Original engineering, shipped as something you can actually buy.
From medical imaging to AI infrastructure — patent-backed work, packaged into a product with a price.
Solipher Triage
Turns a 50,000-frame capsule-endoscopy study into a reviewable shortlist.
A single capsule-endoscopy study produces tens of thousands of images per patient, and reviewers can't manually check every frame. Solipher Triage sits between your imaging pipeline and your reviewers, deciding what gets reviewed first, what waits, and what needs a second opinion — under a fixed memory and storage budget.
Impact
In our own evaluation on a 47,161-frame clinical dataset, the choice of scoring model alone swung the flagged-for-review rate by more than 40 percentage points — proof that this decision layer, not just the classifier underneath it, determines what a reviewer actually sees.
Works with any per-frame scoring model you already run
Fixed, predictable memory and storage footprint regardless of study length
Full decision record for every frame, ready for clinical review workflows
Integrates into existing PACS/viewer pipelines
Pilot
Custom
For a single site validating on retrospective studies.
- Up to 500 studies / month
- Retrospective validation support
- Email support
Clinical
PopularCustom
For imaging centers running triage in live review workflows.
- Unlimited studies
- PACS/viewer integration
- Priority support + onboarding
Enterprise
Custom
For hospital groups standardizing across multiple sites.
- Multi-site deployment
- Custom retention policy
- SLA-backed support
Solipher Index
One engine for exact lookups, priority ranking, and range queries — under a fixed budget.
Teams building latency-sensitive systems — fraud detection, real-time bidding, telemetry ingestion — usually end up running two or three separate data stores to cover exact lookups, priority ranking, and range queries, and paying a synchronization cost to keep them consistent. Solipher Index covers all three from one engine with a fixed, predictable resource ceiling.
Impact
In our own benchmarking on production-grade cloud hardware, concurrent throughput scaled from 54.8k to 1.83M operations per second across 1–64 threads, overtaking common alternative approaches once real contention kicked in — validated across 1,414 real trial runs, not simulation.
Single embeddable engine, replaces two or three separate stores
Fixed resource ceiling — predictable behavior under load, not best-effort
Scales with concurrent demand instead of degrading under it
Drop-in for latency-sensitive ingestion and lookup pipelines
Developer
Free
Single-node, non-production use for evaluation.
- Full engine access
- Community support
Production
PopularCustom
Licensed per deployment for production workloads.
- Production license
- Direct engineering support
OEM / Embedded
Custom
For vendors embedding the engine inside their own product.
- Source-available license
- Joint validation on your workload
Solipher Serve
Keeps an AI inference stack predictable when demand spikes.
LLM and other model-serving stacks tend to degrade unpredictably under load — requests queue silently, latency climbs, and by the time it's visible to an operator, users have already felt it. Solipher Serve sits in front of your existing serving stack and makes overload a visible, predictable event instead of a silent one.
Impact
Currently validating against design partners' real production traffic. Early results are promising enough to expand the program, but we're not publishing a performance number until it's measured against a real workload we can stand behind.
Sits in front of existing model-serving stacks — no rewrite required
Makes overload visible and attributable instead of silent
Configurable to your own resource budget, not a fixed default
Design Partner
Custom
Early access for teams willing to validate against their own serving stack.
- Direct engineering access
- Joint validation on your workload
- Input into the production roadmap
Solipher Cache
Cuts the memory cost of serving AI models at scale.
Model-serving costs scale with how much context has to stay in memory, and most caching layers weren't built for how unevenly real production traffic actually hits that cache. Solipher Cache manages that memory pressure so cost scales with genuine demand, not worst-case assumptions.
Impact
Early validation on production-shaped traffic patterns shows meaningful memory-pressure reduction versus a naive cache. We're expanding the design-partner program before publishing a hard number.
Drop-in memory management layer for existing serving infrastructure
Tuned for the uneven access patterns real production traffic actually has
Reduces serving cost without a hardware change
Design Partner
Custom
Early access for teams with real production serving traffic to validate against.
- Direct engineering access
- Joint validation on your workload
- Input into the production roadmap
Solipher Shield
Line-rate intrusion detection that fails predictably, not silently.
Detecting and mitigating intrusions at line rate needs the same fast-lookup, priority-ranking, and resource-budgeting problem we've already solved elsewhere — applied to live traffic. Under an attack-driven load spike, a defended system needs to know whether a flow was dropped by policy or by capacity, every time.
Impact
Core engineering validation is complete. Traffic-scale benchmarking hasn't started, so we're not making a detection-accuracy or throughput claim yet — we're looking for design partners to help us benchmark against a real traffic profile.
Built on the same resource-budgeting approach validated in Solipher Index
Distinguishes policy rejections from capacity rejections under load
Early-access roadmap shaped by design-partner traffic
Research Partner
Custom
For security teams who want early visibility and input as this moves toward benchmarking.
- Direct access to the engineering team
- Input on the benchmark traffic profile
- First access once validated
Solipher ERP
Odoo, customized so your systems of record stop being three separate systems.
Most clinics, imaging centers, and mid-size enterprises run patient records, billing, and inventory as three loosely-connected systems, stitched together by manual re-entry. Solipher ERP is custom Odoo implementation and module development that connects them properly — including a direct integration path for Solipher Triage deployments.
Impact
Every engagement is scoped to the client's existing systems, so impact is measured per deployment rather than published as a general number.
Custom Odoo module development and implementation
Direct integration with Solipher Triage for imaging-center deployments
Migration support from existing fragmented systems
Ongoing managed support and version upgrades
Module License
Custom
A custom module for a single Odoo instance.
- Module license
- Standard integration support
- Upgrade path with Odoo releases
Implementation
PopularCustom
Full Odoo implementation or migration.
- Requirements & data migration
- Custom workflow configuration
- Staff onboarding
Managed Odoo
Custom
Ongoing administration, customization, and support.
- Monthly customization hours
- Priority turnaround
- Version upgrade management