Everything live in the platform today.

• Live AI role-play + the Coach: a real-time spoken AI counterpart with an on-screen avatar that reacts to how you actually handle the moment, moving through tenant-defined backdrops (a lobby, a meeting room, a quiet corner) with cinematic dip-to-black scene cuts as the conversation shifts, and a Coach beside you: a per-turn signal light, a session-tone rail, and guidance grounded in your organization's authored content.
• Mastery + focused drills: per-concept proficiency that persists across sessions and raises the AI counterpart's difficulty as you improve (a first practice is gentle; once you are proficient it presses harder and voices realistic objections), plus 'Practice this' drills that zero in on one skill and debrief how you did on it.
• The learner experience: a coach-guided home that greets you, shows where you stand, and recommends what to practice; a Topic Hub for the Learn, Practice, Master loop; a unified end-of-session debrief with the full turn-by-turn transcript one tap away; and a text 'Ask the Coach' chatbot between sessions.
• Lessons + guided onboarding: the teach side of the loop: each concept opens with a coach video and a short lesson before you practice it, a video can stand alone as the lesson, and a new learner is walked through a Welcome and a platform tour that unlock the rest, with the coach always pointing at the next step. Competency is earned by doing it in the simulator, never by being forced to watch a video to the end.
• Studio + References: ingest your own source material and have the AI draft the standards, dimensions, concepts, and coaching (review and approve each before it goes live), arrange them in a deliberate order that reads cleanly and seeds the learning sequence (fundamentals before advanced), plus a department-scoped references library of loose-leaf facts the engine treats as current truth.
• Branded, stood up from your own material: a guided wizard stands up your space from your source material with no code or deploy (the AI drafts the curriculum and the AI counterpart), wearing your brand: logo (SVG or image), wordmark, and full color theme are edited on the Tenant screen with a live preview and applied at first paint, on your own subdomain, with your content kept fully isolated.
• Portable modules: reusable, opt-in skill packs (e.g. DiSC communication styles) any tenant can switch on, where enabling one merges its dimensions, concepts, and coaching into that tenant's own corpus, side by side with their material.
• Practice modes: each skill can be practiced as a live role-play with the AI counterpart, or as a coach interview in a few flavors: a 'teach-back' that debriefs work you did beforehand and probes for specifics, a 'reflection' that walks you through something you were just shown (a video or a document), and an open 'discussion', with an AI first cut that reshapes the concept to fit the mode, plus a display-only 'how to prepare' brief on the lesson.
• Per-learner data & competency gates: a module can define the data it needs about each learner (e.g. their DiSC style + profile), fed from an outside assessment, and use it to gate progress: requiring it before a learner can reach a 'mastery' level, or keeping certain concepts locked until it is present.
• Scenarios & personas, separated and reusable: the situation (opening, goal, start screen) is now its own editable thing, separate from who the AI counterpart is, so one persona can appear in several scenarios, plus an AI 'first cut' that drafts a full persona (character, arc, voice) from a one-line description. Export a tenant's content and cast as files and import them back (additive or full replace, parent links + scope preserved and every import re-checked against your standards before it goes live), and a chat assistant can even author a loadable cast file you drop straight in.
• Leveled competency ladder: tag each concept Core, Intermediate, or Advanced and the Topic Hub orders fundamentals before advanced, showing the tier on each concept for both the author and the learner. We deliberately left off a live 'how far up the ladder' meter: in free exploration a learner's mastery is a genuine spread across levels, so a single bar would mislead.
• A full guided track: an opt-in mode a tenant switches on, where the coach walks a learner up the ladder one rung at a time, suggesting the next skill they have not mastered with honest 'you are close, one more strong rep' encouragement. It recommends, never gates: every concept stays open to practice in any order.
• Source library + curation: every source ingested into a tenant's corpus is listed (name, type, size, when it was added) and any one can be removed, so re-generating a rubric grounds in exactly the material you want.
• A four-door practice launcher, with Bring Your Own: the home Simulator is a launcher with four ways in: Coach's Pick, Free Practice, Ask Coach, and Bring Your Own. In Bring Your Own you describe a real situation you are facing and the coach builds a role-play against the right kind of person, mapped to a skill so it still counts toward your mastery.
• Credit for overlapping skills, with honest pacing: demonstrate a skill and it counts wherever your authored content says it overlaps, not only on the drill you launched. And mastery now takes repeated strong performance across sessions: one good conversation develops a skill, but never mints mastery of it on its own.
• Ask Coach grounded in your modules: the between-sessions Ask Coach answers from an enabled module's material too, not just the tenant's own, so a question about a module skill (say a DiSC style) is answered from your authored coaching instead of generic advice.
• DiSC assessment: take the official DiSC assessment as a required, gated step, with your personal profile report retrieved, stored, and available to view anytime; configured per tenant by an admin (webhook + course code) and fed into the per-learner data that can gate progress.
