Clarity first. Future-proofing.(Co-)Design › Build › Launch › Scale.

What was breaking

The state’s epidemiology program was robust and widely admired. Its data collection was not.

Disparate disease-specific surveillance systems sat alongside paper-based reporting, county by county. Each disease community had its own instrument, its own case definitions, and its own working idea of what “reported” meant, which made a statewide picture something you assembled by hand, late, and then argued about.

What stayed with people

Case definitions, readiness and the judgment about what the data means were never handed to the platform. The platform moved reporting; epidemiologists kept interpretation. Regular convening, active listening and transparency meant stakeholders felt included, and offered the expertise that made the base implementation stronger than the vendor configuration alone.

What we brought

• Product leadership
• Change management
• Leadership coaching
• Team mentorship

What we did

Balanced people, process and technology across a multi-year engagement, and brought the discipline of product leadership into public health.

• Sequenced by community, not by moduleEach disease surveillance community got its own phase: its own configuration, its own case definitions, its own readiness date inside the program plan.
• Prototypes before production, every phaseRapid prototyping sprints with user acceptance testing ahead of each release kept the platform relevant to the people who would live in it.
• Gates before scaleData-quality and model-evaluation gates, set once and adopted across programs.

What came of it

• Go-live was seamless and uneventful, on the calendar date that had been set.
• The state’s implementation became the baseline for an adjoining state’s.
• Data integrity and the reporting needs of an ongoing pandemic response held throughout, compliant with state and federal guidelines.
• Customizations the state arrived at were folded into the platform itself, approved by its governance body for any future implementation.

What was breaking

A general-purpose model answers with the same confidence whether it knows or not. For partners whose credibility rests on citing their sources, that is the one failure they cannot afford.

What stayed with people

A tool that cites its sources leaves the authority with the corpus and its authors. Partners provide the feedback and testing that hold the platform to the movement’s rigor, and separate learning pathways serve new and advanced practitioners.

What we brought

• Multi-tower agent architecture
• HITL governance design
• Corpus curation and source attribution
• Co-design facilitation
• White-label service model

What we did

The same architecture carries every deployment: a multi-tower conversational core, an HITL governance layer that routes uncertain answers to expert reviewers, and a corpus attribution system that ties every claim back to a source.

• Trained on an attributed corpus, not a promiseFor the Earth Law and Rights of Nature ecosystem, the corpus was drawn from trusted, attributed sources, so accuracy and traceability are properties of the corpus rather than promises about the model.
• One platform, extended into a new ecosystemThe Earth Law deployment runs on the same HITL-governed architecture as every other, extended with corpus curation and source-attribution protocols.
• Co-designed with the people held to itRapid prototyping and global engagement with Earth Law partners, who would answer for the platform’s accuracy.

What came of it

• Trusted-response rate of 92%, with hallucination risk down 40%, across deployments.
• Non-hallucinatory by design: the Earth Law deployment answers from an attributed corpus, not a general model’s recall.
• A B2B, white-label model for partner sites, proven first on the Earth Law ecosystem.
• Websites and PDFs repurposed as markdown for the Earth Law and Rights of Nature ecosystem.

What was breaking

Decisions about landscapes and seascapes are made locally, but the evidence that should inform them is scattered across public climate and ecological datasets that were never built to be read together.

What stayed with people

Action happens at the watershed, so that is the scale the evidence had to reach. Of the two systems, investors chose to scale the one that reached it. Both are built toward life in the loop: human in the loop is a transitional stage, not the destination.

What we brought

• Co-steward
• Living-systems-aligned AI
• Stigmergy
• Active inference
• Planetary co-intelligence
• Life-in-the-loop
• Climate and ecological data
• Lean, agile team

What we did

AI-driven decision support, powered by multiple public climate and ecological data streams, and built with a lean, agile team.

• Public data, fusedMultiple public climate and ecological data streams brought into one decision surface.
• Two scales, built side by sideOne system local in scope, coordinating through stigmergy and active inference; one a planetary world model for planetary co-intelligence, resolving biome vitality from the planet down to the watershed.
• Evidence picked the winnerAdvanced-stage investor diligence selected the second system, and resources were concentrated on scaling it.

