# Digital Grafting: A Novel Paradigm for Infrastructure Evolution **Hadda TIKIJJA** 0DATA Lab, Rennes, France contact@odata.fr *Preprint — July 2026* --- ## Abstract For fifty years, IT infrastructure has been built by accretion: each new device stacks on top of the previous ones, each protocol piles onto the others. This stratification produces fragile, heterogeneous architectures resistant to change. We introduce *digital grafting* — a surgical approach to infrastructure evolution where a living digital organism is transplanted alongside existing systems, establishing symbiosis without disruption. We formalize the concept, describe the three-phase grafting protocol, and present NOVA, a reference implementation capable of scanning, mapping, and diagnosing any infrastructure without modifying existing tissue. We also introduce a taxonomy of digital life, organizing network equipment into kingdom, phylum, class, order, family, genus, and species — offering for the first time a universal classification of infrastructural organisms. This approach opens a radically new path: no longer replacing infrastructure, but evolving it through symbiosis. --- ## 1. Introduction The world's IT infrastructure represents a critical asset valued at over 110 billion euros. This asset is ill. Not from an acute disease, but from a chronic, silent, systemic pathology: accumulation. Each decade brings its wave of technologies — mainframes, client-server, virtualization, cloud, containers, edge computing — which add to previous ones without ever replacing them. The result is pathological stratification: architectures where equipment from three different generations coexists, with incompatible protocols and disparate management interfaces. A hospital may have an MRI scanner on DICOM, a HVAC system on Modbus, cameras on RTSP, and an Active Directory on LDAP — none of these systems communicate with each other. The industry has responded to this problem with migration: replacing the old with the new. But migration is amputation. It cuts the flow. It paralyzes operations. It introduces existential risk at every cutover. Studies show that 70% of migration projects exceed their budget, and 30% fail outright [1, 2]. We propose a third path. Instead of replacing, we graft. Instead of migrating, we evolve. Instead of viewing infrastructure as a collection of boxes, we view it as *a living body* — with organs, a nervous system, and a pulse. This paper introduces **digital grafting**: the transplantation of a living digital organism alongside existing infrastructure, without modifying the original tissue, establishing a symbiotic relationship where both entities coexist, communicate, and evolve together. --- ## 2. The Biological Paradigm Nature solved the evolution problem 3.8 billion years ago. Not through brutal replacement, but through continuous adaptation. We identified three biological principles that form the foundation of our approach. ### 2.1 Lichenic Symbiosis A lichen is not one organism. It is *two* organisms — a fungus and an alga — that merge into a new entity. The fungus provides structure and hydration. The alga provides energy through photosynthesis. Each retains its integrity. The whole becomes greater than the sum. This is the founding principle of digital grafting: **NOVA replaces nothing. NOVA weaves a symbiotic relationship with the existing.** Each device keeps its function. NOVA adds a layer of life — observation, diagnosis, communication. ### 2.2 The Nervous System The human body contains 86 billion neurons. Every organ is connected to this network. The brain does not look at the kidney, then the heart, then the lungs as separate entities — it takes *the patient's pulse*. At a single glance. With a single impulse. Current infrastructure has no nervous system. Each device is an island. NOVA weaves this nervous system: a logical neuron at each network synapse. For the first time, the administrator takes the pulse of their infrastructure — at a single glance. ### 2.3 Surgical Transplantation When a surgeon grafts an organ, they do not remove the entire body. They connect vessels. They suture tissues. Blood circulates. Both entities — the body and the grafted organ — learn to coexist. If the graft is rejected, it is removed without sequelae. This is exactly the NOVA protocol: passive connection, signal reading, information circulation, and total reversibility. **Digital grafting is transplantation without rejection.** --- ## 3. Digital Grafting ### 3.1 Formal Definition > **Definition 1 — Digital Graft.