Tape is becoming AI-native. We’d love your input ✨

📢 Info & Announcements
1

Tape.ai
Tape.ai3200×800 31.3 KB


AI is changing how people work.
How decisions are made.
How teams collaborate.
How routine work gets done.

We have been thinking about this shift for a while.
What we see is that AI works better when it understands context.

Not just text, but how work is structured. Who has access. How things are connected.

Tape brings these pieces together. All data in Tape is structured by default, and Tape’s flexibility supports complex processes. At the same time, granular permissions let teams control who can see and edit what. Combined with other core features, Tape makes it possible to manage company-wide work in one place.

A due date is not just text. It is a field with meaning.
A record is not just a note. It is part of a process.
Communication is not separate. It stays connected to the work.
This context is what allows AI to understand how work actually happens.

When we started Tape almost seven years ago, we were not building for AI. We were building to help people work better together. Looking back, that foundation matters more than ever. The same structure that makes work easier to manage also makes it easier for AI to understand and act on it.

That foundation makes it possible to use AI at work in a more useful way.
Tape is becoming AI-native. Right where work already happens.

Status

AI in Tape is not new. Some teams have been using AI through automations and integrations for years. You can find a guide here

What comes next is not a single feature. It is a layer across Tape. That means AI is not limited to one place. The existing features and logic of Tape continue to work.

You can move fast with AI, for example by creating and updating records through a chat conversation. At the same time, the underlying data stays structured and connected. Work does not turn into a black box where everything lives inside a chat thread.

In Tape, you still get notified when a record you follow is updated. You can go back to tables or records to review open tasks, check details, and understand what changed. Visibility will remain part of the experience. This overview makes it easier to stay in control and helps users feel confident that work is progressing as expected.

The foundation is already there. While development stays focused on the new record experience and the mobile app, the product team is starting to define the first native AI experiences in Tape.

How we see AI in Tape

Right now, we see five main areas.

  • Build with AI
    Use plain language to create workspaces, automations, dashboards, or calculations. Build even complex setups like a CRM, then refine them with AI or manually.

  • Agents
    AI teammates that can handle work in the background. They handle multi-step tasks and take action based on your setup and permissions.
    Example: A new support ticket comes in. The agent checks your help center or existing records, suggests a reply, routes it to the right person, or sends a response.

  • AI assistant
    Chat with an assistant that understands your organization and its data. Create and update records through conversation, analyze your data for insights, or write a meeting agenda.

  • Smart features
    Search across Tape using natural language and full questions. Use AI in text fields to improve writing, reformat text, translate, and create summaries. Fill database fields automatically with AI Autofill.

  • AI integrations
    Connect external AI tools and services directly to your workflows. Keep your data in sync and use the right tools where they add value.

Building with the community

Tape is community-driven. Real use cases and feedback help decide what gets built.
The last major release, relation fields in the new record experience, came from your ideas and feedback. That is how we want to shape AI in Tape, too.

Here’s what we’d love to hear:

  • What use case would create the most value for your team?
  • Which AI tools or setups have worked well for you?

Feel free to share anything in the comments. Ideas, examples from other tools, screenshots, videos, workflows, or early thoughts. Even if something has already been mentioned, repeating it helps us understand what matters most.

We will not build everything at once. It is important to understand the right use cases early so we can make the right decisions from the start.

Nothing is fixed yet.
The direction is there.
Now is the right time to help shape what AI in Tape becomes.

Tape will help your team roll out AI across your company in a safe and flexible way, without losing structure or control. Thanks for being part of what comes next. We’re excited about what’s ahead.

16 Likes
2

This is exciting.

A native MCP would be great so we can pull and push data from inside/outside of Tape.

5 Likes
3

Would this run on API keys? If so, please include all the major players so users don’t get locked out for using a provider that’s unsupported.

3 Likes
4

One thing I’d love to see from the AI assistant: natural language record creation with gap detection.

Imagine typing “Joe from Joe’s Restaurant needs 100 eggs and 50 bacon, net30, ship to 123 Main St” - and the AI parses that into a client lookup, order record, and line items automatically. Then it pops up asking the questions it couldn’t resolve from context, like carton size or whether they are white or brown eggs.

Unstructured input in, structured records out, smart follow-up for anything ambiguous. That would be a game changer for teams doing intake over the phone or in the field.

…

We currently use a tasks app to dump new work details into so it doesn’t get lost in the shuffle. This could “put the toys away” so to speak.

4 Likes
5

Yeah, that sounds really promising! AI could really help here and make Tape even more essential for us.

This AI use cases came to mind that we’ve been discussing in our team over the last few days and that would be especially useful for us:

Create records from natural language
Something like: “Create a meeting next Tuesday at 10 am with this title, these attendees, and this context.” It would be even better if this also worked through voice input.
I think a lot of CRM documentation is missing not because teammates do not care, but simply because they are busy.

AI models and providers
In my team, there is always discussion about which AI model is best. We all started with ChatGPT licenses, then some wanted to switch to Gemini, and now others prefer Claude. In the end, it often feels like a matter of personal preference, and sometimes one provider is simply ahead for a certain use case. It would be great if Tape did not lock users into one model or provider, but instead made it possible to switch depending on what works best.

