Skip to content

Gavio roadmap: v3.2 → v5.0 ​

Status: planning document · baseline 3.1.0 · planned 2026-10-18

Seven releases that turn Gavio from a broad runtime into the compliance-grade AI runtime for regulated EU teams, with proof of real use at every step. Dates are targets, not commitments; each release also has an adoption gate and ships only when that gate holds.

Strategy in one line: win two categories outright (EU compliance depth and verifiable trust architecture), stay credible on eval and agents, and stop competing on provider count.

Release train ​

VersionThemeTargetHeadline outcome
3.2Proof and polishDec 2026Real integrations, loop guard, normalised tool calls, full SDK example parity
3.3Semantic evalFeb 2027Tiered LLM-judge eval inside the eval runner
3.4AgentsMar 2027Task-level cost and success across multi-agent runs
4.0EU Compliance EditionMay 2027Data residency plus article-level DORA, NIS2, GDPR and AI Act evidence
4.1EcosystemJul 2027Native framework middleware and enterprise providers
4.2HardeningSep 2027Multimodal PII, streaming judges, adaptive limits, supply-chain evidence
5.0Platform and communityQ4 2027Plugin SDK, outside maintainers, API freeze

Feature IDs follow the F-* convention used in CHANGELOG.md; an ID is recorded there when its feature ships.

Guiding principles ​

  1. Proof before breadth. Every release has an adoption gate as well as a feature gate. If the adoption gate fails, the next release waits.
  2. Keep the design rules. Zero mandatory dependencies, metadata-only audit, lockstep SDKs and shared test vectors (P1 to P8) apply to every new feature.
  3. Parity is part of done. A feature is finished only when Python, JavaScript and Java all pass its test vectors and ship a runnable example. Parity means library parity: CLIs (gavio eval run, gavio workflow release, gavio compliance export) stay Python-only and are listed in the parity matrix.
  4. Compliance claims point to evidence. Anything labelled DORA, NIS2, GDPR or AI Act links to a specific article and a generated artefact, never to marketing text. Mapping evidence helps a compliance team; it does not make anyone compliant.
  5. Measure in CI. Overhead, judge accuracy and cost numbers are produced by CI jobs and copied into docs automatically.

Order of work inside a feature: spec/ schema → shared test-vectors/ → Python reference → JavaScript → Java. A release is cut only when all three pass.

Deliberately not building ​

Not buildingReason
100+ provider adaptersLiteLLM and Portkey already win here; Gavio adds the enterprise few and leaves the rest to the v5.0 plugin SDK
A hosted SaaS gatewayBreaks P6 (embeddable library) and the self-hosted compliance story
Its own agent frameworkGavio governs agents; building one would compete with the frameworks it integrates with
Vector database, fine-tuning, distillationOutside runtime governance; at most, export redacted traces as training data
A general-purpose eval platformBraintrust and Langfuse lead; Gavio's eval stays tied to CI gates and audit evidence

v3.2: Proof and polish ​

Credible to a stranger: real integrations, honest numbers, full SDK parity, and two runtime features developers feel on day one.

IDFeatureDone when
F-DOC-10Docs truth sync: README tables, version map and benchmark numbers generated from spec/, an examples matrix and benchmarks/latest.jsonA stale number fails CI
F-DX-13Example parity: port the missing examples (04, 05, 15, 20 to 25) to the remaining SDKs, port the Python-only platform workflow release to JS and Java, add an example-matrix CI checkEvery example runs in all three SDKs against the in-repo builds
F-REL-09Loop guard interceptor keyed by root run: max steps, max cost, max wall time, repeated tool-call detection; on breach stop, fall back or dry-run log, recorded in the audit recordShared vectors cover all four limits; demo stops a runaway agent at 25 steps or a cost cap
F-ADP-03Normalised ToolCall / ToolResult model in spec/; adapters translate OpenAI, Anthropic, Gemini and Azure formats both ways, including parallel calls; tool arguments pass through PII and secret scanningSame tool-calling code runs unchanged on four providers; vectors per mapping
F-TRUST-02Reproducible interceptor-chain overhead benchmark with fixed seeds and benchmarks/latest.jsonAn outside rerun is linked from the README
F-DOC-11ADOPTERS.md listing real users and how they use GavioAt least three named entries, each with a one-line use

Exit gate: three adopters listed, one external benchmark rerun, parity CI green.