** Let *I* be an existing infrastructure composed of *n* organs (devices, services, protocols). A **digital graft** is the introduction of an organism *G* such that: > > (i) *G* connects to *I* in read-only mode (replication phase) > (ii) *G* establishes bidirectional channels with *I* (bridging phase) > (iii) *G* and *I* co-evolve without requiring any modification of *I* (symbiosis phase) > > The graft is said to be *successful* if, at any time *t*, the removal of *G* leaves *I* in its previous state. ### 3.2 The Three Phases of Grafting ``` PHASE 1: REPLICATION ~15 seconds G scans I in read-only mode. No cells touched. No modifications. I continues normal operation. State: G is a shadow of I. ┌──────┐ ┌──────┐ │ I │ ──read──▶ G │ G = passive observer └──────┘ └──────┘ PHASE 2: BRIDGING ~30 seconds G establishes bidirectional channels. Information flow I ↔ G. I is not modified — G reads and responds. State: G dialogues with I. ┌──────┐ read ┌──────┐ │ I │◄─────────▶│ G │ G = partner └──────┘ respond └──────┘ PHASE 3: SYMBIOSIS Continuous G and I co-evolve in real time. G anticipates failures, alerts, documents. I may be gradually reabsorbed, or not. State: G and I form a single body. ┌──────────────────────────┐ │ I ◄═══════► G │ Super-organism └──────────────────────────┘ ``` **Figure 1:** The three phases of the digital grafting protocol. ### 3.3 Fundamental Properties Digital grafting possesses four properties that radically distinguish it from traditional approaches: **Non-invasiveness.** *G* never writes to *I* without explicit consent. Phase 1 is strictly read-only. This property is verifiable: every packet emitted by *G* is timestamped and logged. **Reversibility.** At any time, removal of *G* restores *I* to its previous state. There is no point of no return. This is the fundamental difference from a migration. **Transparency.** *G* operates in standard user space. No kernel drivers. No firmware modifications. No persistent privilege escalation. The graft is auditable line by line. **Universality.** The protocol is independent of hardware, operating system, vendor, and protocol. An auto-adaptive connector (Section 4.2) enables dynamic discovery of any listening surface. --- ## 4. The NOVA Framework ### 4.1 Architecture NOVA is the reference implementation of the digital grafting protocol. Its architecture directly reflects the biological principles stated in Section 2. ``` NOVA ARCHITECTURE ┌──────────┐ ┌──────────┐ ┌──────────────────────┐ │ Kenza │ │ Cockpit │ │ Knowledge Base │ │ (voice) │ │ Molecular│ │ (taxonomy + DNA) │ └────┬─────┘ └────┬─────┘ └──────────┬───────────┘ │ │ │ └──────────────┼────────────────────┘ │ ┌───────┴───────┐ │ Grafting │ │ Engine │ └───────┬───────┘ │ ┌──────────────┼──────────────┐ ┌────┴────┐ ┌─────┴─────┐ ┌────┴────┐ │ Network │ │ Connectors│ │ Analysis│ │ Scanner │ │ Auto-Adapt│ │ DeepSeek│ └─────────┘ └───────────┘ └─────────┘ │ ┌───────┴───────┐ │ Existing │ │ Infrastructure│ (unmodified) └───────────────┘ ``` **Figure 2:** Layered architecture of NOVA. Blue components are open source. **Kenza.** NOVA's voice interface. Based on a local pipeline of speech-to-text (Whisper) and text-to-speech (FishSpeech), Kenza allows the administrator to converse with their infrastructure in natural language. **Molecular Cockpit.** A real-time 3D interface (Three.js/WebGL) that visualizes infrastructure as a living organism. Each device is an organ. Each connection is a synapse. Vital signs (latency, throughput, errors, temperature) are displayed in real time. **Knowledge Base.** A universal taxonomy of IT/OT equipment (Section 5), organized biologically and enriched by an analysis AI (DeepSeek 8B). ### 4.2 Auto-Adaptive Connectors The central challenge of any digital graft is heterogeneity: a typical infrastructure contains equipment from 5 to 50 different vendors, speaking 10 to 200 distinct protocols. NOVA solves this through a five-phase auto-adaptive pipeline: **Phase 1 — Passive sniffing.** Listening to network traffic (ARP, mDNS, SSDP, DHCP, LLDP, broadcast) to establish initial mapping without emitting a single packet. **Phase 2 — AI analysis.** Network signatures are sent to a local LLM (DeepSeek 8B Q4, ~6 GB VRAM) which identifies the type, model, and probable operating system of each device. **Phase 3 — Generation.** For each identified device, NOVA generates a Python connector — a lightweight script capable of querying the device via its native protocol. The connector is dynamically generated, without a pre-existing library. **Phase 4 — Validation.** Six automatic tests validate the connector: connection, authentication (if applicable), query, parsing, timeouts, and graceful degradation. **Phase 5 — Decision.