MLDSAI Inc.
MLDSAI Inc. builds OpenAdapt, an AI automation company and platform for building and maintaining repeatable workflows across existing business systems. OpenAdapt turns demonstrated GUI work into inspectable, locally executable programs, verifies the intended effect independently, and halts for review when it cannot prove the result.
At a Glance
- Operations teams
- Healthcare and cardiology administration
- BPOs and service teams
- Document-heavy business operations
- +2 more
AI Tools by MLDSAI Inc.
(1)
OpenAdapt
GUI Demo to Automation Compiler
Discussions
No discussions yet
Be the first to start a discussion about MLDSAI Inc.
Latest News
OpenAdapt corrected its MockMed benchmark claim and announced checksum-based verification of benchmark claims; the retained baseline recorded 13 model calls per run rather than 24.
OpenAdapt v1.16.0 released with lifecycle derivation from signed admissions and explicit inactive-admission state.
OpenAdapt v1.15.1 released with governed-execution quickstart documentation and refreshed platform BOM.
OpenAdapt v1.15.0 released with browser quickstart in the base install, updated Flow manifest, and OpenEMR benchmark-result documentation.
Products & Services
MIT-licensed local installer and command-line interface for recording or demonstrating GUI work, compiling it into a workflow, replaying it, and producing run reports. Local use is free and requires no account for the tutorial.
Compiler and governed runtime that converts demonstrated GUI work into inspectable programs with identity, policy, postcondition, and system-of-record effect checks.
Local desktop application for demonstrating workflows, operating them locally, and reviewing evidence when human attention is needed.
Hosted control plane for approved workflow versions, run history, team access, managed browser execution, and customer-controlled or hosted execution boundaries.
Market Position
OpenAdapt positions itself between brittle scripted/browser automation and free-form computer-use agents: unlike a macro or selector-only automation, it compiles a demonstrated task into an inspectable, governed program with identity and effect checks; unlike an agent that re-interprets every run, its healthy path replays deterministically and halts when evidence changes. The company's own comparisons discuss AutoHotkey, Selenium, Playwright, Microsoft Power Automate, and computer-use agents as adjacent alternatives. Its differentiation is strongest for UI-only, cross-system, consequential workflows where independent verification and human review matter.
Leadership
Founders
Richard Abrich
AI architect and founder of OpenAdapt; he holds a Master of Applied Science from the University of Toronto. Public professional information also identifies him as OpenAdapt's Founder and CEO/CTO and as having worked as an AI architect with Tribe AI.
Executive Team
Richard Abrich
Founder; Founder and CEO/CTO of OpenAdapt.AI
AI architect and applied machine-learning engineer; Master of Applied Science from the University of Toronto. He leads OpenAdapt's software and business implementation work.
Founding Story
OpenAdapt was started to put AI automation into production for business workflows that repeat across browser, desktop, remote-application, and virtual-desktop interfaces. The initial vision was to let a person or local agent demonstrate a task once, compile it into a reusable program, and verify the real system-of-record effect rather than trusting a GUI success message; the project's public code dates to 2023.
Business Model
Revenue Model
Open-source local engine is free under the MIT license. Revenue comes from hosted software subscriptions, scoped enterprise deployments, paid workflow implementation, and separately scoped ongoing maintenance/support; third-party usage and hosting costs can be identified in an engagement proposal.
Pricing Tiers
MIT engine on the customer's machine; record, compile, and replay with no account.
OpenAdapt Cloud managed runner with dashboard/run history, API key, verification URL, and agent access.
Qualified workflow and environment; identity/effect checks, customer-controlled or dedicated runner, private verification URL, BAA, Citrix/VDI, volume, and SLA options.
Target Markets
- Operations teams
- Healthcare and cardiology administration
- BPOs and service teams
- Document-heavy business operations
- Software and engineering teams
- Enterprises with regulated or consequential workflows requiring auditability
- Healthcare administration, including EMR evidence collection and RVU/revenue audits
- Document-heavy intake, field validation, reconciliation, and approved updates to business systems
- BPO and service-team work across client portals and desktop applications
- Browser, native desktop, RDP, Citrix, and VDI workflows that APIs cannot reach
- Agent-invoked last-mile execution where the workflow must be governed and independently verified
- Software and engineering teams building reusable workflow execution into an existing product or agent
- A US cardiology electrophysiology practice led by Dr. Victor Abrich, using OpenAdapt for a monthly Cerner PowerChart RVU audit; the case study reports about $75,000 per year in missed billables recovered and several hours of physician audit work saved monthly.