Progress (2026-10-20). Releases 3.2.0 and 3.2.1 ship most of the slices, not all of them: F-ADP-03 in all three SDKs (spec, shared vectors and the Python reference in 3.2.0; the JavaScript and Java adapters in 3.2.1), F-REL-09 with its loop guard in all three SDKs and its runaway-agent demo (example 26-loop-guard) in 3.2.1, and the F-TRUST-02 benchmark harness (the outside rerun is still pending). F-DX-13 (more example ports) and F-DOC-11 are not done, and F-DOC-10 is only started: CI now checks the versions in docs, the spec schema list and the generated benchmark tables, but the Inspector overhead percentages are still typed by hand. So the v3.2 exit gate above is not met yet.

v3.3: Semantic eval ​

Adds a judge case type to the eval runner so suites can check meaning and criteria, not just exact output, while most cases stay cheap. A case moves to the next tier only when the current one is unsure.

  1. Deterministic: schema, regex, hash, length, cost and latency limits. Free.
  2. Classifier: a fixed answer space with a confidence score.
  3. Criteria: a cheap LLM checks a list of rubric points and explains each verdict.
  4. Frontier: a strong model, only for disputes or a sampled share of cases.
IDFeatureDone when
F-EVAL-06JudgeAdapter interface beside ProviderAdapter, reusing the fallback chainInterface in spec/, vectors in three SDKs
F-EVAL-07Classifier judge with confidence threshold and escalate / fail / pass_with_flagUnavailable backend falls back without failing the run
F-EVAL-08Criteria judge: per-rubric-point verdicts with short reasonsPer-criterion results appear in JSON and JUnit reports
F-EVAL-09Pairwise judge comparing prompt version N with N-1, order randomisedRegression gates accept pairwise results
F-EVAL-10Batch collector grouping classifier questions into one callCost and time per suite drop; number published
F-EVAL-11Judge cache keyed by prompt, response and rubric versionRerunning an unchanged suite makes zero judge calls
F-EVAL-12Versioned rubrics with labelled calibration sets, including deliberately bad answersA rubric that passes known-bad answers fails CI
F-EVAL-13Judge audit: every verdict written to the hash-chained audit record as metadata onlyVerdicts verify with the existing chain verifier

Exit gate: published judge agreement on the calibration sets and cost per 1,000 cases by tier, generated in CI from recorded fixtures.

v3.4: Agents ​

Answers the question teams ask about agents: did the whole task succeed, and what did it cost end to end?

IDFeatureDone when
F-AGT-01TaskRecord aggregating every call under one root run: cost, tokens, latency, tool calls, retries, outcomeInspector and gavio cost report show per-task totals
F-AGT-02Per-agent budgets inside a task using Cost Governance v2 decisionsPlanner and coder agents hold separate caps in one run
F-AGT-03Trajectory assertions (required, forbidden and ordered tools, max steps) checked against Tool Runtime replay recordsSuites fail when an agent skips a required approval step
F-AGT-04Optional judge on the final task result, linked to the task recordTask success rate reported per suite and prompt version
F-TOOL-09MCP governance: PII and secret scanning on MCP arguments and results, per-server allow-lists, MCP calls in the audit chainA tool call carrying an IBAN is redacted before it reaches an MCP server

Exit gate: one real multi-agent project running on v3.4 with task records published as a case study.

v4.0: EU Compliance Edition ​

Proves where data went and maps runtime evidence to specific regulatory articles. Requests that cannot be routed in-region fail closed.

