The Edge
Intelligence
Grid.
Divide the payload. Aid the edge.
We transform Cisco Catalyst 9300 Series Switches and CW917X Access Points into a sovereign, decentralized AI execution farm via native Docker containers. Leveraging asymmetric pipeline parallelism to execute high-density AI workloads with zero dedicated GPUs.
Absolute Data Sovereignty.
Consider a financial institution processing highly sensitive PII. Cloud-based LLMs leak context via external APIs. Divaid keeps the prompt entirely localized. The inference swarm is physically sandboxed within the walls of the branch.
Cisco IOx & Native Docker Execution
Catalyst 9300 series and modern edge APs possess robust ARM64 compute capacity. Cisco IOx allows native Docker containers to run directly on the switch hardware, but restricts memory footprints via strict cgroups. Running colossal AI architectures natively in this sandbox was deemed impossible. Until now.
GGUF Encoding
We subject state-of-the-art dense transformer models to aggressive INT4 quantization. By dropping precision on the weight matrices, the payload is algorithmically compressed, fitting inside the Docker image without perplexity degradation.
The DivGraph™ Algorithm
Our proprietary DivGraph Distribution Algorithm resides in the C9300 Orchestrator container. When a heavy AI task hits the network, DivGraph shatters the underlying neural network layers into mini-tasks. Rather than passing parameter weights, it distributes fractional matrix multiplication tasks across the Docker nodes running on your AP swarm.
C9300 Merge
Once the AP nodes complete their fraction of the calculation, the tensors return to the C9300. The Orchestrator acts as the ultimate verifier, recombining the mini-tasks, decoding the hidden states, and returning the definitive final AI output with mathematical certainty.
The 1,000-Node Compute Matrix.
Scale changes everything. Imagine a corporate campus with 100 Catalyst Switches and 1,000 Access Points. When a colossal Generative AI task hits the network, the C9300 Orchestrator slices the mathematical workload into thousands of shard-groups (represented by colors) and routes them across the mesh.
Zero Wi-Fi Disruption: Thanks to strict IOx cgroup isolation, the Access Points seamlessly process massive background AI inferences while continuously serving high-performance Wi-Fi to connected laptops and mobile devices.
Real-World Flow: Heavy Workload Orchestration.
Imagine processing a massive, confidential 500-page Financial Audit through an advanced LLM to detect compliance anomalies. Sending this to an external API violates banking sovereignty.
Instead, the prompt hits the Catalyst 9300 Orchestrator. The DivGraph Algorithm tokenizes the document and shatters the heavy attention-matrix calculation into 50 mini-tasks.
These micro-payloads are pushed instantly to 50 local CW917X Access Points running lightweight Docker workers. They execute their fraction of the neural network layer simultaneously. The returning tensors are instantly verified and recomposed by the C9300, delivering a comprehensive anomaly report locally, securely, and in seconds.
1. Heavy Ingestion
C9300 receives the 500-page Audit (PDF).
2. DivGraph Matrix Split
Algorithm divides the mathematical workload into 50 micro-tensor sequences.
3. Docker Swarm Distribution
Micro-tasks distributed via IOx to 50 AP nodes. Mini-inferences executed in parallel.
4. C9300 Verification & Recomposition
Orchestrator receives returning tensors, verifies hash integrity, and merges outputs.
5. Final Output Yield
Local Zero-Trust compliance report generated.
Hardware-Level Isolation.
How do we run Generative AI on an Access Point without destroying Wi-Fi performance? By leveraging the Cisco IOx architecture to create a strict, hardware-enforced sandbox.
90% Dedicated to Wi-Fi 7
The vast majority of the AP's CPU, RAM, and NPU are completely untouched. Core network services have absolute priority at the hypervisor level.
10% divaid.AI Sandbox
We deploy the DDA Worker node via Docker cgroups. It is mathematically bounded to never exceed 100MB of RAM or interrupt network thread execution.
Micro-Orchestration on Catalyst.
The Catalyst 9000 series switches act as the central brain. Instead of routing standard packets, our lightweight IOx app routes tensor states.
Untouched ASIC Performance
The UADP ASIC continues to switch enterprise traffic at line rate. divaid.AI does not interact with or degrade the data plane hardware switching.
Lightweight Master Node
A tiny fraction of the switch's application hosting CPU handles the DDA algorithm, keeping track of which AP is processing which neural layer.