Build

Custom behavioral engineering and AI systems.

For organizations that want to own the measurement and decision systems at the core of their work — not rent them from a vendor.

Start a conversation

Point of view

Why this work matters

The measurement infrastructure behavior analysis depends on is being rebuilt right now, by people outside the field. The categories, defaults, and data-access rules that will define how behavior is measured for the next decade are being decided by vendors designing for venture returns, not by clinicians designing for behavior-analytic validity.

Organizations that notice that early do not want to rent the result.

The same pattern shows up in any field where vendors are rebuilding the measurement and decision layers faster than the operators of that domain can keep up. Behavior analysis is where I have written this out most fully, in the paper The field that taught machines to learn. The work itself generalizes.

Engagements

The shape of the work

Engagements are scoped per client — most run several weeks to several months, with advisory work typically shorter. What follows describes the shape of the work, not a product catalog.

Measurement systems

Build the continuous measurement apparatus a serious program actually depends on — cameras, microphones, edge compute, dashboards, data schemas. Starts from what your organization actually needs to measure and why, not from vendor defaults. In clinical contexts that means behavior-analytically valid response classes and treatment-integrity signals; in other domains, whatever the equivalent is for you.

AI integration into existing workflows

Make frontier models work inside the workflows your organization already runs — documentation, supervision review, report drafting, funder-specific billing, internal knowledge retrieval. Built so PHI stays where it belongs, so vendor lock-in does not compound silently, and so the system you end up with is inspectable rather than a black box you rent.

Operator-level AI strategy

For executives and founders deciding what to test, defer, build, buy, or ignore. Closer to advisory than to implementation. Engagements are short, substantive, and end with concrete direction rather than a deck.

Research and evaluation

Independent evaluation of vendor AI claims. Research collaboration on applied behavior-analytic measurement problems. Peer-reviewed publication where the work supports it.

Dave has scaled real therapy organizations and built the systems behind them. As CEO of Ausblick Therapie, he tripled locations and employees in a year. He then co-founded Knospe-Lerncenter and built it into Germany's largest ABA therapy provider in two years — 8 facilities, 80+ staff, ABA alongside speech, physical, and occupational therapy. He is also a researcher by training, with peer-reviewed work on decision making in the experimental analysis of behavior and neuroscience, including EEG, TMS, fNIRS, and brain–computer interfaces. Since 2024 he has been presenting on AI in ABA at international conferences, including the Best of ABA and the European Association for Behavior Analysis conferences. Few people sit at the intersection of building real ABA organizations, doing research-grade behavioral and neuroscientific work, and using frontier AI and multi-agent systems — including the open-source ClawSuite Relay he builds and maintains — as daily operating tools. Engagements are direct access to that intersection, applied to the systems your organization actually needs. Full CV →

Start a conversation

If you want to build the measurement layer instead of rent it, write to me.

hello@multiplicity.dev

Describe what you are trying to do in your own words. I will tell you honestly whether this is the kind of work I can help with — usually within a few business days.