Skip to content
Customer Portal
One record, two apps

Planned in the portal. Live on the board.

Aivota is one clinical record with two front doors - a portal where the team plans, authors and reviews, and a device the AAC user speaks through. This page follows the places where the two meet.

Book a demo

One loop, not two products

Most AAC is authored in one tool and used in another, and the two never speak to each other. Here a single record drives both - so the work a clinician does lands on the device, and what happens on the device comes back as evidence.

  1. The team plans

    Goals, boards, programs, calendar and permitted apps are all authored in the portal - mostly by talking to the assistant.

  2. The session runs

    The device carries that plan into the moment: the board rebuilds around what is happening, and the AI helper works toward the objectives the team set.

  3. The evidence returns

    Session notes, new vocabulary, tracked skills and progress data land back in the record - which is where the next plan starts.

The clinician portal

Where the work gets set up

Four places where what a clinician does in the portal changes what the AAC user meets on the device.

Talk to the portal instead of filling it in

Nearly every operation - creating an AAC user, editing a goal, authoring a board, scheduling an event, changing device settings - can be done by talking to the assistant, with the panels updating live in the same turn. The panels and the assistant are two views of the same data, under the same permissions.

The loop

Ask for a change here and it reaches the device on the AAC user's next session. There is no separate publish step to forget.

  • Specialized personas - clinical advisor, speech-language pathologist, occupational therapist, pediatric physical therapist, behavioral analyst, teacher - tune the assistant's expertise to the question being asked.
  • A curated reference library grounds its answers: conditions, clinical practice and approved medications, maintained centrally rather than recalled from a general-purpose model.
  • A small set of high-security operations stays deliberately behind explicit UI. The assistant can prepare them; a person completes them.

Author a board by describing it

Describe the board you want and the AI drafts it; refine it by dragging. Multi-page boards, composed glyphs and per-button behavior stay editable by hand whenever you want that control.

The loop

Attach a context hint and the board loads itself on the device when that context arrives - music class, a meal, a visitor - without anyone pressing anything.

  • Words are built as glyphs: a main symbol plus modifiers for colour, quantity, negation, tense and possession, so a board can say far more than its icons literally contain.
  • Symbols resolve from the canonical library, from the AAC user's own photos, or from AI-generated art held to a single house style.
  • Boards travel as packages between members of an organization - and reach a public catalogue only after human attestation and review.

Programs that reach into the session

Annual goals, measurable objectives, baselines, Goal Attainment Scaling with aggregate T-scores, ICF functional profiles, data collection with trend visualization, related services, meetings and transition planning - assembled as an IEP.

The loop

The objectives written here are injected into the live session's context, so the AI helper is working toward this term's goals rather than improvising its own.

  • Progress data accumulates from real sessions instead of end-of-term recall, and charts its own trend as it goes.
  • Medical, functional, educational and progress reports carry a draft, review and finalize workflow, version history and print-ready output.

A second pass over everything on record

Deep Analysis periodically re-reads the whole picture for one AAC user - sessions, notes, incidents, progress data - and surfaces the patterns a busy team never had time to look for.

The loop

It reads what the device produced across the whole period, then drafts the clinical report in the same pass for a clinician to review and sign.

  • Correlations across time: what tends to precede a good session, which contexts a skill actually appears in, what stopped working and when.
  • The report arrives drafted rather than blank, so the clinician's time goes to judgement instead of transcription.
  • Video analysis, on the same premium tier, turns uploaded footage into a timeline of relevant events and attention targets.
The AAC user's device

Where it is actually used

The same record, met from the other side - as a board that moves, a helper that knows what they are working on, and two ways to reach a word nobody put there.

A board that rebuilds around the moment

Conventional AAC offers a fixed grid somebody authored in advance. Here the board regenerates on each conversational beat - from what the camera sees, who is in the room, what was just said, what is on the calendar, and what this AAC user has said before.

The loop

Calendar events and context-hinted boards from the portal arrive on their own, so the board knows it is music class before anyone mentions it.

  • When the right word is not offered, a parts-of-speech sentence builder - laid out for eye gaze, no scrolling, stable target positions - composes it symbol by symbol.
  • Every press speaks in the AAC user's own chosen voice, distinct from the AI's, with an on-device fallback so a press is never silent.

An AI helper, not an AI friend

The helper has a professional personality: warm, patient, and plainly a working relationship. It remembers what this AAC user needs, what they are working on and what they have told it before, and it uses that to help them say things and make progress. It does not court attachment or offer itself as a companion for its own sake.

The loop

What it remembers is the clinical picture the team maintains, handed to it by a supervising agent at the start of a session - not a private history the device invented.

  • It reads the room: engaging the AAC user directly when they are alone with it, and stepping back to manage the board in silence when they are talking to a person.
  • The agents that see and hear the AAC user never hold clinical data, and the agent that reads the record cannot speak - situational awareness without exposing a diagnosis to whoever is in the room.

Social skills, at a difficulty you set

The Social Trainer generates an AI peer - its own face, voice, personality and memory - to practice conversation somewhere nothing is at stake. It is a training instrument rather than a playmate: the clinician chooses the focus skills and sets a challenge ceiling, and the peer plays to that level.

The loop

Nineteen tracked social skills feed back into the record, so the challenge can be raised deliberately as the AAC user improves instead of by guesswork.

  • Turn the difficulty down for an AAC user who needs an easy win, or up to rehearse the hard parts - interruption, disagreement, repairing a misunderstanding.
  • Afterwards the AI debriefs warmly with the AAC user about how it went, so reflecting on the attempt is part of the practice.
  • The peer is generated rather than real, so an attempt that goes badly costs nothing and can simply be run again.

Reaching a word that is on no board

When a concept is on no board and cannot be built, the Word Finder narrows in on it through a guided game of twenty questions. The AI asks the questions, but a deterministic engine on the device constrains the search, so it cannot wander off or talk itself into an answer.

The loop

A concept found this way is remembered and offered proactively in later sessions - and the vocabulary an AAC user reached for becomes something the team can see.

  • The model supplies the questions; the search space stays under the device's control. The AAC user is not at the mercy of an improvisation.
  • A word found once stops being hard: it surfaces on the board by itself the next time the context matches.
  • What an AAC user tried to say and could not is often the most useful thing a session produces.

Apps and games the AI plays along with

The device is more than a conversation. A clinician-curated apps page - calls, media, creative tools, a restricted browser and a set of games built for gaze - extends it beyond communication, and each AAC user gets only what their team assigned.

  • The AI helper knows what is on screen: it can open an app when the conversation calls for one, talk with the AAC user about what is happening inside it, and hand the board back when they are done.
  • Everything on the page is permitted per AAC user, down to which channels, playlists and sites are reachable.
  • All of it is built for the same access methods as the board, so nothing on the device needs a hand.

See it on a real caseload

The quickest way to judge whether the loop holds up is to watch a session, and then look at what landed in the record.

Book a demo