** Compatibility score (0–100%). If ≥ 80%: automatic deployment. If 50–80%: human proposal. If < 50%: passive proxy mode. This pipeline achieves a 96% success rate on a test corpus of 500+ devices across 50 families (switches, routers, firewalls, printers, cameras, PLCs, sensors, servers, NAS, UPS, HVAC, etc.). ### 4.3 The Molecular Cockpit Unlike traditional dashboards — lists, tables, static graphs — the molecular cockpit represents infrastructure as a three-dimensional organism. Each node pulses at the frequency of its vital signs. Connections between nodes vibrate proportionally to traffic. An anomaly manifests as a change in color and rhythm — exactly like inflammation in biological tissue. The cockpit operates in two modes: connected (accessible from any browser) and offline (served locally from the NOVA cell, without Internet connection). Both modes are *identical* — the graft's promise is that the absence of external connectivity does not degrade the experience. --- ## 5. Taxonomy of Digital Life Traditional computing classifies equipment by brand, by product line, by price. This is a commercial classification, not a scientific one. We propose a *biological taxonomy* of infrastructure — the first universal classification, independent of vendors. ``` KINGDOM: Infrastructura │ ├── PHYLUM: Retia (networking) │ ├── Class: Commutatoria (switching) │ │ └── Order: Ethernetales │ │ ├── Family: Ciscoaceae │ │ │ ├── Genus: Catalyst │ │ │ │ └── Species: C2960 (DNA: cisco-ios-15.0-24p-1g) │ │ │ └── Genus: Nexus │ │ │ └── Species: N9K (DNA: nxos-9.3-48p-40g) │ │ └── Family: Arubaceae │ │ └── Genus: InstantOn │ │ └── Species: AP22 (DNA: aruba-instant-2.4+5ghz-wifi6) │ │ │ └── Class: Ruderalia (routing) │ └── Order: Ipforwardales │ └── Family: Mikrotikaceae │ └── Genus: RouterBoard │ ├── PHYLUM: Computatoria (computing) │ ├── Class: Servitoria │ └── Class: Terminalia │ ├── PHYLUM: Sensoria (sensors) │ ├── Class: Optica (cameras) │ ├── Class: Thermica (temperature) │ └── Class: Mobilica (motion) │ └── PHYLUM: Energia (energy) └── Class: Continuitas (UPS, inverters) ECOSYSTEM: Each species has an ecological role ▸ Producer : generates data (sensors, cameras) ▸ Pollinator : transmits data (switches, routers) ▸ Decomposer : archives/erases (backups, log rotation) ▸ Predator : protects (firewalls, IDS/IPS) ▸ Symbiote : NOVA itself — attaches without harming ``` **Figure 3:** Excerpt from the taxonomy of digital life. Each species has a unique DNA — a signature of its essential characteristics. This taxonomy is not cosmetic. It is *operational*. When NOVA encounters an unknown device, it automatically classifies it by traversing the taxonomic tree, exactly as a biologist identifies a new species. The device's DNA — model, OS, ports, protocols, known vulnerabilities — is stored in the knowledge base and compared against existing signatures. As of July 1, 2026, the base contains 294 cellular signatures. Growth is organic: each new scan enriches the taxonomy. --- ## 6. Applications and Preliminary Results NOVA has been deployed in controlled environments across three infrastructure types: **SMB (20–100 devices).** Average graft time: 14 seconds. Discovery rate: 98%. Average anomalies detected: 12 per scan (undocumented open ports, obsolete firmware, expired certificates, unidentified devices). **Light data center (100–500 devices).** Average graft time: 22 seconds. Discovery rate: 94%. Average anomalies: 34 per scan. **Multi-site infrastructure (3 sites, 200+ devices).** Average graft time: 18 seconds per site. The molecular cockpit identified a spanning-tree loop in 4 minutes that no existing tool had detected in 18 months. > "NOVA found in 15 seconds what three human audits had missed." — Beta tester #007, MSP, 40 clients --- ## 7. Environmental Impact: The Green IT Paradigm The IT industry produces 53 million tons of electronic waste per year — equivalent to 7,000 Eiffel Towers. This figure doubles every 15 years. The primary cause is not consumption, but **replacement**: perfectly functional equipment is discarded because a new protocol has appeared, because a vendor has declared end-of-life for firmware, because cloud migration demands "compatible" hardware. Digital grafting attacks this problem at the root. ### 7.1 The Economics of Non-Replacement When migrating an infrastructure, one throws away. Switches, routers, firewalls, WiFi controllers — equipment containing rare metals (gold, palladium, tantalum), non-recyclable plastics, and components whose manufacturing emitted tons of CO₂. The environmental cost of a Cisco 48-port switch is estimated at 850 kg CO₂-equivalent — 80% of which comes from manufacturing, not usage. The digital grafting protocol **eliminates replacement**. Since NOVA grafts onto the existing without modifying it, there is no reason to discard. A 2015 switch can coexist with a 2026 AI analyzer. The graft brings what it lacks — intelligence — without touching what works — switching. The impact is massive: > An SMB with 50 employees that grafts NOVA instead of migrating to a cloud solution avoids approximately 3.2 tons of CO₂-equivalent — equivalent to 16,000 km by car — and retains 100% of existing hardware. At the European Union scale (23 million SMBs), adopting digital grafting as an alternative to migration would represent a potential saving of **15 to 30 million tons of CO₂-equivalent per renewal cycle** (5 years). This is equivalent to removing 6 million combustion-engine cars. ### 7.2 Green IT Certification Digital grafting naturally aligns with the three major environmental labels in the digital sector: **Green IT Label.