IP, finance, legal, healthcare, and regulated operations need AI workflows without sending private data to cloud AI APIs.
Pitch deck narrative
A slide story for operable local AI.
The deck frames the company around a blocked buyer, a narrow current proof base, an Alpha 1 product layer, and benchmark evidence as the path to trust.
Slide-like narrative
Each slide answers one investor question.
Chat, approved documents, workflows, model/runtime controls, diagnostics, offline updates, and benchmark proof in one deployable stack.
Ollama + OpenWebUI baseline, single-node Docker Compose deployment, and no public claim of customer production proof.
Private knowledge retrieval, workflow automation, model/runtime controls, RBAC + audit logs, operator diagnostics, and offline updates.
- X: at least 80% pass rate on fixed domain evals.
- Y: N=3 users with p95 text and image-style workload targets.
- Baseline:
gpt-4o-minion non-sensitive text/domain inputs.
Benchmark report v1 should include model/runtime versions, hardware profile, workload profiles, and pass-rate/latency distributions.
Agents, voice, observability, extension packaging, richer media workflows, and hardware-aware profiles are roadmap only until validated.
Use of funds: Alpha 1, benchmark harness, 2 design-partner pilots, first paid deployment, and initial compliance/ops assets.
Demo evidence status
Do not fill the deck with fake product screenshots.
Public media should be captured only from a sanitized Alpha 1 flow. Until then, the deck should state the proof boundary and show the product-layer map instead of pretending the evidence exists.
Lab baseline, documented Compose deployment, and target Alpha 1 workflows.
Install, approved-document workflow, support chatbot, operator diagnostics, and benchmark harness.
No private prompts, customer data, browser sessions, secrets, or unverified benchmark results.
Latest version
Request the current investor deck.
The public route is the story spine; shareable deck media should wait for sanitized Alpha 1 evidence.