What is an AI instructional designer?

What an AI instructional designer is

An AI instructional designer is an AI system that does the design and production work a human instructional designer would do. It takes a training request, works out what people need to be able to do, proposes objectives and a structure, builds the training, and revises it when the business changes. Your team directs the work and approves it. The system does the hours in between.

The phrase gets used for two different things, and the difference matters when you are buying.

The first is an AI feature inside an authoring tool. You still open the tool and build the course yourself. The AI drafts a paragraph, writes quiz questions or generates an image while you work. It makes a person who is already operating software a little faster.

The second is an AI coworker. You brief it the way you would brief a designer on your team, in conversation, and it does the design and production from start to finish. Nobody learns a new interface. Subject-matter experts share what they know and the AI turns it into training.

This guide is about the second kind. Katie, the product behind this site, is one. The Katie compared page sets the two kinds side by side in a feature matrix, and sets Katie against hiring a designer or an agency.

What an AI instructional designer does

The job is the same one a designer on your team does, so it helps to walk through it in order.

It clarifies the request

Most training requests arrive vague. A department head wants "something on the new pricing rules" and sends a deck. An AI instructional designer starts there. It helps define the outcome, the audience and the key behaviours, then writes the strategy and a training plan for your approval before anything is built. If you only have a business change or a performance gap, that is enough to start.

It gathers knowledge from your experts

Subject-matter experts provide the knowledge, answer the designer's questions and review drafts for accuracy. The AI runs that back-and-forth itself. For a human designer, chasing experts for answers is where the hours go.

It designs and builds the training

From an approved plan it produces the whole program: text, video, practice activities, simulations and assessments. In Katie's case that includes AI presenter video, short animated explainers, images, and software simulations built from screenshots or a screen recording. Alt text, captions and audio description are part of every training rather than an extra.

It revises in plain language

Your team and your experts give feedback the way they already talk at work, in Teams, Slack or WhatsApp. "The policy section is too dense for frontline teams. Split it into two parts and add a decision scenario." The AI makes the change and shows you the new version.

It publishes to your LMS

The finished training goes to your existing LMS as a SCORM package, or out as a direct link. Translations and localised versions follow the same route. Katie works in more than 20 languages with full right-to-left support, and local teams can review each adaptation before rollout.

It keeps the library current

One policy change can make several programs wrong at once. You tell the AI what changed. It finds the affected lessons, videos and practice activities, prepares the updates, and your team reviews and publishes. The same applies to older programs: you can assign it to refresh, translate or modernise any of them.

It teaches inside the training

This is the part a human designer never had time for. Instead of clicking through slides, learners can talk to the AI. It asks questions, role-plays realistic scenarios and gives feedback grounded only in the material your team approved. It also watches where learners make errors, ask questions or drop off, and recommends fixes for you to review and publish.

What it does not do

Vendors are less keen to list this part, so here is ours.

• It does not own the business context or the strategy. It needs your team for both. It can propose objectives and a structure. It cannot know what the business is trying to change this quarter unless someone tells it.
• It does not publish without you. With Katie, your team approves the objectives and structure before production, the complete training before publishing, and every later change. Learners never see something nobody on your team signed off.
• It does not set the quality bar. You do. It learns your terminology, business context and quality standards from each review and approval, so the bar it works to is yours, not a generic one.
• It does not need to be wired into your live systems to teach them. Katie builds system training from an existing user guide with screenshots, without integration with your live operational system. Your team removes sensitive data from the screenshots before uploading them.
• It does not take your content. You own your source materials and the final training programs. If you stop working with Katie, everything she built for you stays yours. Ann Education owns the technology, not your training.
• It does not skip your IT and privacy review. Before rollout, Trust Katie starts a security review covering your content, learner data, the AI providers involved and your security requirements.

How it differs from an AI authoring tool

L&D teams are being offered a lot of AI tools, and most of them are authoring tools with AI added. The difference is who operates the software.

An AI authoring tool still needs your team to spend hours in it. Someone lays out the screens, writes or edits the generated text, records or assembles the video, builds the quiz and exports the package. The AI shortens some of those steps. The job stays with your people, and so does the learning curve: somebody has to become the course builder.

An AI instructional designer takes on the work itself. You brief it in plain conversation, the way you would brief a colleague, and review what comes back. Your experts never see an authoring interface.

Pricing usually reveals which kind you are looking at. Authoring platforms often charge by creator seat, and learning management systems often charge by learner seat. Katie is priced around the workload she takes off your team, never per creator or learner. The comparison page puts the two side by side, row by row.

Will AI replace instructional designers?

This is the question people ask most, so it deserves a straight answer. Not in the sense the question usually means.

Instructional design is two kinds of work. One kind is judgement: understanding the business, deciding what the training should change, working with stakeholders, and deciding whether the result is good enough. The other kind is production: writing, building screens, making video, writing assessments, exporting packages and fixing every version when a rule changes. Backlogs are mostly made of the second kind: the design and production work that keeps requests waiting.

