Independent applied-AI advisory by David Scheler · Berlin, Germany
I help product and engineering leaders connect data, people, processes, and ownership—so AI removes friction, improves the flow of intelligence, and strengthens the organization instead of optimizing for quick ROI alone.
10+ years leading applied AIAcross product, engineering, and enterprise delivery
Voice AI since 2008From machine-learning research to production systems
From prototype to enterprise programHands-on technical work backed by product leadership and complex delivery experience
Junglebrains design perspective
AI systems carry assumptions about data, workflows, roles, interfaces, incentives, and exceptions. As those systems are iteratively refined and embedded more deeply into an organization, they encode and amplify those assumptions, increasingly constraining how the organization operates.
That can strengthen a connected organization—or harden its fractures. Apparent ROI of an isolated AI deployment may therefore prove usefulness without showing whether the intervention improves the organization’s longer-term trajectory.
Junglebrains uses this as a working design perspective and a reason to make operating models, data relationships, ownership, and feedback loops explicit early.
Common failure
The automated task improves, but delay and overload move to another person or process.
Junglebrains designs for
Map the complete flow and measure whether friction actually leaves the system.
Common failure
Higher throughput creates monitoring, exception-handling, and rework that absorb the apparent gain.
Junglebrains designs for
Design ownership, escalation paths, correction capacity, and learning loops from the beginning.
Common failure
Financial KPIs rise while stakeholder needs, resilience, adoption, or internal capability erode.
Junglebrains designs for
Evaluate financial outcomes alongside workload, latency, resilience, adoption, and long-term ownership.
How I can help
Junglebrains works with teams that already own the outcome. I step in where product, technical, and organizational decisions have become inseparable, bringing hands-on depth to test critical assumptions while the internal team retains delivery ownership.
01
Identify stakeholders, information flows, delays, handoffs, overload, conflicting incentives, and the evidence that should define success.
System map · Decision criteria
02
Clarify operating models, ownership, decision rights, data models, system boundaries, interfaces, and make-or-buy choices.
Decision memo · Data and interface model
03
Build or specify the smallest useful prototype that can test an important interaction, integration, or organizational assumption.
Prototype · Feasibility evidence
04
Define a roadmap and measures that capture immediate value without ignoring correction effort, human workload, displaced bottlenecks, resilience, and long-term ownership.
Measurement framework · Roadmap · Handover
Selected evidence
These examples focus on my responsibility and the public evidence while keeping unapproved customer and implementation details private.
Conversational AI · Production
Designed, built, and deployed a conversational system for a German customer-acquisition business. It reached prospects who had remained unavailable by phone and generated direct revenue during its production run.
Product framing · Conversation design · Implementation · Deployment
SoundHound AI · Automotive Voice AI
At SoundHound AI, I served as the primary onsite technical interface during the MVP phase of a global automotive Voice AI program, connecting engineering teams with the OEM stakeholders, the systems integrator, UX, and technology partners across architecture, requirements, dependencies, and risk.
Technical program interface · Architecture trade-offs · Partner alignment
Medical Voice AI · Product leadership
Shaped the roadmap for a Voice AI SaaS product and led a cross-functional product and engineering team spanning product operations, design, conversational AI, machine learning, frontend, and platform development.
Product strategy · Team leadership · Cross-functional delivery
Earlier work at Volkswagen Group includes production speech and NLU systems, patents, and a multi-user automotive Voice AI prototype shown at CES 2016. My background also includes peer-reviewed machine-learning research and university teaching.
Natural flow of intelligence
Automation is an intervention in an organizational system, not an end in itself. The work stays close to the people who own that system and leaves them with greater clarity and capability.
Technical feasibility is considered alongside stakeholder needs, information flow, adoption, ownership, and correction costs.
Every engagement is led and delivered by David Scheler. There is no account-team handoff between the question and the work.
Engagements are bounded. The rationale, artifacts, and ownership remain with the internal team.
You have a consequential applied-AI question, an existing team, access to the relevant context, and the intention to own the outcome.
You need high-volume outsourced delivery, generic AI training, staff augmentation, or validation for a technology choice that has already been made.
Low-friction first step
In a free 30-minute call, we’ll assess the situation, whether Junglebrains is a useful fit, and what a sensible next step could look like. No preparation or polished project brief required.