• Context-aware coaching: the coach names a skill's framework (say a price-objection technique) only once the learner has actually met it; if a scenario surfaces that moment early, the coach gives plain in-the-moment help, never penalizes the gap, and the debrief afterward offers the skill as a next step, so a learner is never told to use a technique they have not been taught yet.
• Read the room: a people-reading drill where the colleague's communication style is hidden: you converse, adapt, and make your read, and the debrief reveals who they actually were, how well you read them, and offers the matching adapt-to-style skills so you can go deeper with that same colleague.
• Coach figure selection: each tenant chooses its on-screen coach: a persona with its own avatar, voice, and personality, reserved so no role-play counterpart ever reuses the coach's face.
• Passwordless sign-in: operators and learners sign in with a one-time magic link, no password. It is invite-only, so only people you have provisioned can get in, and the verified email becomes the learner's identity. This is the first piece of full single sign-on.
• DS-Checkpoint (external assessment): our first Volution Core channel: a public, phone-friendly prompt behind a QR code or short link: someone scans, writes their response, and it is scored on the spot against the same authored rubric that trains your team, so an in-person class can measure at the beginning, middle, and end and compare the rounds (high, median, low per checkpoint). Checkpoints group into a series you can duplicate in a click and drag to re-order, grading runs on a model you choose with a cost estimate shown up front, results export for the facilitator, and a demo test mode reveals the scoring machinery to whoever submits. Checkpoints can also be bundled, so one QR code walks a participant through several prompts in sequence ('1 of 2').
• Launch from your LMS (SCORM): a Cornerstone-style LMS launches DriveState as a standard SCORM 1.2 course and receives the learner's proficiency back on its own transcript, updating automatically as they practice (In Progress until the pass mark, then Passed) with no buttons to remember. Learners arrive already signed in under the LMS's own user id; where the assessment needs a real email (DiSC), the platform can look it up automatically from the client's roster via a per-tenant webhook, or simply ask the learner once. Per-tenant opt-in, tested end to end against a mock LMS and ready for a Cornerstone sandbox.
• Interviews that remember: leave an interview unfinished and the next run continues the conversation instead of starting over: the coach greets you as returning, builds on what you already shared, and focuses on what is still missing. The coach also will not wrap up while any required item remains uncovered, yet it reads the room: if a learner clearly is not engaging, it closes gracefully rather than dragging the conversation on. Scoring reflects meeting the interview's requirements, so one genuinely complete conversation can earn full marks.
• Audio check + self-healing sessions: a pre-session Audio check tests your microphone AND your speakers, with the speaker test voiced by your own coach: pick a device, watch the meter, and get a clear verdict, including warnings about background noise or speaker echo before you start. It remembers the capture recipe that works for your hardware. In the session itself the platform watches the audio chain live: a dead microphone is detected and repaired automatically where possible, echo and noise problems surface as gentle advisories, and if the browser blocks the avatar's sound a one-tap fix appears right on the video.
• A coach that reads the conversation live: open the coach panel mid-session and the first thing you see is a fresh read of THIS conversation: what is happening right now and one concrete adjustment for your next turn, grounded in your organization's authored coaching. The full skill card sits one tap deeper, and the scenario brief stays re-readable in the panel so you never lose track of the goal.
• Ask Coach as a standalone page: a tenant can mint a public Ask Coach link (another Volution Core channel): anyone with the link gets coached answers grounded in that tenant's authored material, phone-friendly and rate-limited, with no sign-in and no app access needed.
• Feedback, always one tap away: a Feedback control rides on every screen, sessions included: general feedback or a something-went-wrong report goes straight to the team with the screen you were on attached, and your email is included only when you check the box. An admin inbox triages every report through a simple status workflow, and your own submissions and their status are visible in your profile.
• Weekly team practice reports: a printable Team Practice Report for client admins: who practiced and how much, each learner's current proficiency and focus, a team skill map with collective coaching opportunities, DiSC completion, and week-over-week movement including newly proficient skills called out by name. Generated automatically every Monday over a consistent week so the comparisons are honest. Individual practice conversations stay private, always: reports show outcomes only.
• Coach video recorder: record a coach video by driving the tenant's coach avatar to read an authored script and capturing the clip, with upload progress, so a concept's lesson video matches the live coach exactly. No studio, no film crew.
• Scenario packs in modules: a module now carries ready-to-run Free Practice scenarios alongside its dimensions, concepts, coaching, and personas: switch the module on and its authored situations light up Free Practice with no per-tenant authoring, and the tenant's own scenarios always sit alongside and win any conflict. The DiSC pack ships five universal workplace dramas, one per style posture, so editing the pack once updates every tenant that runs it.
• Session movement, shown honestly: the end-of-session debrief and the home recap now show how much each skill actually MOVED this session ('+6 pts this session') next to its overall standing, so one conversation's contribution is never confused with the cumulative score.

