Screenshot Tutorial Companion
Diagnose an unfamiliar interface by turning a screenshot into guided instruction
- Difficulty
- Starter
- Time to result
- ~days to results
- Steps
- 5
- Confidence
- 94%
The Screenshot Tutorial Companion turns a static image of a software interface into contextual instruction. Capture the state where you are confused, upload it to a vision-capable learning assistant, and ask the tool to explain what the interface is doing. Add the intended outcome so it can compare the current configuration with the desired result, then request potential inefficiencies, best practices, and related concepts worth learning. Because the image supplies labels, branches, and layout that users may struggle to describe, the assistant can often reconstruct the workflow and produce more relevant guidance than a generic text query. The process should end with cautious validation inside the real product, especially when the interface controls automation, customer data, permissions, or other consequential operations.
Origin
Extracted from Marketing Against The Grain after a host uploaded a HubSpot workflows screenshot to Google Learn About and asked it to explain and critique the configuration.
Core principles
- 01Show the system the actual interface rather than describing it from memory.
- 02State both what you see and what you are trying to accomplish.
- 03Use visual interpretation to identify components, branches, and likely inefficiencies.
- 04Ask for explanation, diagnosis, advice, and next learning steps.
- 05Treat the response as guidance that must be checked against the live product.
How to run it
- 1
Capture the stuck state
Take a screenshot that includes the relevant controls, branches, labels, and surrounding context. Exclude unrelated windows and sensitive data.
Pro tip Capture enough of the interface to show how the visible element connects to the larger workflow.
Watch out Screenshots may expose customer data, credentials, internal URLs, or other confidential information.
- 2
Ask for interpretation
Upload the screenshot and ask the assistant to explain what is happening. This establishes a shared description before requesting advice.
Pro tip Ask it to identify visible triggers, actions, branches, and dependencies.
Watch out Visual interpretation may misread small labels or hidden configuration.
- 3
Supply the intended outcome
Explain what you are trying to accomplish and ask whether the shown configuration is a good way to do it. The desired outcome gives the diagnosis a decision criterion.
Pro tip Describe expected inputs and outputs rather than merely saying the workflow is broken.
- 4
Request improvements and lessons
Ask for inefficiencies, best practices, alternative structures, and the concepts you should learn next. Use the answer as a targeted tutorial.
Pro tip Request explanations of unfamiliar terms before making changes.
Watch out Do not blindly apply recommendations that depend on unseen settings.
- 5
Validate safely
Compare the advice with official documentation and test the proposed change on a reversible or non-production example.
Pro tip Change one element at a time so you can identify which recommendation affected the result.
Watch out Automation changes can trigger unintended actions when tested against live records.
In the wild
A user uploads a screenshot of a HubSpot workflow and asks what is happening, whether it is a good implementation, and what other advice applies. The assistant identifies workflow triggers, actions, branches, linear versus non-linear structure, potential inefficiencies, and best practices before explaining the visible go-to action.
→ The user receives a contextual tutorial without first knowing the terminology needed to formulate multiple search queries.
Common mistakes
Uploading sensitive screenshots
A screenshot can disclose far more context than a typed question, so confidential details must be removed first.
Omitting the desired result
Without knowing the intended outcome, the assistant can describe the interface but cannot reliably judge whether the configuration is appropriate.
Treating visibility as completeness
The assistant cannot evaluate settings, data, or dependencies that are not visible in the captured image.
Is it for you?
Best for
It is best for unfamiliar software interfaces, workflow builders, dashboards, and configuration screens where visual context matters.
Not ideal for
It is not ideal for screenshots containing secrets or for changes whose consequences cannot be safely tested or reversed.
From the transcript
“Can you explain to me what's going on?”
“Is this a good way to use HubSpot workflows? What other tips or advice could you share?”
“just take a screenshot and use AI, in this case, learn about to basically tell you like, oh, this is what I'm doing. What am…”
From the episode
Will Google's AI Experiment Replace Search Engines?