Brett LindenbergMindstamp records what each viewer does in an interactive video, including their answers, clicks, path, and how much they watched, and it sends those records by webhook after each viewing session. You can send them to any webhook receiver that can call an AI model, such as your own service or a workflow platform. The receiver asks the model to classify, summarize, or rank each record and passes the result to the system your team works in. We first built this setup in 2024 with Contextual, our AI workflow partner at the time. That partnership is no longer active, and the steps below work with any receiver. The workflow runs after the session ends, so it shapes what your team does next and leaves what the viewer sees to the video. The short demo below starts with the first decision to make: what should the workflow help someone on your team do?
What Mindstamp sends to an AI workflow
Mindstamp sends three kinds of webhook, each as signed JSON. You turn each one on per video, and the field names match the Mindstamp API (see what each exported field means).
| Webhook | What it contains | Plans |
|---|---|---|
| View | The finished session: watch progress, scores, identity, location, variables, a list of interactions, and each question with its answer | Core and higher |
| Interaction | One answer, click, or response, with the prompt, the value, the correct answer, whether the answer was correct, and the time in the video | Pro and higher |
| Lead | The viewer record: name, email, phone, custom ID, variables, and totals. Mindstamp sends it once for each known viewer. | Pro and higher |
Mindstamp sends all three webhooks during background processing after the session is finished, which happens when the viewer reaches the end, closes the page, or stops for about two minutes. Interaction webhooks arrive together after the view, not one by one during playback, so allow for a delay and do not depend on the order. Your endpoint must reply with HTTP 200 or 204 within 15 seconds. Mindstamp records any other reply as a failure and has no scheduled retry, so log what arrives and use the API to fill gaps. Our help page on how Mindstamp webhooks work has the setup steps and a signature check you can copy.
How to send Mindstamp data to a webhook receiver
The receiver can be a small service on your own server, a serverless function, or a flow on a workflow platform. Whatever you choose, it needs four parts: a web address that accepts the webhook, storage for each record, a second step that calls the model, and a safe place for the credentials of the model and of the system that gets the result.
- In Mindstamp, open Integrations, select Mindstamp API, and copy Your Webhook Key.
- Create an endpoint that checks the X-Mindstamp-Signature header against your webhook key, saves the record, and replies with HTTP 200 right away.
- Back in Mindstamp, enter the endpoint's address as the default URL for each webhook type you need. Then open the video, select Integrations > Mindstamp API, and turn on Send View Webhooks, Send Interaction Webhooks, or Send Lead Webhooks. A default URL on its own sends nothing.
- Start the AI step from the saved record, for example with a queue or a database trigger. Give the model a fixed set of outputs, such as five categories or a yes-or-no flag, and write the result back to the record.
- Send the result to the place someone will act on it, such as a CRM task, a help desk ticket, or a team inbox.
- Watch the video in a private browser window, answer its questions, then close it or wait about two minutes and check the endpoint's logs.
Saving the record first keeps a slow model call from pushing the reply past 15 seconds. For a worked example with screenshots from our 2024 build, see our walkthrough of an interactive video and contextual AI workflow that sorts free-response answers by topic.
What an AI workflow can do with video answers
Each use below starts from data the video already collects. If the video does not ask for something, the workflow cannot use it, so plan the questions and the workflow together.
Route open-ended answers to the right team
A free-response question lets viewers type anything, which is useful and hard to sort by hand. The workflow reads each answer, assigns a category such as product question, support, or partnership, and creates a task for the team that owns it.
Write a session summary for the sales rep
The view webhook holds every answer, the path a prospect took, and how much they watched. An AI step can turn that into a three-line note: what the prospect asked about, which option they chose, and where they stopped. Mindstamp's direct HubSpot integration already writes contacts and timeline events on Pro and higher plans, so use the workflow for the summary. Salesforce integration is quoted per account; confirm its scope with Mindstamp.