Actively being built or refined right now.

• Ask Coach → training opportunities: Ask Coach answers in text and also surfaces the specific Topic Library lessons and drills that address what you asked.
• Module-aware generation: when a tenant has a module like DiSC switched on, content generation weaves the module's styles into the tenant's own concepts + coaching (e.g. 'with a fast, results-driven counterpart, lead with the bottom line') and skips re-drafting any dimension the module already delivers, so the two genuinely overlap instead of sitting side by side or doubling up.
• Mobile practice layout: on a phone the AI counterpart fills the width and the coach videos scale up as large as they fit without hiding the coach light, so the opening coach moment is not a tiny thumbnail.

Designed and queued; building soon.

• Progression tiers: a passing badge and a 'mastery' tier above green that recognizes sustained excellence.
• Verified identity on assessment checkpoints: an optional magic-link step on a public checkpoint: the participant taps the link in their email before submitting, so the response carries a verified address instead of a typed one. That makes cohort reporting trustworthy, e.g. comparing how people who train on the platform score against those who do not, using the same checkpoint.
• Facilitator tools: instructor-facing channels, starting with action-plan improvement for team leads.
• Consistent debrief everywhere: the last-session recap on the home page now shares the end-of-session debrief's layout, per-skill results, and session movement; the remaining gap is the conversation itself: storing both sides of the transcript so the recap can offer the same Review Turns and download that the live debrief has.
• Client admin access (scoped roles): client admins sign in to their own tenant's back end with roles scoped to just their workspace: manage their roster, open their team's practice reports directly instead of receiving them, and tune their content, with no visibility into any other tenant. The prerequisite for self-serve everything.
• Full tenant onboarding: grow the new-tenant wizard beyond the corpus to cover scenarios, personas, department(s), and branding, so a tenant is stood up end to end in one guided flow.
• AI scenario mapping: as you build a scenario, the AI maps it to the skills it teaches and the prerequisites that should come first, then pressure-tests that it actually plays out as intended for a typical learner, so a sound, well-sequenced curriculum doesn't require a curriculum designer.
• Authoring flow polish: surface the reference-fact scan during source ingestion (rather than as a separate later step), so loose-leaf facts are flagged as the material comes in.
• xAPI / cmi5 launch: classic SCORM 1.2 launch already ships (see the live list); this adds the modern xAPI / cmi5 flavor some LMS pipelines run on (Cornerstone via Rustici SCORM Cloud, for example), so completion and scores report back through whichever standard your LMS actually speaks.
• Flexible learner identity: sign in through your corporate single sign-on, or be identified automatically by your LMS at launch (the SCORM/xAPI handshake), with no email required. A data-minimal mode keeps the LMS or identity provider as the system of record and stores no learner email at all, so a learner's personal data stays where your organization already manages it. (The first piece, passwordless magic-link sign-in, already ships; corporate SSO and the LMS handshake are the open part.)

