01
Lower AI cost
Target less unnecessary model spend while keeping the workload outcome comparable.
Coming soon · In development
RHEA Kernelas a Service.
A future commercial service designed to bring RHEA Kernel efficiency to enterprise AI workloads — with a savings-based commercial model and clear private-data boundaries.
Commercial direction
Reduce the cost of useful AI work.
RHEA Kernel as a Service is being designed around measurable efficiency: lowering AI compute, model usage, token consumption, latency, monetary cost and energy while preserving the required output quality.
02
Less model work
Target fewer calls and fewer tokens where the required result can still be preserved.
03
Lower latency & energy
Efficiency targets include time and energy as well as monetary cost.
04
Private by boundary
Private company knowledge remains within its authorized private scope.
Proposed plans
Private, or Shared Network + Private.
Both plans are designed around isolated private company knowledge. The Shared plan adds access to permitted shared knowledge without giving one tenant access to another tenant's private data.
Private
Private onlyBuild efficiency from your own authorized private knowledge.
€149 one-time entry
Standard proposed entry price.
20% of verified savingsor 18% with annual settlement.
- Private knowledge scope only
- No Shared Knowledge Network access
- Entry price does not rise with network maturity
- Savings-based fee after entry
Shared Network + Private
Network + privateKeep your private scope and add access to permitted, safe shared knowledge.
From €399 one-time entry
Proposed range: €399–€798 depending on network maturity when joining.
20% of verified savingsor 18% with annual settlement.
- Everything in Private
- Access to permitted shared knowledge
- No direct tenant-to-tenant access
- Entry terms would be locked when accepted
Proposed pricing only. These figures are a product-development preview, not a binding commercial offer. Taxes, contractual methodology, billing mechanics and final eligibility rules remain subject to launch terms.
Savings-based model
RHEA would get paid when eligible savings exist.
The intended commercial model compares an agreed equivalent workload without RHEA against the corresponding workload with RHEA. The success fee is applied only to eligible monetary savings, not directly to calls, tokens or energy figures.
01 · Comparable baseline
What the workload would cost without RHEAMeasured using an agreed methodology and equivalent required output quality.
02 · RHEA-enabled cost
What the comparable workload costs with RHEAIncluding agreed comparable execution costs and relevant overhead.
03 · Eligible savings
The positive monetary differenceThe applicable RHEA success-fee percentage is then applied to eligible savings.
Illustrative example: if verified eligible savings are €1,000 for a month, the standard monthly success fee would be €200. Under annual settlement, the standard rate is 18% rather than 20%.
Proposed founding-customer terms
First global customer
50% entry reduction. Proposed success-fee schedule: 8% in customer year 1, 15% in year 2 and 20% from year 3. Annual settlement applies a 10% relative reduction to those rates.
Second global customer
25% entry reduction. Proposed first-year success fee starts at 8% if joining in the same month as the first customer and rises by one percentage point per month until the standard 20% rate is reached at 12 months. From year 2: standard rate. Annual settlement applies the same 10% relative reduction to the applicable success-fee rate.
Only the first two global RHEA customers would qualify. Eligibility and terms would be assigned and locked by RHEA, not self-selected by a customer.
Knowledge boundaries
Designed to keep private data private.
The public product concept distinguishes private company knowledge from any future shared knowledge layer. This page intentionally describes product boundaries, not the proprietary mechanism used to achieve them.
Private
Private tenant material remains within its authorized scope.
- No automatic pooling of private documents.
- No automatic sharing of private conversations.
- No direct access between private tenants.
Shared Network
Any future shared layer would require controlled, permitted contribution and reuse.
- Only authorized, generalizable knowledge.
- Contribution permission and reuse permission are distinct.
- Provenance and validation would be required.
Disclosure boundary. RHEA Kernel remains proprietary. This preview does not publish internal algorithms, scoring, thresholds, data structures, routing heuristics or other reconstruction-enabling implementation details.
Future product
RHEA Kernel as a Service is being built.
Commercial enrollment is not open yet. Follow the public project record for approved milestones and future availability updates.