Claude peut rendre riche n'importe qui, voici comment (conférence à la Sorbonne)

Company / Business Model Overview (Zenitia)

• Company: Zenitia AI, co-founded ~3 years prior to the talk by Clément and Enzo Donati.
• Business model: An AI automation agency focused on high-margin delivery and scalable process design.
• Revenue milestone: They “recently exceeded one million euros in revenue generated” (also described as “a little more than one million”).
• People model / cost structure:
  • They claim to achieve scale without “10 or 15 people” on permanent contracts.
  • Gross margin target: “60–80% gross margin.”
• Vision / thesis:
  • Use automation + systematization to gain freedom, improve margins, and make decisions faster.
  • Treat the business side (marketing/sales) as primary—not an afterthought.

---

Frameworks / Playbooks / Operating Principles

B2B AI Automation “3-step offering”

1. AI audit
   • Identify what can and cannot be automated.
2. Training
   • Offered in in-person and online formats.
3. Build AI systems/workflows
   • Prefer deterministic automation over unreliable autonomous agents.

Deterministic vs agentic automation (reliability playbook)

• Use deterministic workflows for reliability-critical processes (claims need around ~99.7% reliability).
• Use “agents” only when autonomy levels still meet required reliability.

“Don’t start with tools—start with market”

• Start with positioning, value proposition, and acquisition channels.
• Then select the appropriate tooling/workflows.

Sales-first sequencing

• “Sell first, then build”
  • Avoid 3–6 months of building without validated demand.
• Use V1/prototypes to de-risk delivery before full implementation.

---

Key Products / Services (Actionable Examples)

1) AI Audit (B2B)

What it does

• Interviews stakeholders across departments (e.g., marketing, billing, HR, sales).
• Maps day-to-day tasks → identifies automatable vs non-automatable processes.
• Produces a report including:
  • Problem description
  • Proposed solution(s)
  • Estimated gains (e.g., time saved → redirect to higher-value work)

Typical deliverable economics

• Price range: “€2,000 to €10,000” (also cited repeatedly as €3,000–€10,000)
• Time investment: stakeholder interviews + report + restitution
  • Later converted into demos/prototypes

Concrete KPI example

• Estimates like: “recover 200 hours per year” (time saved from automated tasks).

Operational recommendation

• The audit is the market-facing wedge: once the automation roadmap is presented, Zenitia transitions into delivery or prototypes.

---

2) Training (Automation Literacy)

Two formats

• In-person (physical) training

  • Lower technical complexity; mainly awareness and change readiness.
  • Price: “€1,000–€3,000 per day”
  • Scaling constraint: physical delivery can’t easily be multiplied via hiring.

• Online training

  • Higher scalability and profitability (because the product exists).
  • Delivered via platform/ecosystem/community (mentions ~380 members).

Strategic positioning

• Training is valuable, but the agency value ultimately comes from built systems, not just education.

---

3) Built AI Systems / Workflows (Core)

Target processes

Digitized, repetitive “tertiary sector” workflows such as:

• data retrieval
• templating
• prompting
• report generation
• invoice/proposal creation
• document processing

Tooling / stack (as described)

• N8N for low-code workflow automation (nodes/“knots”).

• NocoDB as an internal database alternative to Airtable (they claim lower cost + self-hosting).

• Document + e-sign components (mentions electronic signatures such as DocuSeal/Docu…)

• Vision models (e.g., Google Gemini) for document extraction.

Deterministic design approach

• Chain steps with controlled reliability.
• Emphasize reliability over fully autonomous agents.

---

Concrete Case Studies & Measurable Impacts

A) Automating Commercial Proposals (Sales Ops)

Problem

• Proposals are templated but still require heavy manual tailoring.
• Manual effort: 4–7 hours per proposal.

Their system

• A button triggers:
  • call data retrieval
  • generation of a customized proposal document
  • ROI/diagrams
  • quote details (incl./excl. tax)
  • tailored email + signature insertion
• Output time: ~5–10 minutes instead of 5–7 hours.

Business outcomes / KPIs claimed

• Responsiveness KPI: send proposals in <24 hours
  • contrasted with “90% of companies” sending in a week or 10 days
• Conversion uplift: “increase conversion rates by 40–60%”
• Economic example (speaker estimate):
  • quotes of €5k–€20k
  • customers gain “€10k–€15k up to €50k–€100k per month extra”
• Social prospecting KPI:
  • 30–60% higher conversion when sending next-day vs one week later
• Speed-to-reply KPI:
  • if responding in <3 minutes on social platforms → ~400% higher chance of reply/discussion
  • they mention handling hundreds of conversations concurrently; speed drives booked calls

Actionable sales operations recommendation

• Use proposal turnaround optimization as a growth lever:
  • reduce time-to-proposal
  • increase number of conversations handled
  • align sales process speed with lead intent

---

B) Management Control for Restaurants (Cost + Margin Visibility)

Problem

• A restaurant with many invoices/suppliers couldn’t track:
  • unit economics by menu item
  • true profit margins (cost-to-serve vs prices)

Their system

• Ingest supplier invoices (Drive/email).
• Auto-extract invoice line items → standardize into a structured sheet.
• Generate monthly summaries and supplier breakdowns.
• Uses computer vision extraction (they claim very high reliability).

