# Lindy AI for Building Apps: Internal Tools From a Chat Request

> How Lindy builds internal apps and live pages from a chat request: credit bands, per-app permissions, three visibility levels and the restore limit.

_Source: https://professionalstoolkit.com/articles/lindy-app-builder — The Professional's Toolkit · updated November 2026_

---

Lindy (often searched as Lindy AI) builds internal tools on request. You describe the tool in Slack or in Lindy's chat, and it returns a working web app with its own link, saved data and versions you can restore [1]. No code is written by you. Lindy's pricing page places internal tools in its big-build band of 1,000 to 2,500 credits, while basic HTML artifacts fall between 250 and 1,000 credits [2]. Our read: Lindy Apps, launched in October 2026, fits a team that already runs Lindy on its connected tools and wants a tracker, a lookup screen or a dashboard without waiting on engineers. It is built for internal tools.


**1,000-2,500 credits** — Lindy's pricing-page range for internal tools and complex HTML dashboards.



- **1,000-2,500** — Credits for an internal tool, per Lindy's pricing-page band
- **3 levels** — App visibility: only you, the workspace, or a public link
- **Oct 2026** — Lindy Apps and app webhooks launched


## Step one: ask for the tool in a Slack thread or the chat

Building starts with a plain request. You ask Lindy in Slack or in its web chat for an internal tool, and it builds a working app that saves its data and comes back as a link you open in a browser. Lindy's home page labels Apps as new and frames them as simple internal tools built on the apps a team already uses. Take that word seriously. The docs describe apps as web apps made from a chat description, which suits trackers, lookup screens and team dashboards far better than anything resembling a product launch.

Every build draws on the workspace's shared credit pool. Lindy's pricing page gives ranges rather than exact costs: complex HTML dashboards and internal tools sit in the 1,000 to 2,500 credit band, and basic HTML artifacts in the 250 to 1,000 band. Iteration adds up. Lindy's docs say every message in a thread re-reads everything above it, so a long build conversation costs more credits with each turn, and Trustpilot complaints describe credits spent while iterating on a design. When the pool runs out, Lindy pauses credit-using work until the monthly reset or a top-up, as our [Lindy pricing guide](/articles/lindy-pricing) explains.


> 💡 **Write the full spec first:** Name the fields, the users and the source tool in your first message. Every later turn re-reads the whole thread, so fewer corrections cost fewer credits.


## Step two: grant each app access to the data it needs

An app is only as useful as the tools it can reach. Lindy Apps read from and update connected tools such as Salesforce or QuickBooks, and each app gets its own permission for each integration it uses [3]. Connecting a tool to Lindy does not hand every app access. The same rule covers saved API credentials, since an app needs its own permission before it can use one. That permission is the main control, because Lindy's stored credentials are write-only, private by default, shareable with the workspace and outside the reach of guardrails.

Shared connections deserve a second look. Owners or admins can share an integration connection with the workspace, members can see it but not change it, and a shared connection acts with the connected account's permissions. An app built on a shared CRM connection therefore works with that account's access. Lindy lists 3,000+ integrations plus MCP, so most sources will be reachable. Its per-integration guardrails (Always allow, Require approval, Do not offer) apply in shared Slack threads, not in web chat or Slack DMs. Separately from those guardrails, Lindy by default waits for approval before any action with outside impact, such as sending an email or updating a record.

## Step three: decide who can open it

Each app carries one of three visibility levels: only the creator, active workspace members, or public to anyone with the link, with no Lindy account needed. Workspace sharing is the natural choice for internal tools, but it has a billing side. Every person who joins the workspace, or who @mentions Lindy in Slack, becomes a billed seat, so the audience for a workspace app is your paying users. Public visitors need no Lindy account at all. They also see whatever the app shows to anyone who holds the link, which makes the public level a data decision first.

Offboarding takes one step. Removing someone from the workspace removes their access to every shared app at once. On compliance, Lindy says apps run under SOC 2 Type II, GDPR and PIPEDA, and HIPAA is the exception that changes the plan: apps that handle covered health data need the Enterprise plan with a signed BAA, which Plus, Pro and Max do not include. Lindy's docs also warn that its AI can make mistakes and that users should review actions before relying on them. That warning weighs more when an app writes back into a finance or CRM system.


> ⚠ **Public means anyone with the link:** A public app opens for anyone holding its link, with no Lindy account required. Check what data each screen shows before choosing that level.


## Step four: change the app without losing its records

Lindy keeps every app version. A manager can restore an earlier version or undo the last change, but saved data stays as it is and deleted data is not recovered. Read that twice. A restore rolls back how the app works, not what it holds, so a bad change that deleted rows leaves those rows gone even after the rollback. Ask for changes that touch stored records with that rule in mind, and let the connected tool remain the system of record rather than the app alone.

