Got a question? Start here. If yours is not on the list, email us and we will answer it.
A training platform for staff in residential, crisis, runaway and community youth services. Staff practice difficult conversations out loud with simulated young people. Every response is graded against the young person’s behavioral state at that moment, and the results build a competency record over time.
In three ways. First, a separate rules layer decides how the young person’s behavioral state changes. The language model only proposes, and the rules layer can overrule it. Second, difficulty follows a published rule. Three consecutive responses below standard lower it and three at or above standard raise it. A critical miss starts remediation in the same session.
Third, it is built to grade against your organization’s own rubric and protocol rather than against whether a model thought a reply sounded reasonable. Loading yours is part of upfront facility procedural integration: you send your documents during setup, KnightByrd loads them, and nothing is applied until your administrator approves it. Private document upload is built, with policy version control: every document is a dated version that its administrator attests is the current approved one. It opens for each organization once its pilot agreement is signed, and every document is approved before use. Until a rubric is loaded, scoring uses a placeholder rubric, and the scorecard says so.
The demonstration is rehearsed. It is five exchanges with Marcus, who is fifteen. His lines were written, voiced and lip-synced in advance, which is why his mouth forms the words.
The live simulation generates every reply on the spot from what the staff member actually says. HeyGen and Tavus both prohibit avatars that appear to be under eighteen. No vendor will lip-sync a synthetic minor in real time. So in a live session the mouth follows the audio and is visibly approximate.
Not yet. The engine is built, and the live simulation page shows what it does, but it is not open to visitors. It stays closed until every young person’s voice matches their age, gender and background, because a fifteen-year-old who does not sound fifteen undoes the rest of the session.
The rehearsed demonstration is open now.
No. All 15 personas are fictional, aged 5 to 17. Each has an authored background and language profile, and behaves according to the model for their developmental age band. No real record, real child’s likeness or diagnosis about a real person is used, and the setup forms collect no information about any young person.
Yes, as an optional developmental profile layered onto any persona for one session. There are 4: autistic (sensory overload), ADHD (impulsivity and attention), intellectual disability (processing and understanding) and complex trauma (hypervigilance). Each is written as behavior rather than a label, with rules the engine enforces, and each names the misreading it trains against.
All of them are drafts awaiting a named clinical, simulation and safety reviewer. Until then they run in practice mode only and produce no competency record. None is about a real child, and the intake asks which profiles your staff need to practise with, never who has them.
A way to make competency assessment checkable. The system makes its determination from the session’s evidence and seals it with a timestamp before a supervisor reviews the session. The supervisor’s score is then compared with it.
After at least 8 independent comparisons it reports how often an evaluator agrees with the evidence. It also shows which direction they differ in and flags possible leniency, severity and rubric drift. A score given after seeing the system’s determination is not counted. Recording supervisors’ scores inside the platform needs accounts and storage, which are not built yet.
No, and we do not claim that. Automated scoring can inherit bias from the rules and ratings behind it, including ours. What changes is the order: the evidence-based determination exists before anyone forms an impression, so the distance between the two can be measured. A flag is a statement about scores, never a judgment about a person.
Assessment Integrity is set up for 5 fields. Youth behavioral health, healthcare and nursing, child welfare, crisis response and juvenile justice. Only youth behavioral health is built. It has scenarios, personas, a behavioral model and a competency framework. The other 4 have none of those yet. Each would need them written or approved by people who do that work before it could be offered.
No. The curriculum is mapped to the INACSL Healthcare Simulation Standards of Best Practice and to SAMHSA’s trauma-informed and crisis care guidance. None of those bodies has approved, certified or endorsed it.
COA, CARF and the Joint Commission accredit organizations and programs, not software. What a platform can do is produce the records an organization puts in front of a surveyor.
Not yet, and we will not imply otherwise. Simulation-based training is well established in child welfare: Illinois trains new child protection investigators in simulation labs at three state universities, and the first of them has been independently evaluated since 2016. Whether this platform delivers comparable results is unproven. Finding out is the purpose of a design partnership.
Partly, and within limits. A language model proposes each turn’s grade against your rubric. It must quote the staff member’s own words or the grade is refused. The rules layer lowers any pass on a turn where the young person became more frightened and less trusting. The model never sets the young person’s state or the difficulty, and it never approves a scenario. Three named people do that: a clinical reviewer, a simulation reviewer and a safety reviewer.
Each voice is chosen for that persona, then checked by measuring its pitch against the range for their age. Pitch can be measured and background cannot, so background is judged by listening.
Marcus’s custom-designed voice measured too low for a fifteen-year-old, so he uses a stock voice chosen by audition that measures within his range.
Indicative pricing scales on three things: sites, users and usage. There is an annual platform fee of $2,400 for the first site, and less for each additional site. Each user costs $139 a year, falling to $79 at volume. 30 practice sessions per user per year are included and pooled across the organization.
For example, a 25-person home pays about $5,875 a year. A 60-person program pays about $10,540, and a six-site agency with 250 staff about $36,850. Upfront facility procedural integration (onboarding) is $2,500 for the first site and $750 for each additional site, once. Practice is never cut off mid-year, and state and regional government is quoted.
Nothing on the pricing page can be purchased yet. Design partners start with a paid pilot, 30 days of setup and then 90 days, priced from the published rate card. The pricing page shows a worked example.
Sometimes. Title IV-E reimburses allowable short-term training for child-welfare staff at 75%. That includes staff at private licensed agencies. The amount is reduced by the share of children served who are IV-E eligible. What qualifies depends on your state’s plan, so your finance lead has to confirm it. We will not assert it for you.
No. Completing a session is not the same as being competent. The platform records which competencies a member of staff has demonstrated, at what level, and across how many different situations and young people.
At the moment, nowhere. This deployment does not store accounts, sessions or organization settings, so a session’s record lasts only as long as the browser tab. Storage, accounts and role-based access are the next major piece of work. No information about any real young person is collected at any point.
A paid pilot: 30 days of setup, then 90 days of practice, priced from the published rate card. What we need most from you is which scenarios matter in your setting, and what a good response looks like on your floor. Your materials are used only for your organization, and separating each facility’s data is being hardened now.
Email us with a line about your program and what you want your staff to be better at. If you would like to look first, watch the rehearsed demonstration or read through a completed intake to see what setup asks.
Become a design partner. It starts with a paid 90-day pilot, priced from the rate card. What we need most is which scenarios matter in your setting and what a good response looks like on your floor.