Claims / outcomes

• Thousands of invoice lines processed monthly.
• Reliability claim: “100% reliability” (with verification controls).
• Outcome:
  • immediate visibility into margin impacts when commodity/supplier prices change
  • renegotiation triggers (example: discovering “€10k–€15k more” costs)

Recommendation

• Use automation for margin control to prevent months-delayed loss accumulation.

---

Pricing & Delivery Mechanics (Operations / GTM Economics)

Project pricing ranges

• Custom automation/system fees: €1,000 to €50,000 (they typically prefer ≥€5,000)

Maintenance / recurring options

• Standard maintenance: 5–20% of setup cost per year
• Typical monthly maintenance for exploitation: “€800–€2,000/month”
• Operations license for high-frequency processes
  • paid monthly
  • positioned as strategic for ongoing business value and recurring revenue

Exploitation / maintenance models mentioned

• Recurring maintenance percentage (updates, bug fixes)
• License for operational dependency (systems used every day)
• Full outsourced maintenance to trusted subcontractors if needed

---

Delivery Capacity & Scaling Solution

Key bottleneck

Even with fast automation, projects require back-and-forth and waiting for client input. Delivery can range from 1–6 months depending on complexity (not 2 days).

Scaling approach

• Build an ecosystem of subcontractors (around 5–7, depending on periods).
• Train + document internally to keep subcontractors reliable.
• Recruit via their ecosystem rather than only external platforms.
• Reliability training reduces micromanagement and preserves delivery speed.

---

Strategic Marketing & Acquisition Playbook (How They Sell)

“Creator economy” acquisition engine

• Acquisition heavily tied to content creation:
  • LinkedIn / YouTube / Instagram
• Claim: over 80% of acquisition comes from social networks.
• Content roles:
  • inbound lead generation
  • conference credibility
  • trust creation (social proof)

Webinars / funnels

• They mention webinar tunnels and registration flows.
• Claims include:
  • 3500 registrations in an evening tunnel
  • €150k–€200k in 10 days attributed to marketing/sales capability

Positioning advice (explicit)

• Don’t “tinker with gadgets” when new tools drop.
• Instead:
  • positioning + value proposition
  • choose acquisition channels aligned to strengths
  • run content/webinars
  • automate prospect conversations
  • pick niches based on fit with your acquisition system, not just “where the money is”

“Unfair advantage” logic

• Differentiation is not only technology:
  • expertise + reliability + references with large clients
  • deliver without failures → customers get peace of mind

---

Agent vs Deterministic Workflow Guidance (Business Risk Management)

• They warn that agents can fail unpredictably, which is unacceptable in financial/billing contexts.
• Near-absolute reliability is required (citing 99.7%; they state lower like 80% would be “catastrophic” for billing).
• Automation credibility depends on:
  • verification controls
  • robust extraction and reconciliation
  • draft/review steps where needed (e.g., generate proposals vs auto-send)

---

Investing / Markets (High-Level Only)

• AI automation demand is framed as long-lived:
  • “automation needs won’t disappear”
• Focus on durable use cases such as:
  • sales responsiveness
  • document processing
  • cost/margin control
  • admin workflow automation

---

KPIs and Targets Mentioned (Collected)

• Revenue / growth
  • “> €1M revenue generated” (recent)
• Gross margin
  • “60–80% gross margin”
• Sales proposal turnaround
  • Manual: 4–7 hours
  • Automated: ~10 minutes (also cited as 5–10 minutes)
  • Speed target: <24 hours after call
• Conversion rate uplift
  • With <24h turnaround: +40–60% conversion
  • Proposal sent next day vs week later: +30–60% conversion
  • Social response speed:
    • <3 minutes → ~400% chance to get reply/discussion
• Operational time savings
  • Audit example: recover 200 hours/year
  • Also mentioned: saves 1500–2000 hours/year in some cases
• Reliability requirement
  • Deterministic workflows aim for 99.7%
• Maintenance benchmarks
  • 5–20% annually of setup cost
• Acquisition claims
  • revenue impact estimates (e.g., customers gaining €10k–€100k/month extra)

---

Presenters / Sources (As Mentioned)

• Clément — speaker; co-founder/presenter of Zenitia
• Enzo Donati — co-founder; credited as technical CTO partner