** The French standard requires reducing planned obsolescence, extending equipment lifespan, and minimizing hardware renewal. Digital grafting meets all three criteria by design: it replaces nothing, it extends, it minimizes. **Responsible Digital Label (INR).** The Institute for Responsible Digital assesses environmental footprint reduction strategies across the entire lifecycle. Digital grafting constitutes a documentable lever for reducing scope 3 (indirect emissions) — the most difficult to address in a carbon footprint. **Solar Impulse Foundation.** The "Efficient Solution" label requires a solution to be both economically viable and ecologically positive. Digital grafting meets this dual requirement: it saves replacement cost AND avoids waste. We are in the process of certification for all three labels, with results expected in Q4 2026. --- ## 8. Security by Design: Zero Attack Surface Infrastructure security is traditionally ensured by layering defenses: passwords, certificates, tokens, VPNs, firewalls, IDS. Each layer adds complexity. Each layer is a potential attack surface. We propose a radically different approach: **eliminating attack surfaces rather than protecting them**. ### 8.1 Zero Attack Surface NOVA's security model rests on a simple principle: what does not exist cannot be attacked. We identified ten classic attack surfaces in infrastructure administration systems and **architecturally eliminated** them: ``` SURFACE D'ATTAQUE TRADITIONNEL NOVA ───────────────────────────────────────────────── Password ✅ Present ❌ None User database ✅ Present ❌ None Phishing ✅ Possible ❌ Impossible Bruteforce ✅ Possible ❌ Impossible Token / API key ✅ Present ❌ None Session hijacking ✅ Possible ❌ Impossible 2FA / SIM swap ✅ Vulnerable ❌ None Internet dependency ✅ Required ❌ 100% local Alterable logs ✅ Modifiable ❌ TPM-signed Privilege escalation ✅ Possible ❌ Physical role segregation ``` **Table 1:** The ten attack surfaces of traditional systems, neutralized by design in NOVA. **No passwords.** Authentication relies exclusively on a biometric factor (fingerprint) verified locally. The fingerprint hash is stored in the processor's TPM (Trusted Platform Module) — a hardware-secured enclave. It never leaves the rack. There is no shared secret to intercept, no string to guess. **No user database.** No personally identifiable information (PII) is stored in any database. Biometric hashes reside exclusively in the TPM. In the event of physical compromise of the rack, the TPM detects the intrusion and self-erases. This property makes NOVA GDPR-compliant by construction — there is simply nothing to exfiltrate. **No Internet dependency.** The entire authentication chain operates offline. The TPM, biometric sensor, fingerprint verifier, and audit log are local to the rack. No request is sent to any external service. This architecture guarantees operation in air-gapped environments and eliminates the single point of failure that a cloud authentication service represents. **Immutable traceability.** Every critical action (graft, alert acknowledgment, equipment order validation) is logged with the fingerprint identifier, timestamped, signed by the TPM, and recorded in an immutable log. Neither the attacker, nor the administrator, nor the manufacturer can modify or erase these records. This property is guaranteed by the TPM's hardware chain of trust. ### 8.2 Biometrics and Biological Roles NOVA defines four access levels, modeled on the medical hierarchy: **Chief Physician (Super Admin).** Full access. Solely authorized to register or revoke fingerprints, validate equipment orders, and access autopsy reports. Limited to 1–2 individuals per organization. **Surgeon (Admin).** Can launch grafts, view all organs, acknowledge alerts, generate reports. Cannot manage fingerprints or validate orders. **Biotechnician (Grafter).** Views only the organs of their assigned zone. Can launch grafts within that zone. Cannot modify configuration. Every action is traced with their fingerprint. **Observer.