On our minds; shape still being decided.

• Tone-aware scoring: grading delivery and tone, not just what was said. (The live session-tone rail already ships; grading delivery vs content is the open part.)
• More channels for the Volution Core: DS-Checkpoint and the standalone Ask Coach page proved the channel model; next, let an external process “call” the Volution Core directly, so other tools (a facilitator's app, a call-center lookup, a website chatbot) can tap the same coaching brain. (Facilitator-requested.)
• Deeper-thinking responses: optionally routing Ask Coach, the drill debrief, and similar moments through a slower reasoning model when depth matters more than speed, balancing response time against answer quality.
• Verified learner identity: making sure the person earning the credit is the registered learner, not a paid stand-in (a known problem in this kind of training that ordinary 2FA can't solve). Because DriveState is already a live spoken role-play, the natural fit is a supervised first-time enrollment plus passive, continuous voice/face verification during the session, with liveness checks that also raise the bar against an AI standing in for the learner.
• Auto-built learning content: once the corpus is loaded the engine has what it needs to draft the rest: propose a starter set of personas + scenarios that exercise the concepts and feed dimension mastery, edited from one combined scenario/persona screen where the author tweaks the seeds and adds scenes + branding. Extends AI scenario mapping from analysis into generation. (Persona generation already ships; full scenario auto-build from the corpus is the open part.)
• Self-healing content: when a learner reports a coaching error, the engine searches the source + reference material for support and proposes a correction for human approval (e.g. the coach warned against a claim the latest source material now actually supports). Speeds the human-in-the-loop fix; full automation stays gated. (Auto-repair during authoring already ships; post-deployment error detection and learner-driven correction is the open part.)
• AI co-authoring in the Studio: when you add a dimension or concept by hand, the AI drafts the description, asks a clarifying question or two, then refines your wording into an approvable version, and you keep either your text or the improved one. (The AI-draft-on-add first pass already ships; interactive multi-turn refinement is the open part.)
• Multi-lingual platform: deliver the whole experience in other languages: the live coach and the AI counterpart converse in the learner's language, and the lessons, coaching, and interface localize with them. The live models are already multi-lingual, so the open work is localizing the authored content and re-rendering the coach videos per language.
• Coach video generation: produce a coach video's spoken track from a script with synthetic voices, so a clip can be authored without recording a live take (the script-driven recorder is already in testing). It is also the building block for re-rendering coach videos in other languages for the multi-lingual platform above.
• Hiring sandbox (candidate screening): a stripped-down mode a hiring manager hands a candidate to gauge how well they already handle a sample interaction before any training. The live role-play runs, but the coach and the on-screen measurements are hidden from the candidate; the results flow to the manager instead, who can review them and compare candidates side by side. Built on the same corpora and grading as the training engine, so a screen and the course it feeds into stay consistent. (The written-response version of this already ships as DS-Checkpoint; this extends it to the live role-play.)
• Cross-talk handling: the live conversation is interruptible by design, which is great, but other people talking nearby can cut it off. The device, echo, and noise side already ships (the Audio check and in-session watchdogs in the live list); removing other people's actual speech, not just noise, is the harder open part and needs background-voice cancellation: Krisp offers a browser SDK that does it, with the open-source RNNoise as a noise-only fallback. A noise-canceling headset remains the practical recommendation meanwhile.
• More lesson types: the coach video already counts as a first-class lesson type; add others over time (interactive walkthrough, reading, quick check) so each concept's Learn beat is authored in the format that teaches it best.
• Product knowledge as a first-class entity: the things being discussed (products, policies, plans, specs) become a structured entity the engine reasons over and can grade against, so the counterpart asks informed questions and the coach can check accuracy of the facts, not just the skill of the delivery.
• Flip the script: a reversed-role mode where you play the counterpart to feel a great interaction from the other side (a new mode in the practice-mode system that already ships).
• Adaptive coaching strictness: the coach's own grading bar rising as you improve. The counterpart already scales its difficulty to your level; this is the coaching side, pairing with the session-based mastery pacing so the standard keeps climbing as you get better.
• Reusable gatekeeper step: turn the Welcome onboarding gate into a setting any dimension can use: a must-finish-first step that then steps out of the way and no longer counts toward your proficiency, optionally keeping selected dimensions (or all of them) locked until it is done. Today the onboarding module delivers this for the one case that needs it; we will generalize it once a second gate, for instance a required assessment before certain skills open, makes the shape clear.
• One QR code per checkpoint series, forever: today a recurring class duplicates its series per run, which keeps cohorts perfectly clean but mints new QR codes each time. Evaluate keeping ONE printed QR in the slide deck permanently and segmenting results by run instead: by date window (the easiest), or by an indicator like a location or training-locator code. Combined with the roster integration, reporting could then also break out by facilitator or facilitator pairing.
• Switching departments, keeping every score: let a learner move between departments (a tenant setting) and hold a separate proficiency record in each. The foundation is already right: mastery is scored per concept and each department's concepts are their own, so records stay separate, nothing is overwritten by a switch, and returning to a department finds it exactly as you left it. Shared skills carry over automatically: module content (like DiSC) and company-wide concepts keep one record across departments, so what genuinely transfers is never re-earned. The open work is the learner-facing switcher, per-department rollups in the roster and reporting, and small calls like whether onboarding repeats per department.

