via OAuth
    const url = `https://api.linkedin.com/v2/people/(id:${profileId})`;

    try {
        const response = await fetch(url, {
            method: 'GET',
            headers: {
                'Authorization': `Bearer ${accessToken}`,
                'Content-Type': 'application/json'
            }
        });

        if (!response.ok) {
            throw new Error(`Error: ${response.status}`);
        }

        const data = await response.json();
        return data.lastName; // Assuming the API returns a structure where the last name is under 'lastName'
    } catch (error) {
        console.error('Error fetching LinkedIn profile:', error);
    }
}

// Example usage
const profileId = 'some-linkedin-profile-id';
getLinkedInLastName(profileId).then(lastName => {
    console.log('Last Name:', lastName);
});

chatGPT said as much:

Remember, this is a highly simplified example. The actual implementation would need to handle OAuth authentication and conform to the specific constraints and capabilities of Fibery's scripting environment.

So all this to say, a Fibery-specific AI for scripts would be helpful I think.

Yes, I didn’t mean to suggest that work would not be needed by Fibery team to make this function well. I was just saying that of the 3 approaches Yuri outlined, the 3rd one that he thought was “most advanced” actually seemed to me to be the least amount of work and likely most effective and flexible. Essentially it would just be Fibery taking advantage of existing GPT capabilities and giving it a bit more info/context about Fibery workflow, primitives, etc.

I could be very wrong, but I even suspect that if you gave the current, public ChatGPT a detailed enough prompt, it might well be able to code your solutions properly already. But prompt engineering can be quite tricky and at times unintuitive, especially for more unusual cases like this. So better to let Fibery team handle it, which hopefully they will (we already have AI formula, so…).

1 Like