Operator-led · Paris

AI in operations,
actually shipped.

We help companies transform their operations with AI. We diagnose where AI actually moves your P&L, pick the right tech and AI partners across the market — vendor-agnostic, no allegiances — and ship them live in production in ninety days. Or we keep working until they are. One methodology across every engagement: The Grid Operating System.

EU-native AI Act–ready Vendor-agnostic
See the offer
Where we operated before Grid
01 — The Offer

One offer.
Two phases. One asset.

We do one thing. We diagnose where AI moves your P&L, then we ship the two or three interventions that matter — live in production, not in a deck. One operating discipline, end to end. No army of juniors, no scope creep.

i. Diagnose
Five-week fixed-fee sprint

Where AI actually moves the P&L.

Interview-led, operator-driven, vendor-neutral. We map your operations against the AI use cases that compound — not the ones that look good in a deck. You leave with a costed roadmap your CEO and Operating Partner sign in the room.

  • Twelve to fifteen stakeholder interviews across GTM and operations
  • Process and decision-rights audit, mapped to AI-readiness
  • Build-versus-buy per use case, with named alternatives
  • Costed ROI roadmap — two or three scoped interventions, prioritised
  • AI Act risk classification per use case, written in
Delivered via The Grid Operating System
ii. Ship
Eight-to-ten-week implementation

Live in production, not in a deck.

We embed with your operations and go-to-market org to ship the prioritised bets. Buy-then-orchestrate by default — thin wrappers on the tools you already license. Custom only where data, moat or unit economics earn it. We design ourselves out before we leave.

  • AI-native operations: pipeline, forecasting, CS, support, back-office
  • Frontier or open-weight models, picked per task — not per partnership
  • EU-hosted inference where regulated data flows demand it
  • Change management built in: champions, rituals, decision rights
  • Hand-over week: documented runbooks, trained owners, thirty-day warranty
Delivered via The Grid Operating System
iii. The AI-Readiness Index
What compounds across engagements

Your peer benchmark.
Not just your roadmap.

Every diagnostic places your business against a cross-portfolio benchmark of peer scale-ups — twelve dimensions, sixty scored items, anchored rubrics. You see not just where to go, but where you actually stand against the companies your investors already back.

  • Twelve-dimension AI-readiness score — data, decision-rights, ops, change
  • Anchored against anonymised peer benchmarks across PE portfolios
  • Frozen instrument — comparable across diagnostics, versioned annually
  • Score deltas tracked from diagnostic to hand-over
  • PE-portfolio roll-ups available on request
Delivered via The Grid Operating System
02 — Who we fit

Three audiences.
One operating discipline.

Scale-ups building AI-native operations from the start, mid-market leaders putting AI into established functions, and PE-backed portfolios where the operating partner needs the same playbook across multiple companies. Different industries. The operating discipline does not change.

i.
For scale-ups

AI-native operations, from the start.

B2B SaaS and tech-enabled services, €15–80M ARR. The next inflection depends on how much your team can run with fewer people and more leverage. We embed, audit, ship the two or three interventions that compound — and leave you with the system, not the consultants.

ii.
For mid-market leaders

AI where it moves the P&L.

Established companies (€100M–1B revenue) putting AI into ops, support, revenue and back-office without rebuilding the whole house. We help you pick the right tech and AI partners across the market, then ship the work in production — not as a deck for the steering committee.

iii.
For PE-backed portfolios

One playbook across portcos.

Operating Partners running AI value creation across multiple portcos. We run the diagnostic once, calibrate it across the portfolio, and benchmark portco-by-portco — so value-creation theses defend themselves at the next IC.

03 — How you buy

Three engagements.
Fixed fee, gated next.

Three productized SKUs. Each fixed-scope and fixed-fee, each earns the next. Diagnostic gates Implementation. Implementation gates the Operating Partner retainer. No open-ended scopes. No army of juniors. No slides as deliverable.

04 — The Grid Operating System

We diagnose.
Then we ship.

One methodology, two phases, ninety days end-to-end — every engagement. Boring on purpose: no frameworks for the sake of frameworks, no deck as a deliverable. We stay in the room until the work runs without us.

01

Diagnose

2 weeks

Before we scope, we listen. No generic templates. No precooked answers.

02

Prioritize

1 week

Ruthless focus on two to three interventions with clear expected ROI.

03

Ship

8 weeks

We stay in the room until the work is live in production, not just approved.

04

Hand over

1 week

Internal capability, not consultant dependence. We design ourselves out.

05 — How we choose

How we choose.
Criteria, not preferences.

Every stack choice we make is a decision against criteria you can audit. No vendor allegiances. No “preferred partner” badges. Here are the four criteria we apply on every engagement — so you can pre-qualify us before the call.

Criterion · cost per useful output

Picked per task, not per partnership.

Frontier models where reasoning quality earns the unit cost. Open-weight where privacy, EU residency, latency or margin demand it. We benchmark both on your real tasks and commit on cost per useful output. We never lock you into a model family we have economics in.

Criterion · failure-mode tolerance

Copilots first. Agents when failure is cheap.

Most mid-market value still unlocks with copilots — the model sits next to a human who keeps the keyboard. We move to agents only when the workflow is deterministic enough to wrap and the failure mode is recoverable. That order is almost always the right one.

Criterion · EU residency by default

Engineering constraint, not legal memo.

Paris-native by default. For regulated data flows, we run on EU-hosted inference with risk classification and logging designed in from week one — not retrofitted the week before audit. Sovereignty as a posture, not a slide.

