Jo Flores

Chief of Staff · Operations · New York City

Building operating functions from zero to 1.

I put company knowledge where agents can execute against it, then run the company on it. At AncestorIQ that meant a version-controlled knowledge base agents query reliably, a skills library that constrains what they produce, and a daily competitor monitoring agent that runs unattended. At JumpStart it meant four agents handling recurring admin that used to live in someone's memory, every one of them drafting or proposing and waiting for a human on anything irreversible.

Before operations I was embedded in healthcare itself, in NYC emergency departments and ICUs and then medical school. In 2023 I joined ExactRx (Techstars '25) as the founding operator and built the function from scratch. Being a 0–1 operator is a lot more like working in the ER than anything most MBA grads and consultants have seen.

Jo Flores
Seeking: Chief of Staff, Head of Operations Stage: Seed to Series A Where: New York City, in person or hybrid. Authorized to work in the US. Sectors: Health tech · Gov tech · Social services · Wellness · Open to others Note: Looking to move from fractional into full-time, embedded with one high-growth company
Email me Writing → Download CV (PDF) Book 20 minutes

Forwarding me to someone? Steal this.

Jo Flores, Chief of Staff and operations, NYC. Built ExactRx's ops function from zero as the founding operator and the team went one to six, Techstars x Northwestern Medicine '25. Now runs operations across three organizations: took one from three clients to eight and four times monthly revenue. Ships agent infrastructure: a daily competitor monitoring agent, a skills library that constrains agent output to production standard, and four recurring-admin agents at JumpStart. Every one of them drafts or proposes and waits for a human on anything irreversible. Looking for seed or Series A in health tech, gov tech, or social services, in person or hybrid in New York. joflores.com

One to six.

Team growth at ExactRx while I built the operations function as its first operator

8

Accelerators, awards, and federal programs ExactRx won on my award strategy

Four times.

Monthly revenue growth at JumpStart, with clients from three to eight

How I operate

Every example below is from an actual engagement. Nothing here is illustrative.

AI operations

I make agent output trustworthy enough to act on

  • Skills library in the repo constrains agent output to production standard
  • Daily agent monitors ~12 competitor sites unattended, with typed error states and a scraping layer I specified once bot protection kicked in
  • Authored the guardrail spec that stopped the product's own agent from silently persisting a date it inferred from ambiguous evidence
  • Everything else runs gated on purpose

Result

Recurring research runs with nobody in the loop. A system with write access to user data stops inventing facts.

Founder leverage

I turn two weeks of scatter into one decision pass

  • Commits, product plans, and GTM work pulled from across channels into one sourced memo
  • 13 open decisions distilled to 10, ranked Blocking / High / This month / Roadmap
  • Every claim attributed inline so nothing had to be taken on faith
  • Source-linked knowledge base built from the channel where decisions actually get made, not curated later into a wiki nobody opens

Result

The founder decides in one sitting. The same argument doesn't resurface six weeks later.

Technical judgment

I read the PRs, not the status update

  • Use Claude Code CLI to read actual commits, diffs, and branch history directly
  • Reconciled stakeholder feedback against 62 commits on main, separating fixed-but-unpulled from genuinely open
  • Surfaced a 28-commit branch-sync gap hiding real fixes
  • Read one sprint of eight commits and returned the strategic read: cache TTLs varying by data source were deliberate margin engineering, which made a four-tier pricing architecture viable

Result

The team stopped re-litigating solved problems. Pricing came from real marginal cost, not a competitor's page.

Risk

I catch what ships before it ships

  • Flagged a third-party people-search integration for privacy backlash, regulatory exposure, and litigation risk
  • Specified the mitigations, not just the objection: gate contact disclosure behind verified family relationship, get terms reviewed, build a removal flow
  • Caught a fairness defect in my own verification design: users whose ancestry predates digitized records would have looked under-verified through no fault of their own

Result

Two pre-launch catches, one commercial and one ethical, both found by reading what was actually built.

Spend

I say no to spending before it becomes a commitment

  • Re-scoped a foreign archive integration from ~$100K to ~$30K
  • Then recommended pausing entirely and sequencing distribution first: two interested test users is not a demand signal
  • Logged with an explicit revisit condition so it was a pause, not a veto
  • Canvassed seven usability-testing platforms against a $500 monthly budget; argued for a paid panel once organic recruiting stalled

Result

Capital preserved without closing the door. Engineering agreed on the record.

Systems of record

I make one place true and connect the rest to it

  • JumpStart: Monday.com across 7 boards joined by a shared short code, with triggered automations for contract signature, invoicing, and auto-creating work items from intake answers
  • CPC: 13 workspaces and ~150 boards built separately by four staff, audited into a single source of truth per category

Result

JumpStart went from 3 clients to 8, revenue 4x. CPC phase one exceeded original scope.

