Built by a CFO, not a software company
Inflect is the practice we wished we'd had — a senior fractional CFO with a named analyst that remembers every decision, correction, and board preference, and carries them forward when the CFO rolls off.
Three months of memory, gone in a Dropbox handoff
A fractional CFO rolls off a Series B SaaS engagement after fourteen good months. Twelve careful board packages. Three hundred Slack threads documenting the “why” behind tricky calls — the reason NRR is computed without services revenue, the reason the Q2 hiring-plan variance is called a “temporary investment” rather than a headwind, the reason the Rule-of-40 narrative leads with the operating-margin variant.
The next CFO inherits a Dropbox folder, four PDFs from three different tools, and a context doc that nobody updated after month three. Month one is worse. Month two is the same. By month three, the new CFO has re-derived half of what the last one knew, and the client is paying for the tuition twice.
Inflect exists because that scene should not be normal. The memory is most of the value of a senior CFO. And the memory is what leaves the building when the CFO does.
Phil Davis
Phil runs the fractional-CFO practice that builds Inflect as it learns. Prior: $2B+ in finance operations across SaaS, marketplace, and vertical-software companies. MS Computer Science, MIT. He built Inflect the way he wished his junior analyst had worked — as a longitudinal memory that cites its sources, not a chat box that starts over every month.
Reach Phil directly: phil@inflect.finance.
Four operating beliefs that shape what Marlow does — and doesn't — do
The CFO owns the judgment.
Marlow drafts. The CFO signs. The workflow refuses to let the two steps merge, at any price. We won't ship autonomous AI output — even if the founder asks, and even if it's cheaper.
Every financial claim should be traceable.
Every number in a Marlow draft has a lineage back to the source row. Every metric card shows its formula and inputs. If a claim can't cite a source, it doesn't make the package.
AI should surface uncertainty, not hide it.
A confidence indicator on every computed metric. A “structurally can't be computed” flag when a formula is missing inputs. We'd rather show a blank with a reason than print a confident-looking 0.0%.
Every board question should make the next package better.
If a board member asked why NRR moved, next month's package opens on NRR movement. If the Accel partner asked for a hiring-plan comparison, it's in the template going forward. The practice should compound; the CFO shouldn't have to remember to make it compound.
Institutional finance memory
Not another dashboard. Not another AI chat box. A named analyst — Marlow — who remembers every decision and preference you've confirmed, attributes her work, and gets sharper every month. The compounding layer that used to live only in the CFO's head.
Design-partner mode, small cohort
We're currently in design-partner mode with a small cohort of Series A–B SaaS founders, two operating partners at growth-stage PE firms, and three fractional-CFO practices. No paid testimonials, no AI-generated logos, no fabricated quotes on the homepage. When we have named references that are OK to share, they'll appear here. Until then, the blog and the sample package are the proof we can offer.
If the shape of that resonates — or if you think we're wrong about something — tell us.