Captured so they're not lost, but honestly, we may never need them.

• Font selection (white-label): let a tenant's brand skin set the typeface, not just colors and logo. Parked because a font change ripples through every layout and needs licensed web-font handling, and the payoff is small until a customer's brand guide demands it.
• Department-level branding: cascade the brand skin below the tenant, a department overriding colors or logo the way content scopes do. Parked until a real customer has sub-brands distinct enough to need it; tenant-level branding covers everything we've seen so far.
• Cross-training across roles: let a learner practice a neighboring team's scenarios and be graded on that role's competencies, by having each scenario carry the skill set it should be scored against. Parked until a tenant has mature multi-team content and a customer asks for it.
• Module dimension shape: settle whether a module is one rolled-up competency or several dimensions. Today it defaults to one (DiSC = communication styles, with the four styles as concepts), which is plenty for now. Parked: it only bites once there are many modules of different shapes.
• Module dependencies: let an add-on module declare it builds on a base module (e.g. a DiSC-for-Management pack on the core DiSC styles), auto-enabling the base so a learner never gets the add-on without its foundation. Parked until the module catalog grows past one.
• Counterpart asides: the narrator voicing brief non-verbal beats between turns (the counterpart pauses, considers, warms up). Parked: a realism nicety with uncertain feasibility and low payoff against the learning-path work.
• Large-file ingestion: today a deck is read in the browser (text only, media stripped), which already handles 300-700 MB+ decks; for files past the browser's memory ceiling, move extraction server-side / streamed so even very large decks ingest reliably rather than needing the author to split them.