AI for repeated support questions
We get many similar questions from customers through our ticket system. If an AI agent could handle this like in your example, that would save a lot of time.

I’m sure I could think of many more, but overall, this feels like exactly the right direction, and we’re really excited about it.

9 Likes
6

What will the monetization strategy be? You can’t just add it for free because of the up-stream costs.

6 Likes
7

MCP’s suck because of the context penalty. Just point the AI at the developer docs, give it an API key, and tell it to build it’s own skill using bash helpers.

3 Likes
8

@Leo and Team, this is awesome!

So happy to see you are heading toward this AI-native place with Tape.

I completely agree with the five main areas that you listed for utilizing AI in Tape. I think most of my quick gut responses below are covered by the categories you’ve defined.

Based on conversations I have with clients, the initial AI Requests that I think of would include things such as:

  1. Ability for AI to produce accurate data/reports from plain text requests.
    When the CEO asks for some random KPI that was never explicitly tracked before…but a simple query to AI can understand the company specific data structure and context, then provide a quick answer in a fraction of the time that it would take a human user to go run three reports and calculate an anwser.
    The big nuance here is that the AI can’t just understand overall Tape field type structure to be accurate. It will need to understand the individual client data structure and custom definitions.
    eg. I have one client that would want such ‘quick questions’ answered using their field called Gross Sales Amount while others would probably expect Net Sales values to be reported on.
    How is the org specific AI memory structure built so that the AI can learn to get this right without having to ask for explicit clarification every time?

  2. Ability for AI to provide high level summarization of data, or resource allocation and prioritization.
    I think of this as the AI being able to understand everything that is tracked as needing to happen, and then being able to accurately tell users what are the most important activities they should be working on.
    In these early days of AI, lots of people don’t really understand how to tell AI to ‘go do things,’ but rather, they want AI to validate or tell them what they should be doing.
    This can get fuzzy in a hierarchical data environment because these requests may require analyzing info from more than one app/data structure within a Tape account.

  3. Ability to recommend/add effective automations without explicit prompting.
    A major frustration with low-code platforms like Tape is that the tool doesn’t automatically handle ‘easy’ things like checking for/avoiding duplicates.
    Users shouldn’t have to build new Dupe Check flows every time they set up a Contacts app. Or they shouldn’t have to think about adding an Overdue Task Reminder flow anytime a Task app is created.
    It would be great to see an AI tool that can understand the context or desired end result of what is being built, then make recommendations on automations to achieve that goal without any developer having to sit down and remember/build it all manually.

  4. Ability to recommend/add effective automations from plain text.
    Much like the example above, but if a developer does think of something and wants to prompt in plain text they should be able to. It would be great to have the ability to just say “add a email and phone value searches to the duplicate check automation” then the AI understands what is needed there. It needs to know how to get clarity when needed and then build it for the user.

  5. Ability to perform ‘health checks’ and proactively flag if calculation fields or automations are broken in a system. Bonus points if it can provide recommended fixes for the flagged issues.

  6. Ability to create an official ‘onboarding process’ when AI Agents are created.
    We’ve experienced many frustrations when AI tools try too hard to create a seemingly simple and fast process to get up and running. If I onboard a new human team member, I expect several hours will be required to teach them the context and expectations of their assigned job role. When AI Agents are treated that same way, they can be extremely successful. But when there is no opportunity to provide a detailed set of instructions, feedback, and proper onboarding, the only possible outcome is error and frustration. Please don’t allow creation and deployment of ‘Agents’ that get trusted with important business data with only two sentences of instructions.

There is much more that could be said here, but I’ve rambled enough for now.

I’m very excited to see where this conversation goes and what the genius minds behind Tape create for this community.

5 Likes
9

I should also add, for me it is very important that there are clear differences between AI features that are explicitly intended to help end users versus AI that is designed to help developers and ‘super admins’ of a Tape organization.

If I am a developer putting fields together while building a Projects app, I would like to be able to say something such as, “I need a sum total of all invoices connected to this project.” The AI should understand that it needs to add a calculation field getting a mathematical sum from the related items. It needs to know how to reference centralized Tape documentation on how calculation fields work to do this properly.

If I am an end user, however, and I’m just looking at a table front end layout of that Projects app in a Tape workspace and I asked the same question, I wouldn’t want an AI to run off and add a calculation field (potentially duplicating something already there). I would want this AI use case to simply run a filter, pull a report, and display an answer for quick reference, but not change any app structure.

That is probably covered in your differentiation between “build with AI” and “AI Agents”…but a clearly defined and separate process for using those AI features will be critical in my estimation.

2 Likes
10

A few things I’d really love to see.

It would be great to have a chat experience similar to Claude or ChatGPT, where you can talk through ideas and then turn the outcome into records in Tape. For example, if I’m planning an offsite, I could work everything out in chat first. Once the plan is ready, it could be transferred into Tape automatically with the event details, the members involved, and the related tasks.

