OW Pulse
Human-in-the-Loop AI for Market Research at Scale
AI-powered research platform designed to accelerate expert workflows — not replace them
OW Pulse is a B2B research platform built for professional researchers and insights teams who need to move faster without compromising methodological rigor.
Unlike consumer-facing AI tools, our users are domain experts. They don't want AI to think for them — they want it to remove friction, automate heavy setup work, and give them more time to focus on interpretation and decision-making.
This fundamentally shaped how we designed and applied AI across the product.
Traditional research workflows were slow and operationally heavy:
- Researchers spent hours writing questions and programming survey logic
- Survey setup required translating Word/PDF questionnaires into online logic manually
- Complex skip logic, quotas, and randomization increased error rates
- Accessing the right audience required additional tools and coordination
- Existing AI solutions either felt like black boxes or tried to replace expert judgment
The result: long time-to-insight, low scalability, and high operational cost.
From discovery, we defined three non-negotiable principles:
Human-in-the-loop by design
AI should assist, not decide.
Optimize the heaviest work first
Question creation and survey programming were the real bottlenecks — not analysis.
Protect statistical rigor
AI should never calculate or invent statistical results.
- Product owner for OW Pulse end-to-end
- Defined AI strategy and human-in-the-loop principles
- Led discovery with researchers, ops, and commercial teams
- Partnered with Engineering and Data to design AI + algorithm boundaries
- Prioritized features balancing automation, control, and trust
- Oversaw launch and adoption across LATAM markets
Human-in-the-Loop AI for Survey Creation
Researchers don't want AI to replace their expertise — they want it to speed up execution. AI was applied to assist in generating survey questions based on research objectives, help structure questionnaires and logic flows, translate research intent into programmable survey logic, and reduce manual setup and repetitive configuration work. At every step, users could review, edit, and override AI suggestions. This significantly increased trust and adoption among experienced researchers.
AI + Panel Integration to Accelerate Insights
AI alone was not enough. By tightly integrating OW Pulse with our proprietary panel, researchers could reach specific audiences instantly, test hypotheses faster, and iterate on studies more frequently. This closed the loop from idea → audience → insight in a single workflow, driving significantly higher platform usage.
Clear Separation Between AI and Statistical Analysis
A critical architectural decision was never allowing AI to compute statistics. All analytical work is performed by proprietary deterministic algorithms and controlled statistical pipelines for crosstabulation, correlation analysis, and statistical significance testing. Only after the data is fully processed do we pass structured results to AI — strictly for interpretation, insight synthesis, and explanation and storytelling support. This preserved methodological rigor, reproducibility, and trust from expert users.
AI Accelerates Expert Workflows
AI creates the most value when it accelerates expert workflows
Human-in-the-Loop Design
Human-in-the-loop design is essential for trust in expert domains
AI Wins Before Analysis
The biggest AI wins often come before analysis, not during it
Separate Algorithms from AI
Separating deterministic algorithms from generative AI protects rigor
Respect Domain Expertise
Adoption grows when AI respects — not competes with — domain expertise
This project solidified my approach to building AI products:
Design AI as a copilot, apply it where it removes real friction, and never compromise on trust or rigor — especially when your users are experts.