Frog
Synthetic Learner Panel
Started as a question: could AI expose instructional problems before real learners encounter them?
AI made training faster to produce. Someone still has to make sure it’s good.
Actyra helps organizations build better training, test it before learners see it, and understand what is already in their course library.
Not sure where to start? Tell us what you’re working on.
What we believe
Completion isn’t capability.™
Your LMS can tell you whether someone finished the training. It cannot tell you whether the training was designed to make them capable in the first place. That is a question about how the course was built, and unlike the first one it is answerable before a single learner sees it.
Why now
Publishing a course is easy now. Knowing whether it is well designed is not.
AI dramatically lowered the cost of producing training. It did not solve instructional quality. The hard part is still deciding what someone must be able to do, what they need to know first, where they need practice, what feedback they need, and what they must demonstrate before passing means anything.
Production is cheap now. The question stops being how do we make enough of itand becomes how do we know any of it is good. There are three versions of that question, and Actyra is built around them.
Build it
Actyra Course Factory™
Turn expertise and capability requirements into training designed around what people actually need to be able to do.
Test it
Learning Preflight™
Before a course reaches learners, find the instructional failures, gaps and shortcuts your normal review missed.
Understand it
Portfolio Intelligence
What exists, what still opens, what it collectively teaches, and what is worth keeping.
Test it
Stop using your learners as your QA department.
Software gets QA before release. Products get inspected before they ship. Training usually does not. Learning Preflight™ is QA for training: we test the course before rollout and give you a prioritized plan for fixing what we find.
What we check.
Objectives
Does the course actually teach what it says it teaches?
Practice
Do learners actually get to practice it, with feedback?
Assessment
Does passing mean anything?
Failure modes
Where can someone get lost, disengage, or game the course?
Plus three more: capability coverage, prerequisites and instruction quality. All seven, with the scoring →
Then six synthetic learners test it.
Cold Start, the Practitioner, the Mobile Learner, the Second-Language Reader, the Keyboard-Only User, each running the course three times. And one more.
One of the six
Minimum Effort
This learner wants to finish as quickly as possible. It skims, looks for shortcuts, and tests whether completion can be earned without demonstrating the intended capability.
If they can get credit without demonstrating the capability, the course is Not Ready for Deployment, regardless of every other score.
AI finds. Humans verify.
Synthetic learners surface potential failures. Every critical issue is checked against the actual course package by hand before it reaches your report. We do not forward a model’s opinion and call it a finding.
$1,500 per course, up to 3 hours. Ten business days. Retest included.
And if we cannot find a single critical instructional issue, you don’t pay. Not a partial refund, no invoice, and you keep the full report either way.
Preflight is a structured review of instructional design, not a measurement of learning outcomes. It finds design defects before deployment. It does not predict how learners will perform on the job, and it is not an accessibility conformance audit, which is why the verdict is Cleared for Human Pilot rather than “proven”. See exactly what it does and doesn’t do.
Build it
Build your course, and the system for building the next one.
Most experts do not have a content problem. They already know far more than belongs in a course. Actyra Course Factory™ helps you turn that expertise into a complete course architecture, build the course with AI, test it with Learning Preflight™, and leave with a process you can run again.
Extract
Structured prompts surface the explanations, examples and judgment calls you use without thinking about them.
Architect
Capabilities, KSATs, objectives, sequence, practice and assessment. AI can help with every step; it shouldn't be making the final instructional decisions.
Build
Lessons, examples, practice, feedback and assessments, produced in CourseTemplates.ai.
Preflight
Six synthetic learners, eighteen runs. Fix what is weak, retest, clear it for human pilot.
You keep the production system.
Along with the finished course, you keep the prompts, templates, rubrics and workflow used to build it, so you can run the same process on the next one.
$2,500 · all the way to human pilot · coming soonOne finished course, a full Preflight ($1,500 on its own), the remediation cycle and retest, and the Factory system for every course after this.Course Factory and the waiting list
Understand it
The course still works. Until you need to change it.
It launches. It tracks. The LMS reports completion. Then someone asks for an update, and nobody can edit it anymore. Source files disappear, authoring tools change, developers leave. A small update becomes a rebuild, and a course you already paid for gets paid for again.
73
real course packages measured
46.6%
had no editable source
15.1%
already had a dead runtime
Every one of those packages still tracked correctly. The LMS saw healthy courses while the organization could already have been sitting on stranded assets, which is exactly why the dashboard never raised it.
Portfolio Intelligence tells you what you have, what still opens, what the library collectively teaches, and what is worth keeping. The result is not another spreadsheet of course titles. It is a decision map: keep, recover, rebuild, retire.
Seven public-sector publishers, pre-registered, 46.6% with no editable source.
Proof
We publish selected runs, in full.
Complete Preflights against public-sector courses we did not author and were not engaged to review, including a legally mandated compliance course where all fifteen answer keys are readable in the page source and completion means a browser reached the last page.
These are a selection, not everything we run. Client work stays private. What we do publish is published whole: the findings, the evidence behind them, the verdict and the remediation plan.
From the Estuary
Where we test ideas before deciding what they should become.
AI role-play, synthetic learners, learning systems for AI agents, and other experiments that do not fit neatly into a product yet.
Who builds this
Built on decades of learning technology experience.
Actyra combines instructional design, AI-assisted course development, and deep expertise in the standards behind modern eLearning.
Brian Caudill
Founder
Brian was in the working groups when SCORM was being written, and holds one of only nine SCORM auditor certifications in the world. He built and sold JCA Solutions to Rustici Software, the company whose engine runs SCORM content for much of the industry. Twenty-plus years at the intersection of eLearning standards and software engineering (SCORM, xAPI, AICC and cmi5), building and repairing content systems across defense, enterprise and higher education. He co-wrote the first practical guide to SCORM development for instructional designers, and is a multiple Army dL Maverick Award author.
Brian’s track record spans more than two decades. Learning Preflight™ is new.
The testimonials below are from Brian’s SCORM consulting and eLearning work, the practice Preflight grew out of. We are building Preflight’s own track record separately, through published runs and the no-critical-issue, no-invoice guarantee.
Brian has done eLearning work for the U.S. Army, U.S. Navy, Northrop Grumman, Stanford, LexisNexis, FlightSafety, Truist, Infor, Penn Foster, Apollo Group, NREL and Fiat, among others, across his career and including work predating Actyra.
Build one. Test one. Or understand all of them.
Actyra helps organizations build, test and understand training around the capabilities that actually matter.
Not sure where to start? Tell us what you’re working on.