Pedulli SRD Engine is Pedulli's universal SRD intelligence engine for free-form reasoning, domain reasoning, verifier-gated outputs and AI acceleration. Public capabilities are exposed through measured demos, proof objects and verifier-gated product APIs. Every domain below is a precise, measured pass/fail against the input you provide, built on SRD.
input -> SRD reasoning -> verifier-gated output (accept exact or diagnostic)
MEASURED CAPABILITIES - VERIFIER-GATED OUTPUTS
A capability ships only with a measured benchmark and a published proof. Post-LLM / AGI-class intelligence is our research direction and is labelled research, not delivered.
The contract is simple
The SRD Engine reasons over the input you give it - free-form or domain-specific - and every public output passes a verifier before it ships. You bring the object to evaluate: a factor pair, a key, an assignment, a proof step, a model, a candidate restore. SRD evaluates it against an exact relation. If the relation holds the verifier accepts; if not it returns a precise diagnostic.
A free-form problem or a domain object: factors, a key, an assignment, proof steps, a model, or a candidate restore. The engine reasons over what you provide.
The engine reasons over the input using SRD and evaluates it against the stated relation. The check is deterministic and exact - no sampling, no approximation, no probabilistic shortcut.
If the relation holds, the verifier returns an exact accept. If it does not, you get a diagnostic that says where it failed - never a fake pass, never a guess.
The verify.* domains
Each domain is a measured capability: SRD evaluates the input against an exact relation and the verifier returns accept or a precise diagnostic. Every domain ships with a measured benchmark and a published proof object.
Restores a supplied .pdli byte-for-byte and confirms the SHA-256 of the restored output matches the original.
Checks a supplied assignment against the supplied clause set and reports whether every clause is satisfied.
Walks a supplied proof and verifier-checks each step is a valid inference, returning an exact accept or a precise diagnostic.
Confirms supplied factors multiply back to N and returns an exact verifier-gated accept or diagnostic. Cybersecurity verification and research surfaces are verifier-gated and authorized-scope only.
Re-encrypts with a supplied key and confirms it reproduces the AES-GCM ciphertext, gated by the verifier. Cybersecurity verification and research surfaces are verifier-gated and authorized-scope only.
Checks that a candidate model replays the observed traces exactly. If it cannot reproduce them, you get a diagnostic.
MDP / game: measured value and policy equivalence under a stated symmetry, returned as a verifier-gated correctness result.
Diagnostic of OOS, cost, slippage, drawdown and leakage on a supplied strategy. No ROI, no Sharpe - just the checks.
Produces SHA-256 and timestamp proof assets so a result can be checked and dated by anyone, later.
How the capabilities are scoped
The SRD Engine is the universal SRD intelligence engine. Public capabilities are exposed through measured demos, proof objects and verifier-gated product APIs - every claim below is scoped to what has been measured.
One engine for free-form reasoning and domain reasoning. SRD evaluates the input you bring against an exact relation, across every measured domain.
Every public output passes a verifier before it ships: an exact accept when the relation holds, a precise diagnostic when it does not - never a fake pass.
Cybersecurity verification and research surfaces are verifier-gated and authorized-scope only. The RSA and AES domains return exact, gated checks on supplied factors and keys.
SRD reasoning and verifier-gated outputs are exposed as product APIs for AI acceleration, so downstream systems can build on measured, gated results.
Post-LLM / AGI-class intelligence is our research direction. It is labelled research, not delivered - no claim runs ahead of the evidence.
A capability ships only with a measured benchmark and a published proof. Anything not yet measured is labelled research, not delivered.
YOU BRING
input -> exact relation
VERIFIER RETURNS
accept (exact) or diagnostic
FAQ
Run the universal SRD intelligence engine on your own input and see an exact accept or a precise diagnostic. Then check the measured benchmarks for yourself.
Universal SRD intelligence engine - measured capabilities, verifier-gated outputs, bounded by measurement