Small business AI

I increase output and oversight without adding headcount

  • Proposal Follow-up bot: scans JumpStart's signed-adjacent leads for ones going silent, counts business days in real Python, checks the inbox for a reply, and drafts a nudge sequence tagged to the right person. Never sends on its own.
  • Monthly Deliverables Generator: fires on the 1st and creates next month's production rows automatically
  • Daily Wiki Freshness Check: opens a PR whenever a page goes stale, ~30 PRs opened, each decided by a human
  • New-client-setup skill: executes three post-signing steps in one run
  • Every agent drafts or proposes. Every irreversible action waits for a person.

Result

Recurring admin work that used to live in someone's memory now runs on a schedule, with a paper trail and a human in the loop on anything that matters. More on how I build this →

The company keeps running when the founder cannot

AncestorIQ's founder-CEO was pulled back into full-time work elsewhere and could only answer late at night. I ran the operating cadence and made the calls, kept his decisions and the tech lead's in sync asynchronously, and sent him only what genuinely needed him. I coordinate a three-person core plus a rotating bench of eight or more engineers, designers, and specialists across US, European, and Asian time zones, and I own the commercial side of that: rates, scope, allocation, invoicing disputes, onboarding, and saying no to candidates.

Regulated by preference

Healthcare, government, and social services. I worked bedside in NYC emergency departments and ICUs, went to medical school, and hold a master's in ethics and society focused on medical ethics and policy. I also ran a full IP program: authored the patent strategy, sourced and capped outside counsel, and drove three trademark filings across four classes against a litigious incumbent. I treat the constraint as the interesting part, not the obstacle.

Selected work

JumpStart DPC Solutions

Chief Operating Officer

Feb 2026 to present · Healthcare marketing & ops agency

The situation

A marketing and operations agency serving direct primary care practices, with three clients and no system of record. Client facts, brand standards, and delivery method lived in whoever last touched them.

What I built

Monday.com as the company's system of record across seven boards joined by a shared short code, with triggered automations for contract signature, monthly invoicing, and auto-creating branding and website work items from a client's intake answers.

A 25-file operating library that turns agency method into executable standards: per-client brand and compliance guidelines, a messaging framework, and a social production pipeline with Python and Node QA scripts that check hook length, banned characters, unresolved placeholders, and AI-tell density before a document is assembled.

The DPC JumpStart Kit, a $2,800 onboarding package. Client-facing builds in house, including a print-ready trifold brochure to PDF/X-1a spec and a before-and-after website redesign case study. I also hired and trained the account manager who now runs client delivery.

Wired Claude into Monday.com after a written SOP failed for months: account managers kept working client calls off personal phones because logging them was slower. I moved the capture into the tool itself so it updated notes, status, and due dates automatically. Adoption followed once the compliant path was also the easy one.

Four agents handling recurring work that used to live in someone's head: a Proposal Follow-up bot that scans for signed-adjacent leads going quiet, counts real business days in Python, checks the inbox for a genuine reply, and drafts a nudge sequence tagged to the right person (never sends on its own). A Monthly Deliverables Generator that fires on the 1st and creates next month's production rows automatically. A Daily Wiki Freshness Check that opens a PR whenever a page goes stale — roughly 30 PRs opened, each decided by a human. A new-client-setup skill that executes three post-signing steps in one run. Every agent drafts or proposes. Every irreversible action waits for a person.

Result

Clients went from three to eight and monthly revenue multiplied four times. Compliance rules that used to depend on someone remembering, like a client whose providers are nurse practitioners and whose copy must never say "physician," are now written into files the work is checked against.

AncestorIQ

Chief of Staff and Operations

Aug 2025 to present · AI-native genealogy platform

The situation

An AI-native genealogy company with no decision infrastructure: product plans, commits, and GTM work scattered across channels. Then the founder-CEO was pulled back into full-time work elsewhere and could only answer late at night.

What I built

A source-linked knowledge base covering product architecture, brand, GTM, and competitive intelligence, structured so both people and agents query it reliably, with every claim traceable to a commit, doc, or interview. Written from the channel where decisions get made rather than curated later.

On top of that: the decision-memo process, the operating cadence, and a skills library in the repository that constrains agent output to production standard. A daily competitor monitoring agent that watches roughly a dozen sites unattended, with typed error states and a scraping layer I specified once bot protection started blocking it. I also authored the guardrail spec that stopped the product's own agent from silently persisting a date it had inferred from ambiguous evidence — a write-access system inventing facts, caught before it shipped. Plus the IP program and contractor coordination across US, European, and Asian time zones.

Result

The team kept moving on the decisions that did not need him, and his few hours went to the ones that did. The company's output scaled on AI rather than on new hires.