Creating or updating records from natural language +1

Skills or something similar would be a huge improvement because you wouldn’t need to re-explain the same context every time. The AI could already know how I work, what kind of output I need, or what style to use.

An agent for my daily task overview would also be incredibly useful. It would be great to get a clear overview of all the tasks my team has planned for today, along with their current status, so I can immediately see what is on track and what needs attention.

Thanks for the opportunity to share feedback. I really appreciate that we’re always asked for input, and it’s also really cool to read the ideas other people in the community have.

P.S. And for the Tape team: an AI that summarizes all the feature ideas in this post :wink:

3 Likes
11

I’m excited about this!!

5 Likes
12

This is exciting!

We’ve leveraged some workflows using AI, both helping us to create incredible tools, as also integrate to Tape, receiving clean data from AI, OCR extracted.

From this OCR experience, I would add that it would be awesome if AI could also read the files inside Tape, and maybe structure data into other fields / apps from this.

Offering an external chatbot for general public to question things after confirming ID, or any user if asking public apps.
In this use case, I am thinking about a person who wants to know about some application made using Tape webforms( for example), and the chatbot can search for the application and answer the user.

We understand AI needs an accurate drive, so the success of this project will depend on how each tool can be driven by the developer or master user, what will reinforce the necessity of considering user levels for AI usage and setup.

I see a great future ahead, I am happy with what I see!

:brazil: :brazil: :brazil:

2 Likes
13

What about a native CLI? :slight_smile:

2 Likes
14

Thanks a lot, everyone. Your thoughtful comments already make a real difference. This is exactly the kind of input that helps us build something that doesn’t just sound useful, but actually fits into your day-to-day work. :100:

@andreas-globi thanks as well for sharing your experience with MCP’s, and for the very fair question on monetization. We haven’t made final decisions on that yet. Right now, we’re watching the market closely for fair and flexible model options.

So if any of you come across a provider or model worth looking into, feel free to drop it here too.

Please keep sharing. Even a single sentence like “the part that kills my workflow is X” gives us a lot to work with and helps us understand which AI capabilities would create real value inside Tape.

Thanks again. It’s really amazing to see this engagement!

Best,
Leo

5 Likes
15

Sounds brilliant. A few things that would be really useful:

Being able to analyse attachments within items, like PDFs, images or documents, and then summarise them or answer questions based on the file.

It would also be great if AI could look across linked items, comments, activity history and attachments, so it understands the full context rather than just one field or one item.

Another really useful thing would be AI suggesting next steps or spotting missing information. For example: “This task is missing X” or “This project looks blocked because X is overdue.”

Summarising long item histories, comments or meeting notes into something short and useful would also be really helpful.

And being able to ask questions like “What are the biggest blockers across active projects?” or “Which tasks have gone quiet in the last 14 days?” would be great too.

1 Like
16

LLMs will become commodity over time. Would be great if users can bring their own LLM and connect it. Especially given the requirements in Data Privacy, they could hook up their own AI instance.

So clients Tape instance hook up to their BYOAI cloud service?

(I mean thats what we do right now in tape automations using external AI.)

The real difference - not just for AI, but also for building, running and maintaining Tape Apps - would be to get access to calculation field code and tape automations via API too.

3 Likes
17

I am using AI to analyse data in Tape already. But my approach seems a little different: I use the API function from the workflow to connect to my own API server. This server connects with my N8N environment. In N8N I can connect not only to almost every AI provider, but also many other services. The response from N8N is structured and my API server will return this JSON to the workflow. In the workflow fields are updated.
A direct connector from N8N to Tape could be helpful to create AI workflows without the need to use the API. I know it also possible to do this using a webhook, but it is less secure so not preferred.

2 Likes
18

Thanks @knobel for your feedback and for sharing your setup!

Thanks as well, @dirk_s for your ideas.
And good news, writing scripts for calculation fields via the API already works: Calculation Field | Tape Developers

19

Hi @andreas-globi

I’ve been following you for years, and your take on MCP’s really stuck with me.

I’d mostly thought MCP’s were just the right way to connect AI to tools, but your post made me rethink that and look into it a bit more.

When you say “context penalty”, do you mean that MCP’s load a lot of context upfront, things like tool schemas, API descriptions and instructions, even when only a small part is actually needed for the task?
Since that context is defined by the MCP provider, it ends up feeling like a bit of a black box. You can’t really see what’s actually being added and that could naturally make things more expensive and less accurate, since the model has to work through a lot of irrelevant information?

I’d also love to understand your alternative a bit better. Could you walk me through what you mean by pointing the AI at the developer docs, handing it an API key, and letting it build its own skill with bash helpers? Does that mean the AI, via the skill, gets to decide for itself what context it actually needs for this tasks, rather than having everything loaded upfront the way an MCP does it?
In that setup, does the skill ever need to go back and re-check the docs to see if anything’s changed, or is that more of an edge case?

It would be really interesting to hear if I’m asking the right questions here. Thanks for sharing your take, it’s really got me thinking.

3 Likes
20

Hey everybody,

I just wanted to drop a note saying that myself and @Martin_E have launched a new product called Syncello which brings powerful AI features to Tape.

2 Likes