Choose the follow-up resource
After the session, the workflow can match a viewer's answers to the next video, article, or offer and put it in the follow-up email. A choice the viewer should act on while watching belongs in the video itself, as a branch.
Rank leads for a person to review
Answers to discovery questions, such as team size or timeline, give a model something concrete to rank. Have the workflow sort leads into a review list with the reason for each rank, and let a salesperson decide who to call. The ranking arrives after the session ends, so it suits a call list, not a reply while the prospect is still watching.
Flag answers that need a person
In compliance or safety training, a learner's free-text answer can show a misunderstanding that a multiple-choice score misses. The workflow can flag answers that match the patterns you describe and send them to a supervisor. Keep the decision with that person.
Find patterns across many sessions
One answer at a time says little about the video. A weekly job can pull a batch of answers through the Mindstamp REST API, which is on Pro and higher plans, and ask a model to summarize what viewers misunderstood or asked about most. Use the summary to decide which scene to rewrite.
What to keep inside the video
Webhooks arrive after the session, so an outside workflow cannot change what a viewer sees while they watch. Build anything that must happen during playback in Mindstamp:
- Send a viewer to the right section from their answer with branching or conditional logic, on supported plans.
- Answer viewer questions inside the player with Genie AI, which answers from the video's transcript and the notes you add (see how to add an AI assistant to a video).
- Show a viewer's name or company from their link or an earlier answer with personalization variables.
Then point the workflow at what happens after the video: who to call, what to send, and what to review.
Zapier, your own code, or a workflow platform
The right option depends on how many steps the workflow has and who will maintain it.
| Option | How the data arrives | The AI step | Fits when |
|---|---|---|---|
| Zapier (Mindstamp Core and higher) | Zapier checks Mindstamp for new finished views, viewers, and interactions on its own schedule | AI by Zapier's Analyze and Return Data action can summarize, classify, extract, or generate data | One or two steps that someone on your team can build and maintain |
| Webhooks to your own service | Mindstamp posts signed JSON after each session | Your code calls the model you choose | You have developers and want full control of the code and the data |
| A workflow platform | Mindstamp posts signed JSON to a flow on the platform | The platform's model step, with the provider you choose | The workflow spans several systems and you want it built and hosted outside your own code |
If your workflow is one classification step, start with Zapier or a small script. If it spans several systems and needs a team to run it, use your own service or a workflow platform.
Before you send viewer answers to an AI model
- Send only the fields the workflow needs. Answers can contain personal data, so keep whole records out of chat messages and logs.
- Tell viewers how you use their answers, in the video or in your privacy notice.
- Keep a person in charge of decisions that affect the viewer, such as a hiring step or a compliance outcome.
- Check the signature on every request, and keep a log of what arrives so you can find gaps.
- Read how your workflow platform and the model provider store and use the data you send. Our page on how Mindstamp's AI features handle your data covers Mindstamp's own AI features only.
Common questions
Can an AI workflow change the video while someone is watching?
No. Mindstamp sends webhooks after the session is finished, so the workflow acts on the next step, such as an email, a task, or a call list. Use branching and conditional logic for anything the viewer should see during the video.
Which Mindstamp plan do I need?
View webhooks are on Core and higher plans. Interaction webhooks, lead webhooks, and the REST API are on Pro and higher plans. See Mindstamp pricing for the current plans.
What else can AI do with interactive video?
AI can also draft quiz questions, dub and translate a video, and answer viewer questions inside the player. Our overview of what AI can do for interactive video covers each one. If your concern is getting your videos noticed among AI-generated content, read how to make your videos stand out in the age of AI-assisted content.
Plan one workflow first
Write down one decision your team makes after someone watches a video, such as who to call first or which answers need review, and the data that decision needs. Check that the video asks for that data, turn on the matching webhook, and send a test session to the endpoint before you add any AI. To talk through a workflow, contact Mindstamp.
