Skip to content

Direct contractAutism support platform · 2025 · Lead engineerautiparent.com

AutiParent

Autism support platform with screening tools, milestone tracking and a moderated community, scoped tightly for a sensitive audience.

AutiParent is a direct Upwork contract: a focused build of an autism support platform on Webflow and Firebase, with screening tools, milestone tracking, a moderated community and an AI assistant scoped tightly to stay non-clinical for a sensitive audience.

Direct
Attribution
Contribution
Scope
5.0
Status

5.0 rating, public on Upwork, Apr–Jun 2025.

01

The problem

Parents arriving at AutiParent are often early in a difficult process, searching for answers a search engine gives badly and a forum gives inconsistently. The platform needed to offer real structure (screening tools, milestone tracking, a community) without ever presenting itself as a diagnostic or clinical authority it is not.

That constraint shapes everything downstream, including the AI assistant. An unscoped assistant answering health-adjacent questions for a vulnerable audience is a liability before it is a feature; the brief was to build one that helps without ever appearing to advise.

02

Constraints

The conditions the solution had to survive, not a wish list.

Non-clinical scope
Screening tools and the AI assistant support parents; neither can present as diagnosis or medical advice.
Sensitive audience
Parents of autistic children are an audience where a wrong or careless answer carries real weight, not just a bad user experience.
Community moderation
An open community around a sensitive topic needs moderation designed in from the start, not bolted on after a problem post.
Webflow front end
Marketing and content pages live in Webflow, with Firebase and OpenAI powering the parts Webflow cannot.
03

The approach

In plain English, before the architecture.

The AI assistant was scoped narrowly before it was built: what it can answer, what it must decline, and where it hands a parent to real resources or a professional. That boundary was a product decision made before any prompt was written, not a guardrail added after the fact.

Screening tools and milestone trackers were built as structured data collection with clear, plain-language framing throughout, and the community layer shipped with moderation tooling live from day one rather than as a later addition.

04

Architecture

The decisions that mattered, and why they were made that way.

  1. 01

    Scoped AI assistant

    Answers held to a defined, non-clinical boundary, with out-of-scope questions routed to real resources rather than guessed at.

  2. 02

    Screening and milestone tools

    Structured questionnaires and progress tracking built on Firebase, designed for plain-language clarity over clinical terminology.

  3. 03

    Moderated community

    Community features shipped with moderation tooling from launch, appropriate to a sensitive-topic audience.

  4. 04

    Webflow and Firebase integration

    Webflow handles marketing and content; Firebase and OpenAI power the tools and assistant Webflow alone cannot.

05

Outcome

Delivered as a fixed-scope engagement, with the sensitive-audience boundaries treated as core scope from the first conversation rather than a late addition.

Closed at a 5.0 rating.

Stack

  • Webflow
  • Firebase
  • OpenAI

Timeline

April to June 2025, fixed scope

More work

  • Lainey AI

    Multi-agent AI SaaS

    500 hrs

  • BlackAlgo

    Algorithmic trading platform

    Full build

  • Avana

    Multi-tenant booking SaaS

    Full build

Tell us what you can’t staff.

Twenty minutes. No deck, no discovery sequence. If we are the wrong fit we will say so on the call and point you somewhere better.