IDFeatureDone when
F-RES-01Data residency policy: request-level residency: eu (or a region list); adapters declare endpoint regions explicitly; no match fails closed or logs a dry runAn EU-tagged request never reaches a non-EU endpoint in vectors; region recorded in the audit record
F-RES-02Signed per-period residency evidence reportReport verifies with the Trust Package verifier
F-COMP-01Versioned, signed control catalogue: DORA Art. 28 to 30, GDPR Art. 5, 25, 30 and 32, EU AI Act Art. 12 and 13, NIS2 Art. 21Each mapping names the control, the evidence artefact and its limits
F-COMP-02gavio compliance export --framework <name> --period <period> evidence bundleA reviewer gets one verifiable file per framework per period
F-COMP-03Gap report on the Platform Runtime Profile, per article, not as one scoreGaps listed per article
F-COMP-04Machine-readable provider register as input to a DORA register of informationExport matches what the router actually used

Before release: an independent compliance or risk professional reviews the mappings and their feedback is recorded. Exit gate: external review done and one regulated-industry pilot.

v4.1: Ecosystem ​

Replaces metadata-only adapter payloads with drop-in middleware and adds the providers regulated enterprises buy. Integrations live in optional packages so core stays dependency-free.

IDFeatureDone when
F-INT-04LangChain / LangGraph middleware (Python and JS)An existing app adopts Gavio by changing 2 to 3 lines
F-INT-05Spring AI ChatModel decorator and Spring Boot auto-config (separate artifact)Sample app with PII, audit and residency on
F-INT-06Vercel AI SDK middleware and LlamaIndex callbackSamples in examples/integrations
F-ADP-04AWS Bedrock adapter (stdlib SigV4, declared regions)Vectors pass in three SDKs; EU region routing works
F-ADP-05Google Vertex AI adapterSame
F-ADP-06Mistral (La Plateforme) adapterSame
F-ADP-07vLLM / TGI adapter with self-hosted health checksSame

Exit gate: each integration has a CI-tested sample app; time-to-first-request measured with three people new to Gavio.

v4.2: Hardening ​

IDFeatureDone when
F-SEC-11Multimodal PII: PDF input and region masking on top of the image support from F-SEC-09A scanned ID or bank statement is masked in vectors; core stays dependency-free
F-REL-10Streaming judges: incremental post-interceptors on a sliding window; early stop on a hard policy breachA leaked IBAN stops the stream mid-response in a demo
F-GOV-09Adaptive rate limiting from provider rate-limit headers and retry-afterFewer 429s under a load test; numbers published
F-TOOL-10Stateless MCP governance for servers behind load balancersWorks with two or more MCP server replicas in a sample
F-TRUST-03Supply-chain hardening: SBOM per release, signed artifacts on all registriesEach release ships SBOM and signatures, verified in CI

Exit gate: load test results and SBOM verification published with the release.

v5.0: Platform and community ​

IDFeatureDone when
F-PLUG-01Plugin SDK: versioned interfaces for providers, scanners, judges and policy packs, with a conformance kit built on the shared test vectorsA third party publishes a plugin without a core PR
F-PLUG-02Signed plugin registry with conformance statusRegistry lists plugins with conformance results
F-DOC-12Contributor programme: good-first-issue labels, an owner per SDK and per policy pack, a public roadmap boardTwo or more outside maintainers with merge rights on one area
F-API-02Core API freeze: remove APIs deprecated in 3.x and 4.x (tracked in DEPRECATIONS.md), publish a 5.x LTS policyMigration guide plus scripts per SDK

Exit gate: at least one release where an outside maintainer reviewed and merged a feature.

ID corrections made while landing this roadmap ​

Three IDs in the original draft were already in use and were renumbered; all other IDs are unchanged.

Draft IDAlready used forNew ID
F-REL-08 Loop guardModel version pinning (0.3.0)F-REL-09
F-REL-09 Streaming judges(follows the loop guard renumber)F-REL-10
F-GOV-07 Adaptive rate limitingDrift detection (0.5.0)F-GOV-09

Released under the MIT License.