What came of it

• Both systems reached MVP in about ten weeks each.
• The planetary-to-watershed platform was selected in investor diligence and is the one being scaled.

What was breaking

This renowned member-owned institution was struggling to rationalize and modernize disparate services, and members were the ones absorbing it.

Auto loans, home loans, insurance and financial advisory each ran their own path, which produced member frustration wherever a journey crossed a boundary. Executives shifted from multi-channel, a product-centric view, to omni-channel: member-centric, personalized and more holistic.

What stayed with people

An omni-channel strategy makes the next best offer and next best action personal to the member. Its deeper benefit is structural: it breaks down silos and unites product teams around one member rather than separate roadmaps. That reframing outlasted the engagement.

What we brought

• Digital Director
• Change management
• Leadership coaching
• Team mentorship
• Early AI/ML product strategy

What we did

Built an omni-channel strategy with the service division product leads, so the member experience is consistent across every service.

• Product roles redefined around the memberProduct roles became member-experience centric where that was what the work required.
• Click-and-mortar journey maps, co-defined per serviceJourney maps co-defined per service area bridged the digital and physical experience rather than treating them as separate products.
• Early AI/ML in the engagement layerProduct strategy for the digital experience embedded early AI/ML into omni-channel engagement, directing product leaders across product, design and operations.

What came of it

• Personalized, seamless and secure member experiences across new and existing channels.
• Every touch point covered by smart identification, authentication, verification and fraud detection.
• Churn down 7% across the digital experience serving 500K+ members.
• A Plaid publication, “A beginner’s guide to omni-channel,” shared the approach with the wider field, and national forums showcased the broader vision.

What was breaking

In Mexico City, a sizable population lacks access to traditional banking, particularly among domestic workers and caregivers.

Without an account there is no safe place to accumulate, no record to borrow against, and no path to the compounding that makes saving worth doing. The constraint was never willingness; it was access on terms that fit the way this work is actually paid.

What stayed with people

Rather than asking people to trust an unfamiliar institution, the model routed transactions through the neighborhood stores and the workers’ own advocacy organization. The app carried the mechanics; the people who already knew each other carried the confidence.

What we brought

• Business model and incentives
• Partnerships and distribution
• Product leadership and design
• Bilingual app (EN/ES)
• AI-assisted onboarding

What we did

Proved that micro-savings and micro-investment for the unbanked are feasible on their own smartphones, a device already ubiquitous in this population.

• Built with partners on the groundSponsored by executive leadership in banking and finance, with a domestic workers’ advocacy organization and a network of bodega stores acting as banking transaction centers.
• Digital wallets, blockchain and cryptographic securityTechnology chosen to make small amounts safe to hold and move.
• AI-assisted onboarding, tuned to real behaviorOnboarding built around how this population is actually paid and actually saves. That fit is what drove the growth.

What came of it

• Launched despite the pandemic, with only a slight delay, and well received by its audience.
• Seed funding secured for app development, and pilot partnerships across the Mexican financial and retail sectors.
• Sunset on evidence: when conversion and willingness to pay failed, the gap was not funded.
• Presented at a financial education forum and to a banking institution; the model and the platform both proved viable.

What was breaking

A pickup promise is only as good as the handoff between “buy online” and “pick up in store,” and that handoff is usually owned by two teams running two roadmaps.

Where digital and in-store fulfillment are built and measured separately, the customer feels the seam. The program needed one owner across both halves.

What stayed with people

Revenue and loyalty are usually reported as a trade-off in retail. Here they moved together, because the fix was the same fix: a shopper who trusts the pickup will be ready is also the shopper who rates the experience well.

What we brought

• End-to-end program delivery
• Conversational commerce design
• Digital-to-store fulfillment
• Cross-functional delivery

What we did

Drove the program end to end, rather than handing off between a digital workstream and a store-operations workstream.

• Chatbots at the digital front doorConversational chatbots carried shoppers into the pickup flow at the point of intent, which moved new in-store revenue rather than shifting existing sales between channels.
• Digital-to-store fulfillment, orchestrated end to endThe handoff was streamlined, so what the chatbot promised online is what the store could deliver at pickup.