Since October 2026, apps can also react to events from other tools. Stripe and Slack webhooks are built in and verified as coming from that tool, and a custom webhook covers anything else. A payments team could, for example, have an app log Stripe events as they arrive, while a support lead routes Slack events into a shared queue. Those are illustrations, not templates Lindy ships. The verification is the useful part, because an app that acts on a webhook should know who sent it, and for Stripe and Slack Lindy runs that check itself.


> 💡 **Keep records in the source tool:** Restoring an app version does not recover deleted data. Let the connected CRM or ledger stay the system of record and treat the app as a view on it.


## When a hosted Artifact fits better than an App

Not every request needs an app. Artifacts, which Lindy also calls live pages and launched in September 2026, are dashboards, reports, prototypes and internal sites that Lindy builds and hosts on a lindy.link address [4]. A new artifact is private to signed-in workspace members, and outsiders see a private notice even with the exact URL. Making it public needs your approval first, after which anyone with the link opens it without signing in. There is no share-with-one-person mode. Unpublishing takes the link offline, and every change rebuilds on the same URL, so a shared link always shows the current version.

| | Lindy App | Artifact (live page) |
|---|---|---|
| What it is | Working web app that saves data | Hosted dashboard, report, prototype or site |
| Who can open it | Creator, workspace, or public link | Workspace by default; public after approval |
| Editing | Restorable versions; saved data stays as it is | Updates land on the same URL |
| Credit band on the pricing page | Internal tools: 1,000 to 2,500 | Basic HTML: 250 to 1,000; complex dashboards: 1,000 to 2,500 |

Lindy's marketing analyst role page shows the pattern for a recurring page. It describes a live dashboard per client that refreshes daily, with a link emailed to each client, plus a Monday 7:00 AM routine that builds and emails weekly reports, each of which also lives on a link clients can open anytime. For client-facing pages, that means the public setting, approved once, and a routine doing the refresh. Routines, memory and the rest of the setup are covered in our [Lindy AI playbook](/articles/lindy-ai-playbook).

## What Truemed's four tools say about scope

The clearest customer example comes from Truemed, a health and wellness payments company. Its case study says the Head of Support built four internal applications himself, covering support, payments, compliance and operations, without engineering time. That is the customer's account, not a typical result. It does show the intended builder: an operations lead who knows the process, not a developer. Lindy's October 2026 creative skills also added interactive dashboards, which widens what a non-technical requester can ask Lindy to produce.

Sales teams are the other obvious fit, because apps can read and update a CRM under their own permission. A pipeline view on top of Salesforce is a natural example, sitting beside the meeting follow-ups and phone work our [Lindy for sales teams](/articles/lindy-for-sales-teams) guide covers. Plan around one limit. Building and changing apps draws on the same credit pool, seats and approval defaults as everything else Lindy does, so heavy iteration on an app competes with inbox, meeting and research work for the month's credits.


## FAQ

**Can I build a Lindy app without using Slack?**

Yes. Lindy's changelog says an app can be requested in Slack or in Lindy's chat; Slack is needed for the shared team teammate in channels, not for the request itself.

**Is there a free way to try building with Lindy?**

A new Lindy account starts with $50 in free credits that last 7 days, with no credit card. There is no ongoing free plan; after that a paid plan is needed, from $29.99 per user a month (3,000 credits).

**What if the tool my app needs has no Lindy integration?**

Lindy's docs say it can connect to a hosted MCP server, drive the browser on Pro and Max, or write its own integration against the tool's API. An app still needs its own permission for each integration and saved API credential it uses.

**Is data inside Lindy apps used to train AI models?**

Lindy states that it does not sell customer data and that neither Lindy nor its model providers use customer data to train AI models.


## Should your team build its next internal tool in Lindy?

Build it in Lindy when the tool sits on data Lindy already reaches and the people using it are already in your workspace. The per-app permissions, three visibility levels and one-step offboarding are designed for exactly that case. Look elsewhere when the tool must hold records you cannot afford to lose to a bad change, since version restore leaves deleted data deleted. Keep HIPAA-covered work for Enterprise with a signed BAA. And write the request carefully, because every extra turn in a long thread costs more credits.



## Sources

[1] Lindy documentation: Apps — https://docs.lindy.ai/teammate/apps (2026-11)
[2] Lindy pricing — https://www.lindy.ai/pricing (2026-11)
[3] Lindy changelog (What is new) — https://www.lindy.ai/changelog (2026-11)
[4] Lindy documentation: Artifacts — https://docs.lindy.ai/teammate/artifacts (2026-11)
[5] Lindy documentation: Billing — https://docs.lindy.ai/teammate/lindy-teammate-billing
[6] Lindy Terms of Service — https://www.lindy.ai/terms-of-service
[7] Trustpilot: Lindy reviews — https://www.trustpilot.com/review/lindy.ai