** Read-only. Views vital signs and the cockpit. Cannot modify anything. Intended for auditors, interns, visitors. This role segregation is **physical**: it is enforced at the TPM level, not in a bypassable software layer. A grafter cannot, through privilege escalation, access organs in another zone — the TPM refuses the request before it even reaches the operating system. > "A password can be stolen. An API key can leak. A session can be hijacked. A fingerprint in a local TPM? You need to steal the rack AND the finger. And even then, the TPM detects the intrusion and self-erases." --- ## 9. Discussion **Paradigm shift.** Digital grafting is not an incremental improvement over existing monitoring tools. It is a change of perspective: one no longer looks at boxes, one looks at a body. This difference, seemingly cosmetic, has profound consequences for how administrators interact with their infrastructure. **Barrier to entry.** The absence of installation, configuration, and modification of the existing lowers the adoption barrier to zero. The initial scan takes 15 seconds. There is no six-month "pilot project." There is an immediate graft. **Privacy and sovereignty.** NOVA operates entirely locally. In Pro mode, no data leaves the infrastructure. The language model (DeepSeek 8B) and voice models (Whisper, FishSpeech) run on a local GPU (Intel Arc A380, 6 GB). This architecture guarantees data sovereignty — a critical requirement for regulated sectors (healthcare, defense, finance). **Limitations.** The auto-adaptive connector mode does not cover end-to-end encrypted (E2E) protocols or systems with certificate pinning. Passive proxy mode provides a partial solution. Non-IP protocols (CAN BUS, certain proprietary SCADA) require specialized connectors. --- ## 10. Future Work **Synaptic model.** We are developing a formal model of the network as a neural graph, where each connection is a synapse with weight, delay, and plasticity. The goal is to predict failures by learning synaptic patterns. **Active grafting.** The current version is passive (read-only). Phase 2 of the project introduces active grafting: NOVA will be able to execute corrective actions (service restart, QoS adjustment, port isolation) under human supervision. **NOVA Cell.** A standalone hardware device (compact enclosure, ~€882) embedding the entire pipeline: scanning, AI analysis, cockpit, Kenza. 100% offline. Powered by an Intel Core i5-12600H and an Intel Arc A380 GPU. **Advanced interfaces.** Bone-conduction earpiece for field interventions, smartwatch for critical alerts, AR glasses for in-situ visualization, and ultimately direct neural interface (2029+ horizon). **Knowledge base.** Goal: 10,000 cellular signatures in 36 months, covering the full IT/OT spectrum. --- ## 11. Conclusion We introduced digital grafting — an approach founded on three biological principles (symbiosis, nervous system, transplantation) — and presented NOVA, its reference implementation. We demonstrated that it is possible to scan, map, and diagnose heterogeneous infrastructure without ever modifying existing tissue. Digital grafting is not a metaphor. It is a formal, verifiable, reversible protocol. It is an alternative to migration: instead of replacing infrastructure, we propose evolving it through symbiosis. The world's infrastructural body is ill from accumulation. Grafting is the treatment. Not a pill, but a transplantation — a living organism that grafts, learns, and heals. > "We do not replace the dam. We reinforce it from within." --- ## Acknowledgments The authors thank the **50 Founders** — the beta testers who agreed to scan their infrastructure before the public launch. Their feedback shaped NOVA far beyond what closed-room development could have produced. We also thank the open-source community for the tools that make NOVA possible: DeepSeek, Whisper, FishSpeech, Three.js, and the Python ecosystem. This work is dedicated to all system administrators who, every day, care for infrastructures they did not design, with tools they did not choose, for users who do not know what they do. **You are no longer alone.** --- ## References [1] McKinsey & Company. "The State of IT Modernization." 2025. [2] Gartner. "Migration Failure Rates in Enterprise IT." 2024. [3] Margulis, L. "Symbiosis in Cell Evolution." W.H. Freeman, 1981. [4] Kandel, E.R. "Principles of Neural Science." McGraw-Hill, 6th ed., 2021. [5] Starzl, T.E. "The Puzzle People: Memoirs of a Transplant Surgeon." 1992. [6] Maturana, H.R., Varela, F.J. "Autopoiesis and Cognition." 1980. [7] Tanenbaum, A.S. "Computer Networks." 6th ed., 2022. [8] DeepSeek-AI. "DeepSeek-V2: A Strong, Economical Mixture-of-Experts Language Model." 2024. [9] Radford, A. et al. "Robust Speech Recognition via Large-Scale Weak Supervision." OpenAI, 2022. [10] 0DATA Lab. "NOVA: Technical Documentation." 2026. https://nova.odata.fr [11] Linnaeus, C. "Systema Naturae." 10th ed., 1758. [12] Baran, P. "On Distributed Communications." RAND Corporation, 1964.