Criterion · transparent economics

Vendor-agnostic by contract.

We pick tools per job, never per partnership. If we ever take referral economics from a vendor, it is disclosed in the engagement letter — line item, before signature. Buy-then-orchestrate by default; custom only where data, moat or unit economics earn it. Your team owns the code we leave.

06 — Team

Two operating partners.
No leverage on juniors.

Grid Advisory is led by two senior operating partners, backed by a trusted network of fractional experts pulled in for specific engagements. You work with the people you hired — not a rotating cast of associates.

Co-founder & Operating Partner
Sébastien
Van Hoof
Based in
Paris, France
Languages
French, Dutch, English
Serving
France, Benelux, UK, DACH
Education
Solvay · HEC Montréal
LinkedIn
linkedin.com/in/sebvh

Ten-plus years operating at the intersection of commercial strategy and frontline execution. BCG strategy consultant and active project leader across Europe and Asia — including a year at BCG Digital Ventures building companies from zero to one. Google EMEA product marketing lead for Grow with Google, training fifteen million people and businesses across the region.

Co-founder, CRO and COO of Sprynt, a digital asset payments company (bootstrapped, seven hires, thirty-plus customer agreements). Engagement Manager at BNP Paribas CIB Consulting & Transformation, advising C-suite on growth strategy and operational improvement. Most recently Head of Revenue & AI Operations and Interim VP of Operations at Spendesk, leading twenty-five-plus people across RevOps, AI ops, customer operations and support.

The through-line is the same: take commercial strategy from PowerPoint into production — and stay long enough for it to stick.


Co-founder & Operating Partner
Victoria
Vergnaud
Based in
Paris, France
Languages
French, English
Serving
France, UK, DACH, Iberia
Education
EM Lyon
LinkedIn
linkedin.com/in/victoriavergnaud

Ten years operating at the intersection of strategy and execution. Transformation consultant at PwC Paris, advising CAC 40 groups — including Eutelsat, Sage, Chanel, and LVMH — on large-scale change programs across Europe.

Seven years at Spendesk, a €50M ARR B2B SaaS scale-up serving SMB and mid-market (~300 people across France, the UK, Germany and Spain), including three years on the Executive Committee. Exposure across go-to-market, tech, product and people — supporting the company from 150 to 500 employees through hypergrowth, professionalisation, restructuring, and change management. A 360° view, from the inside, of what it takes to make a scaling organisation actually work.

The through-line is the same: partner with founders and investors to turn strategy into measurable execution — organisational design, performance, operational excellence — and stay long enough for the change to stick.

07 — Questions

Answers that
usually come up.

If one of these is still unclear after reading, it is probably worth a thirty-minute call.

What do your engagements cost?

The Diagnostic is fixed-fee, typically €60–120k depending on company complexity. The Implementation that follows is fixed-fee, typically €150–280k per Phase-1 use case shipped to production. The Operating Partner retainer starts at €20k per month for two days per week, six-month minimum. Every engagement is priced to fit the problem — book an intro call and we will give you a straight answer within twenty-four hours.

How do we pick between Diagnostic, Implementation and Operating Partner?

Start with the Diagnostic if you need a defensible plan before committing budget. Implementation follows the Diagnostic — same team, same playbook, the bets it scoped. Operating Partner is the retainer for compounding once Phase 1 is shipped. We don't sell them out of order.

Who owns the IP we build together?

You do. All code, prompts, playbooks and documentation belong to the client. Anonymised scores feed our cross-portfolio AI-Readiness benchmark — we retain perpetual rights to that aggregated dataset; you retain exclusive rights to anything that identifies your business.

Do you take referral economics from AI vendors?

If a recommendation we make to you carries referral economics from the vendor, it appears as a line item in your engagement letter before you sign. You accept those economics with full visibility, or we recommend something else and skip them. We never hide them. We never let them choose the tool.

How do you handle data and the EU AI Act?

Paris-native by default, EU-hosted inference where required. We classify every use case against the AI Act before we build — not after audit. No client data ever goes to training. DPAs on request.

What if we're in a regulated industry (insurance, banking, healthcare)?

For ACPR-supervised, Solvency II, EBA, EIOPA or healthcare-regulated engagements where AI touches scoring, claims, underwriting or patient data — we refer to specialist partners with the right compliance posture and step back. For the rest of your operations, we can still ship. We won't pretend a SaaS playbook applies to a mutuelle's actuarial models.

How available are you after we sign?

Embedded. You get one of the two co-founders in your Slack and leadership meetings. No associates, no outsourced juniors — the only scale lever is our fractional expert network, pulled in by name for specific lifts.

What happens when an engagement ends?

A hand-over week. Documented runbooks, trained owners, and a thirty-day warranty where we patch anything that breaks. We explicitly design ourselves out — consultant dependence is a failure mode.

How are you different from Artefact, Sia, Capgemini Invent or a Big-3 AI practice?

Three things. Price: a BCG or McKinsey kickoff deck costs roughly what our entire Diagnostic costs. Pyramid: no associates — the people in your room have run revenue and AI operations inside a unicorn. Lane: Artefact and Capgemini Invent sell data-science capacity to your CDO; we sell operator judgement to your CFO and Operating Partner. Different buyer, different conversation.

What languages do you operate in?

French, English, Dutch (Sébastien) and French, English (Victoria). Delivery is in French or English by client preference; DACH and Iberian engagements run in English unless otherwise agreed.

Let's talk

Want to see how we ship?

A thirty-minute intro call. No slides, no pitch. We will tell you honestly whether we are the right fit — and who to call if we are not.