Section Coach Learner Data

Overview

Section Coach keeps what learners build and can come back to, such as their use cases, artifacts and assessment results. It deletes the answers it only needed along the way.

This page lists, feature by feature, what Section Coach keeps from the text learners type into it, who can see it, and what is deleted. Here, your admins means everyone with organization-wide reporting access in Section HQ, such as admins and viewers.

Feature	What is kept	Who can see it
Superpower artifacts	The finished artifact only	The learner
Other exercises	Nothing	Nobody
Use cases	The learner's answers and everything generated from them	The learner, colleagues they share it with, your admins, the Reporting API and the Section MCP tools
AI Fitness assessment	Every answer, the email entered, and the result	The taker, your admins, managers for their own team, the Reporting API and the Section team running the assessment
Profile	The learner's description of their company and role	The learner; your admins can request these descriptions through the Section team

Across every feature, the AI requests that generate content (personalized exercises, grading feedback) are also kept in Section's engineering logs for 40 days for troubleshooting purposes. See Processing.

📘
Section does not train or fine-tune AI models on learner data. See Processing for how answers reach the model providers.

Superpower artifacts

Some superpowers end with an artifact the learner keeps, such as a set of project instructions, a scheduled prompt, or a custom AI assistant designed in the ChatGPT Plugin, Gemini Gem and Copilot agent superpowers. The learner builds it over several exercises.

What is kept: only the finished artifact, the fields the learner reviews on the last step.

What is deleted: the answers the learner gave in the earlier exercises, as soon as the artifact is finished. Until then, they stay saved so the learner can pick up where they left off. Starting a new artifact clears them.

What the learner can do: open the artifact again, edit it, download it as a PDF and create more.

Who can see it: only the learner. Artifacts are not shown to your admins, and are not in any HQ report, export, the Reporting API or the Section MCP tools.

Other exercises

Quizzes and practice exercises that aren't part of building an artifact, including prompt writing practice, don't save the learner's answers. Section Coach records which exercises the learner completed and their progress, not what they wrote, their score or the feedback they got.

Answers that get AI feedback pass through Section's LLM pipeline, so they are in its logs for 40 days. See Processing.

Use cases

A use case is what a learner builds in the use case builder: how they would apply AI to a real task in their job.

What is kept

The learner's answers: their company, their role, the objective they describe, the choices they pick, and any note they add.
What Section Coach generates from those answers: the prompt, the plan, the feedback and the insights.
The use case output, for example a skill file, with the learner's edits.
Its name, color, icon, tags, the library template it started from, and how it is shared.

What the learner can change

Learners can rename a use case, edit its prompt and output, change who it is shared with, and delete it. Their answers can't be edited; building it again creates a new use case.

A deleted use case disappears from the learner's library, sharing, search, the Section MCP tools, the HQ Use Cases export and the Reporting API, but is still listed in the Progress Dashboard's use case download. Its contents stay in Section's database.

Who can see it

Where	What they see
The learner	Everything above.
Colleagues the use case is shared with	Everything above. Learners share with specific colleagues, or by link with anyone at your company. People outside your company can't open it, and your admins get no special access.
Your admins, in the Progress Dashboard's use case download	The name, objective, goal tags, template, whether it was saved and when it was created, with the learner's name and email. Includes deleted use cases.
Your admins, in the HQ Use Cases export	The name, level and goal tags, template, whether it was saved and when it was created, with the learner's name, email and department.
The Reporting API	The name (or, if unnamed, the first 30 characters of the objective), type, template, whether it was saved and when it was created, with the learner's name, email and employee ID. JSON responses add the level and goal tags. No prompt, plan or output. The Coach report returns only a count of use cases per learner, when `include_use_cases` is requested.
The Section MCP tools	The name, objective, prompt, plan, choices and skill file, to the learner and to colleagues it is shared with.

AI Fitness assessment

What is kept

Every answer, including written answers. Written answers are scored by an AI model.
The segmentation answers, such as function, job level or region.
The result: the level, the overall score, the score for each dimension and the points for each answer.
When the assessment was started and submitted, and which assessment link was used.
The email address the taker entered on the assessment link, or the account email of a signed-in taker. Section Coach uses it to attach the results to the taker's account when they sign in with the same email, now or later.

Who can see it

Where	What they see
The taker	Their level, shown against the full set of levels, right after they finish. No account is needed.
Your admins, in HQ	Totals for your organization: how many people took the assessment, how many are proficient, and the breakdown by level and by department. Where these features are enabled, they also see each person's level in the Coaching page's team view, with their name and email, and in the Employee Directory export, with when they last took it. The Assessment Results export lists each taker's level and overall score with the same fields as the Reporting API.
Managers with Coaching access in HQ	The same team view and Employee Directory export, only for people in their own reporting line. They don't see overall scores or the Assessment Results export.
The Reporting API	Each taker's latest level and overall score and when they took it, with their email, name and employee ID.
Section	The Section team that runs your assessment can open individual responses, including each answer and its score, export them, and link a response to the right account. At your request, it can also remove responses from your reports, for example duplicate submissions, or correct the email on a response. A removed response is kept, but no longer appears in HQ or the Reporting API, and the taker's own result doesn't change.

Answers, segmentation answers and dimension scores are not shown in HQ or returned by the Reporting API, but are available to Section employees.

Anonymous assessments. When an assessment is set up as anonymous, takers still enter their email, and it is stored so results can be attached to their account. HQ then shows only totals, with any group of fewer than 5 people hidden, and never shows an anonymous result next to a name: the team view and the Employee Directory export leave it out. The Reporting API and the HQ Assessment Results export still return each taker's level and score, but leave out their name, email and employee ID. The Section team running the assessment can still see the email.

Profile

Learners can describe their company and their role and responsibilities in their profile, so exercises and recommended use cases use examples from their own work. If your organization has set a company description, it is filled in for them. Section Coach keeps these descriptions with the learner's name and email, and includes them in the AI requests that build that learner's exercises and use cases. Your admins see the learner's name and email in reports, and can request the company and role and responsibilities to be shared by the Section team.

Processing

AI models. To generate prompts, plans, feedback and artifacts, and to score written assessment answers, Section Coach sends the learner's answers to an AI model through OpenRouter. OpenRouter routes each request to the provider that hosts the model, for example Anthropic for Claude models, which Section Coach uses by default.

Engineering logs. Section's LLM pipeline logs each request in full, including the learner's answers and the model's response, so the engineering team can investigate problems. A daily job deletes these logs after 40 days. The one exception is a small number of individual requests that engineers star while debugging an issue; those are kept until an engineer unstars them. Only Section staff with log access can view these logs.

Search. The name, objective and tags of each use case are turned into an embedding, a numeric representation used for search, by an OpenAI embedding model through OpenRouter. The search index stores only the numeric representation. Search terms learners type are embedded the same way, and are stored with their text, without the learner's name, so repeat searches are faster.

Product analytics. Section Coach sends usage events to Section's product analytics. Some events include what learners enter: their name and email, the profile details they give when they sign up (company name, job title, job level, function, industry and company size), use case names and search terms, the reason picked when reporting an exercise, and the email and result of an AI Fitness assessment.