What came of it

• In-store revenue up 7%, driven by chatbot engagement into the pickup flow.
• NPS up 5%, alongside revenue rather than traded against it.
• One program, one owner, digital and in-store.

What was breaking

A working product with a proven first market is the hardest thing to grow: what got it from 0 to 1 is rarely what takes it from 1 to 10.

Retail category managers and vendor reps planned trade in silos. Data arrived after it was needed, the impact of joint trade decisions against forecast was hard to see, and there was no single system of record: multiple ad hoc entries, with errors and inconsistencies.

What stayed with people

Agile product management and human-centered design were adopted across the organization, not only on this product.

What we brought

• Global Business & Product Director
• P&L ownership
• Design-driven product leadership
• Agile product management
• Human-centered design
• Omni-channel merchandising
• Sponsor-user validation
• Go-to-market and pricing

What we did

Owned strategy and the P&L together, so the direction and the numbers answered to the same person.

• Repositioned, not rebuiltThe offering was reframed around ML-driven customer insights for CPG and retail.
• 1-to-10 stage strategyGrowth sequenced as a scaling problem, not a second launch.
• A joint planning portalProject Janus gave retailers and vendors one real-time system of record for joint event planning, new item introductions and trade spend by category, with analytics on past performance and insight for future planning.
• Validated with sponsor usersThe business model was tested with sponsor users among existing grocery customers, with prospects lined up next.
• Hub-and-spoke go-to-marketTiered, à la carte pricing with retailer- and vendor-based revenue, a mobile add-on, and a path from existing customers to new retailers worldwide.

What came of it

• 8X market expansion across the CPG and retail sector.

What was breaking

Deep-tech ventures at this stage are usually strong on the science and thin on the story. The technical differentiation is real; what is missing is the language that turns a sensor spec or a material property into a reason a customer or an investor acts.

What stayed with people

The technology does not sell itself. The founders who internalized that were the ones who could repeat the exercise on the next feature or the next market without an advisor in the room.

What we brought

• Go-to-market advisory
• Commercial narrative and positioning
• Market and buyer identification

What we did

Go-to-market advisory across several ventures at once, in unrelated technical domains, translating innovation into commercial value in each.

• Drone surveysA geospatial data-capture capability translated into a commercial offering for the industries that need that data.
• Biopolymer packagingA materials-science formulation positioned as a commercial alternative buyers could specify against.
• Reusable market-analysis frameworksBuilt once and applied across the portfolio, halving strategy formulation time.

What came of it

• Revenue outcomes of 3–5X across the portfolio.
• A commercial story per venture: technical differentiation translated into value a customer or investor could act on.

What was breaking

A high-visibility initiative with a fixed public deadline, a new market, and a technical partner not yet chosen: every one of those is a way to arrive late.

The gap it answers is connection, not capital. Despite record climate and nature finance commitments, capital still struggles to reach high-integrity, locally led solutions, held back by fragmented data, costly due diligence and limited visibility of credible opportunities.

What stayed with people

A high-visibility initiative with a fixed public deadline arrived early, because scope, not the date, was the thing allowed to move.

What we brought

• 0-to-1 product roadmap
• Technical-partner due diligence
• Feature-timeline guardrails
• Expert AI advisor
• Explainable, human-in-the-loop AI
• EU AI Act and GDPR readiness

What we did

0-to-1 product roadmap and technical-partner due diligence for a global deal-flow discovery marketplace.

• A roadmap with guardrailsEvery feature carried a timeline guardrail, so scope gave way before the deadline did.
• Diligence before commitmentTechnical partners were tested against the roadmap before the initiative was bet on them.
• An AI-enabled, science-guided marketplaceConceived as a Luxembourg-led international initiative, built to connect investors, funds and public capital with validated, investment-ready, locally led climate and nature solutions, so projects move from visibility › validation › investment.
• Expert AI advisor on the build tenderEvaluated competing builders in an open tender on AI/ML capability, explainability and audit trails, EU AI Act and GDPR readiness, EU data residency, human-in-the-loop design, Global South accessibility and non-extractive data stewardship.

What came of it

• Delivered 20% ahead of schedule, against a COP31 deadline.