ExactRx

Head of Operations, Founding Team

Jan 2023 to Aug 2025 · Techstars '25 · B2B healthcare SaaS

The situation

An early-stage company automating pre-operative coordination for ambulatory surgery centers. No operations function and nobody to build one. I was on leave from medical school when I came in as the first operator.

What I built

The operations and hiring functions from zero. I also owned the award and visibility strategy, which is how a company this small got in front of health systems, federal programs, and a state governor.

Result

The team grew from one person to six. ExactRx secured selection into eight programs, including Techstars x Northwestern Medicine and NY Gov. Hochul's Empire AI Roundtable, where it was the only startup invited. Northwestern Medicine planned a pilot through the Techstars cycle.

CPC

Project Management, Operations and Monday.com Consultant

Dec 2025 to present · NYC policy and advocacy nonprofit

The situation

Thirteen workspaces and roughly 150 Monday.com boards, built separately by four different staff, with no agreed source of truth for any category of data.

What I built

I did not build those boards. I audited them, decided which one is the source of record for each category, and connected the rest to it. On top of that: nine internal apps, nine fiscal automations, and a production MCP server with scoped OAuth. I structured the engagement as a phased statement of work with milestone-based payment.

Result

Phase 1 delivery exceeded the original scope. A legal entity name that had been silently wrong across several boards surfaced in the audit and was corrected the same day.

Worth saying out loud

I turned AI down twice on that engagement, in writing, where automating would have degraded the client's data. I told them why. That is the same judgment behind everything else I shipped there, and it is the part I would want a founder to hire.

Programs ExactRx won on my award strategy

Techstars x Northwestern MedicineSpring 2025 cohort
Empire AI RoundtableNY Gov. Hochul · only startup invited
ARPA-H Investor Catalyst HubFederal health research
MassChallengeUS Early Stage
NVIDIA InceptionAI startup program
NSF Innovation CorpsI-Corps
Femtech World AwardsSector award
Femovate by GuideaSector program

What the first 90 days look like

Every company I have joined was already running on something. The job is not to replace it on day one, it is to find out what is load-bearing before I touch it.

1–30

Audit before I build anything

I interview everyone, then map where decisions actually get made against where they are supposed to get made. I find what is tracked in someone's head, what is tracked twice, and what nobody owns. At CPC that surfaced a legal entity name that had been silently wrong across several boards. You get a written picture of the operation and a ranked list of what is costing you most.

31–60

One source of truth, then the cadence around it

I pick the system of record per category and connect everything else to it, rather than letting every tool claim to be canonical. Then the operating rhythm: planning, prioritization, retro, and a decision log so the same argument does not get relitigated in six weeks. This is the part that lets you stop being the router.

61–90

Put the knowledge where agents can run on it

Company context, standards, and method move into version-controlled files that both people and agents execute against, with QA that runs before work ships. Recurring work starts moving to agents. I will also tell you plainly which parts should not be automated, because automating the wrong step quietly degrades your data and you find out months later.

Before operations

I came to operations closer to the OR than a consultant desk. Bedside ER & ICU to health tech. And to be honest, being a founding operator is much more like covering the ER than anything most consultants have seen.

2020 to 2022

Loyola University Chicago Stritch School of MedicineDoctor of Medicine program. I took a leave of absence in good standing to care for family, and that leave led to ExactRx.

Dec 2022 to Aug 2023

NYU Langone HealthSimulation Operations Associate.

Dec 2017 to Sept 2020

Weill Cornell MedicineSimulation Specialist, Research Coordinator, Research Assistant. I ran interdisciplinary in-situ simulations across emergency medicine, internal medicine, OB/GYN, and pediatrics, including the SARS-CoV-2 PPE and contamination response in March 2020, and developed remote telesimulation techniques during COVID-19. Member of the NIH All of Us ELSI working group.

Aug 2016 to Jan 2019

ScribeAmerica at NYP / Weill CornellED Medical Scribe and Certified Floor Trainer across the adult and pediatric emergency departments at two hospital sites. I trained and evaluated incoming scribes.

Fordham University

MA, Ethics and Society · BS, BiologyGraduate work focused on medical ethics and policy.

Publication

Engaging Virtual LearnersUsing an Obstetric Emergency Training Initiative as a Framework for Interprofessional Telesimulation Modeling.

What I am looking for

Role

Chief of Staff or Head of Operations. The first real operating hire, or the person who replaces the founder's improvised system with one that holds.

Stage and sector

Seed or Series A. Health tech, gov tech, social services, wellness, and other sectors where regulation and trust are the hard part — the constraint is the interesting part, not the industry label.

Where

New York City, in person or hybrid. Not remote. I would rather be in a room with people building something demanding, and operations work is harder to do well at a distance. Authorized to work in the US.