Where I'm most useful.
Most of my work falls into a few patterns. If you see yourself in one of these, we should talk.
0-to-1 Product Strategy
When you need a founding PM to take a system from a mandate to a roadmap a CTO, partners, and sales can all get behind, I'll define the bets, the sequencing, and what's deliberately out of scope for v1.
Integrations & Platform Architecture
When bespoke, one-off integrations are eating your engineering budget, I design a canonical data model and API layer so the tenth build costs less than the first, not just faster.
Workflow Automation
When a feature's shipped but barely used, I dig into activation and usage data to find the highest-friction manual step and automate that first, instead of guessing at what's next.
Agentic AI, Grounded
When you want AI in the product without the hallucination risk, I define the orchestration layer and the accuracy benchmarks (with human review) that make it trustworthy before rollout.
Compliance & Data Trust
When employee or customer data starts crossing a regulatory line, I build consent, data minimisation, and audit requirements into the spec itself, not a checklist we bolt on after launch.
Making a product feel right.
It's the little, subtle details that decide whether people trust and actually use what ships.
Reusable before repeatable
I default to a canonical data model or a shared API layer, so the tenth integration costs less than the first. Not just faster to copy and paste.
Consent by default
Data only moves with explicit consent, and only the fields actually needed for the job. Never a full record by default.
Human-reviewed AI
Every agent response is grounded in live product data and checked against human review before it ever reaches a customer.
Metrics that survive scrutiny
I report activation, MAU, and ARR the way finance and engineering would both sign off on. Not a vanity number picked after the fact.
The unglamorous replatforming
I own the migration and schema work nobody wants to spec, because it's usually what makes the next order of magnitude possible.
One PM, full loop
Discovery through release sign-off and QA, with no separate program layer to hand off to. I carry it end to end.
How I usually plug in.
Every team and product is different, but most engagements follow this shape.
Understand
Start from data and the person feeling the problem: support tickets, usage patterns, a partner's monthly closure numbers, before I touch a roadmap.
Shape
Turn the problem into a scoped bet: what gets a reusable architecture (like a canonical data model) versus what's a one-off, and what's deliberately out of scope for v1.
Ship & measure
Own it through release sign-off and QA, then track the metric that actually moved: activation, MAU, ARR, or an escalation rate. Not just whether it shipped.
Common questions.
What kind of B2B SaaS problems do you specialize in?+−
HRIS and payroll platforms, embedded integration marketplaces, workflow automation, and more recently, agentic AI grounded in live product data. I've done this as a founding PM, owning discovery through release sign-off.
Have you managed a team, or been an IC PM?+−
IC and founding PM. I own the full loop myself (discovery, specs, prioritization, QA, release sign-off), while working closely with a CTO and partner teams on things like AI orchestration and compliance requirements.
What's your process for a 0-to-1 feature vs. an established product?+−
For 0-to-1 work, like the integrations marketplace or the AI assistant, I front-load architecture decisions (a canonical data model, an orchestration layer) so the tenth build is cheap, not just the first. For established products, I lead with usage data to find the highest-friction manual step before proposing anything new.
If this sounds right, let's talk.
A short conversation is usually enough to see if there's a fit.