Intelligence Machine learning · Knowledge graph · Continuous retraining

Every lap makes the platform smarter.

The system learns from millions of telemetry samples rather than from static engineering rules. It gets better every day, on the sessions people upload to it.

01 — Asset A motorsports knowledge graph

Every uploaded session becomes part of a continuously expanding body of engineering knowledge.

SRC 01Telemetry

SRC 02Setup files

SRC 03Weather

SRC 04Surface conditions

SRC 05Driver inputs

SRC 06Tire information

SRC 07Vehicle configuration

SRC 08Session metadata

SRC 09Track conditions

SRC 10Vehicle responses

As participation increases, recommendations become measurably more accurate.

Eventually the platform develops statistical confidence that no individual race team can replicate on its own — not because the engineering is smarter, but because the sample size is larger than any single program will ever collect.

02 — Pipeline Continuous retraining

The model is retrained from eight independent sources — and graded against results.

Customer telemetryEvery session uploaded to the platform, with the setup and conditions it was run in.

Reference datasetsReference behaviour from cars driven at the limit by drivers operating at theirs.

Controlled testingDeliberate single-variable runs, where the answer is known before the data is collected.

Controlled sim runsPerfectly labelled telemetry at volume. Every variable known, every run repeatable.

Engineered test programmesDeliberate setup work run as test programmes by Vaquero engineers, where the variable is known in advance.

Validation testingRecommendations tested against the sessions that follow them. Failures are recorded as failures.

Engineering reviewHuman engineers audit the model's reasoning, not just its output.

Historical comparisonCross-session, cross-condition, and cross-vehicle trends over a full program.

03 — Data How sessions are handled

Retention is what makes any of this possible.

What happens to a session

Uploaded to our servers
Parsed, segmented and analysed
Retained as your performance history
Used to improve the models
Aggregated into car and track baselines

Without history there is no trend analysis: no separating a genuinely quicker car from a good night, and no way to check whether last week's recommendation actually worked. A platform that forgot every session would be a graph viewer.

The same applies to the models. One that has seen ten thousand laps of a car at a track gives better guidance on that combination than one that has only seen yours — which is the whole reason the recommendations sharpen over a season.

Full terms covering storage, retention, and use are set out before an account is connected.

05 — Proof Stated plainly

What we can defend today.

P 01Purpose-built AI architecture

P 02Cloud-native engineering platform

P 03Telemetry-first development

P 04Race engineering background

P 05Recommendations validated against your next session

P 06Vehicle dynamics and software in one team

P 07Relationships inside iRacing communities

P 08Growing telemetry collection

As adoption grows, the claims become measurements: lap time improved, consistency gained, engineering hours saved. The product generates the proof — not the marketing.

Every lap teaches it something. Yours included.

Ask what we do with a session before you send one. It is a fair question and it has a specific answer.

Request access