An AI instructional designer takes on the second kind. The first stays with your team. Katie, for example, is trained on expert instructional design principles and learning science and learns your standards from project to project. The training she builds is hard to tell apart from the work of a senior designer. She still relies on your human team for the business context, the strategy and the final approval. Katie is a capacity multiplier, not a replacement.

In practice the day of a designer who works with one changes shape. Less time operating authoring tools and chasing experts for answers. More time on strategy, stakeholders and the quality of what goes out. A request that would have waited in the backlog gets built, with the same person approving it.

One honest caveat. If your team's value to the organisation is mostly production hours, an AI instructional designer will change that balance, and the team will feel it. The question to settle with your designers before you buy is not whether the role disappears. It is what they want to do with the hours that come back.

How L&D teams use one

The requests below are the ones Katie's own demo walks through, and they are typical of what an L&D inbox looks like.

• A policy changed this morning. The approval limit moved. The AI updates the policy training, then reports that the old limit also appears in the manager onboarding and procurement courses and asks whether to update those too.
• The content is too dense for the people who need it. Split the policy section into two parts and add a decision scenario, so frontline teams practise the call instead of reading about it.
• A program is needed in another language. The same training in Spanish, adapted for Latin America or for Spain, with the local team reviewing before rollout.
• New hires start next week. From a slide deck about the new pricing rules, the AI drafts an outline and two practice scenarios for approval, then builds once they are signed off.
• An approved course needs a version for one department. The support policy stays the same; the IT version gets IT-specific scenarios and quiz questions, set for your approval.
• One system process has to be taught, quickly. From a user guide with screenshots, Katie creates an interactive walkthrough in about five minutes, with no integration project. Employees practise on screenshot-based screens, with an AI guide answering questions as they go, including in Hebrew. The system training page covers the details.

The homepage shows the whole cycle as one working morning: a messy request at eight, a defined need and an approved plan by half past, production while the team gets on with other work, feedback in chat, and a SCORM package or a direct link to the LMS before lunch. The point of the timeline is not the clock. It is that nobody on the team opened an authoring tool.

Training companies use the same capacity differently: to deliver more client projects with the team they have, working to their own methodology and brand standards, with each project reviewed against their quality bar before it reaches a client. The training companies page covers that.

How to evaluate an AI instructional designer

Ask every vendor the same eight questions, including us, and get the answers in writing. Then check them on the vendor's own pages. Everything this guide says about Katie is on this site's pricing, trust and product pages, and the comparison page puts the answers in a table.

1. Where are the approval points? You want approval before production, before publishing, and on every later change. If the answer is "you can review it afterwards", the learners are the reviewers.
2. Who owns the output? Your source materials and the finished programs should be yours, including after you leave. Ask what happens to the content if you stop paying.
3. Can IT and privacy review it before rollout? The review should cover your content, learner data, which AI providers are involved and your own security requirements. Katie's starts from the Trust Katie page.
4. Does it fit your LMS and your languages? SCORM export to the LMS you already have, or a direct link when you do not need one. The languages your workforce speaks, including right-to-left scripts, with a local reviewer in the loop.
5. How is it priced, and can you see the cost before the work? Per-seat pricing punishes you for involving more experts or more learners. Katie is priced on workload: one credit is one cent, and the exact cost shows inside Katie before each action. Ask what happens when the budget runs out. With Katie, you are warned first, paid work pauses at zero until you top up at the same rate, there are no overage bills, and unused credits roll over while the plan is active.
6. Does it learn your standards? Terminology, business context and quality bar should carry from one project to the next, so the tenth training needs less correction than the first.
7. Can you try it on real work first? A demo on sample content proves little. Katie's pilot is eight weeks with one team on one real training, from brief to publish, with success criteria agreed up front.
8. What does the learner get? A course to click through, or an instructor inside the training who answers questions, runs practice scenarios and only ever draws on approved material. Ask whether the vendor uses learner activity to recommend improvements, and who approves those.

Where Katie fits

Katie is an AI instructional designer built for L&D teams by Ann Education. She is the coworker kind, not the authoring-tool kind: you brief her in conversation, she does the design and production, and your team approves the work at every stage. She works in more than 20 languages, publishes through SCORM or a direct link, and teaches inside the training she builds.

She is hired through a monthly workload budget rather than seats. The listed plan is $1,000 a month for 100,000 credits, and yearly billing keeps the monthly price and adds 20% more credits. On the pricing page, that budget covers a worked month of 20 trainings, 40 minutes of presenter video, 10 translations, 2 software simulations and 500 learners completing a two-chapter training. The pricing page also has an estimator that compares a month of your workload today with the same month with Katie, with every assumption editable. A pilot program and an Enterprise plan with custom capacity, multiple workspaces and SSO are scoped on a call.

Where to look next:

• Katie compared: feature matrices against authoring tools and against hiring.
• Portfolio: the learning formats Katie builds, with a guided walkthrough on request.
• Pricing: the plan, the rate card and the workload estimator.
• Trust Katie: approval, ownership and the security review.
• System training: interactive walkthroughs from a user guide, without an integration project.

If you would rather talk it through, a 30-minute scoping call looks at what is on your team's plate and whether Katie is the right hire.