What should we build with AI first?
AI Product Strategy
Clarify the customer problem, product surface, and roadmap sequence before defaulting to generic AI features or “chat with your data” ideas.
AI-native product growth · B2B SaaS
Turn AI pressure into product growth.
Most teams have AI tools now, but the work itself has not really changed. I help B2B SaaS teams find one workflow where AI can reduce effort, improve quality, or create a safe experiment worth testing.
Then we connect that learning back to product, growth, and operating decisions.
10+ years product leadership · Toyota · Zalando · reev
Shipped customer-facing AI · built internal agentic systems · connected product signals to expansion decisions
The market problem
The AI-native product shift is here. Most product and growth systems have not caught up.
AI does not create operating change just because tools are available. Most teams already have AI access. The harder question is why the work itself still looks the same.
The practical starting point is the gap between the documented process and the real workflow: the delays, handoffs, judgment calls, exceptions, and control points that usually sit outside the clean SOP.
That is where AI opportunity becomes visible.
Common symptoms:
- AI features ship without a clear customer problem, adoption path, or commercial model.
- Chat-style product ideas appear before teams understand activation, usage, or GTM.
- Pricing and packaging decisions get made before adoption and value are proven.
- Internal AI tools appear without redesigning the workflow, decision points, or quality checks.
- Product usage, customer behavior, and expansion signals stay disconnected from the decisions they should inform.
If the documented process is simpler than the real workflow, AI should not be added blindly. First, inspect where the work actually waits, repeats, breaks, or requires judgment.
See the example outputWhere I help
I work where AI changes the product, the growth model, and the workflow.
How do we activate, package, price, and expand it?
AI-Native Growth
Connect AI usage to activation, product-led growth, packaging, pricing, behavioral signals, and customer-base expansion.
How should work change when AI can execute part of the process?
Agentic Workflows
Design human + AI operating systems with clear decision points, escalation paths, quality checks, and workflow ownership.
How do we move from ambiguity to something working?
AI Build & Operator Advisory
Hands-on help turning a validated AI opportunity into a testable prototype, bounded workflow, agent-assisted process, or product-to-revenue learning loop.
Start here
Start with one real case. Go deeper only when the opportunity is worth it.
Most teams start with the AI Workflow Workshop. Some come directly to the diagnostic or sprint when the workflow decision is already clear.
Step 1 · €1,000
AI Workflow Workshop
Where should we start?
€1,000 to identify whether one workflow is worth changing with AI before you spend weeks building, buying, or debating. You leave with one concrete experiment, the expected business impact, the human controls needed, and a recommendation to test, diagnose, sprint, or stop.
Step 2 · €2,500
AI Workflow Diagnostic
What exactly should we test?
A one-week diagnostic to turn the chosen workflow opportunity into a clear experiment plan with owners, metrics, risks, human checkpoints, and implementation options.
Step 3 · From €5,000
Operator / Advisory Sprint
Can you help us make it happen?
A four-week sprint to help design, prototype, test, or operationalize the workflow experiment.
How I work
Map the product and growth problem → inspect one real workflow case → design the smallest useful experiment → measure what changes.
Selected proof
Proof across product, AI, growth, and workflow decisions.
reev · Customer-facing AI
Shipped customer-facing AI, not a demo.
Delivered reev’s first customer-facing AI feature: a natural-language insights engine for fleet operators, built from the customer problem outward rather than as a generic AI add-on.
reev · Behavioral scoring
Turned product behavior into expansion prioritization.
Built and operationalized a qualified expansion pipeline covering roughly 85% of the annual target, combining product signals with CSM qualification.
Product growth · Pricing architecture
Connected product, packaging, pricing, and customer-base growth.
Led pricing architecture and product growth work at reev, translating usage, segment, and workflow signals into commercially useful decisions.
Operator systems · AI workflows
Built and operated, not just advised.
I use agentic systems in my own work across research, product decisions, operating coordination, and commercial workflows.
“Luis has a rare ability to connect product strategy, AI opportunities, monetization, and execution. At reev, he helped turn complex business requirements into scalable product solutions, drove important pricing and packaging work, and pushed AI-driven workflows forward with strong ownership and strategic clarity.”
Christian Krawczyk
Head of Product, reev
How I build
Working with AI systems, with human judgment in the loop.
I use these systems in my own product, growth, research, and operating workflows. Their value does not come from being magically autonomous. It comes from deliberately deciding what AI can interpret, what software should enforce, what can be persisted or executed, and where I remain responsible.
Mira
A persistent Hermes agent I operate through Telegram and CLI, with memory, tools, session history, and scheduled runs. I use it for research, coordination, briefings, and maintaining operating context across my work.
BoundaryAI interprets, plans, and uses tools. The underlying system validates and records runs. I approve durable changes, external communication, publishing, and business decisions.
Evidence75+ recorded sessions, 900+ tool results, and a recurring weekday briefing workflow as of July 2026.
Growth Signal OS
A Supabase-backed product-growth cockpit I built with Bruno, my coding agent. It turns public B2B SaaS evidence into structured signals, diagnostic hypotheses, dashboard views, and a prioritized human action queue.
BoundaryAI interprets evidence and proposes hypotheses. TypeScript, Zod, Supabase, SQL, and React make the workflow structured and inspectable. I decide whether the evidence is credible and whether to act.
EvidenceWorking data model, validated ingestion workflow, read-only dashboard views, an outcome-learning layer, and 18 passing system-contract tests.
Not every workflow needs an agent. Every AI system needs clear boundaries: what it can read, prepare, persist, and execute, and where a human makes the call.
How to start
Bring one real workflow case.
In 90 minutes, we will review how the work happens today, what AI could produce from the same case, where it fails, what humans must control, and what practical experiment should come next.
Latest writing
Notes on the decisions behind AI-native product growth: what to build, how to price and package it, how to redesign workflows, and how to turn usage signals into growth.
July 24, 2026 · 10 min read
From AI research chat to an inspectable decision loop
How I turned scattered AI research into structured evidence, deterministic scoring, and a human action queue, without building an autonomous outreach machine.
July 4, 2026 · 7 min read
The monetization conversation your team keeps deferring
Most B2B SaaS teams treat monetization as a pricing page problem. In the AI era, that's the wrong frame, and the cost of getting it wrong is already showing up in your margins.
If AI is turning into a product, growth, or workflow decision, start with one real case.
Bring one anonymized workflow case your team is already discussing. In 90 minutes, we will review what happens today, what AI could realistically help with, where it could fail, and what practical experiment should come next.
Book a 30-minute call