The coach that has the numbers.

Like a race engineer on a Formula 1 pit wall, Navigatek talks to your crew while you sail. It reads your sensors, your wind and your position in real time, and turns them into spoken calls you can act on.

LIVE Audio coaching on board

COACHING Coaching: Bear away firmly to hold 40° … next tack in ~500 metres … Wind steady.
VERDICT Gust absorbed: +0.5 kt of speed (peak 12 kt). Good absorption, keep going.
ALERT Plan C, ⚠️ TACK NOW — VMG collapsed (0.5 kt against 5.1 on the other tack) — header too costly
VERDICT Well played, you tacked — I had just recommended it (12 s earlier) … Wind will shift −2°
VERDICT Tack completed in 9 s to regain speed (excellent), with an estimated loss of ~1 s against a perfect tack.
COACHING Coaching: 🟢 Mark rounded — speed first, heading 243° to mark 3.
COACHING Coaching: Bear away 10° to hold 152° … Wind will shift +13°.
COACHING Coaching: Heading good … next gybe in ~1600 metres … Wind will shift +11°.

Real messages, spoken aloud on board during a race.

Measured, never guessed.

Every recommendation comes from a deterministic tool: dual GPS at 10 Hz, onboard wind sensor, polar curves, drift maps learned on your own water. The artificial intelligence computes none of it itself — it reads those results and explains them to you in plain language, during the race or after.

If the data does not exist, Navigatek says so. It never invents a wind, a VMG or a time to the line.

It augments the crew. It does not replace it.

No AI touches the helm, the trim or the call. Navigatek widens what the crew can see and measure, so the decision — which stays yours — rests on facts instead of impressions.

Six AIs. One crew.

Six specialised systems work together, share the same data and speak the same language — live coaching during the race, structured debrief afterwards. Together they raise the crew's tactical ceiling.

Navigatek — Six beasts, one shield — six AIs, one crew.

Wind AI

Studies how the wind behaves during the race, and race after race. It offers the crew an optional bold route — a bet on the side of the course that pays — always stated as an option, never as an order.

Race-director AI

Speaks out loud on board: live advice through the boat's speaker while you sail. After the race, it answers any question about what happened, in conversation.

Tactician AI

Trained on tactical computation: VMG, polars, laylines, and headers. It knows when a shift is worth a tack — and when it is not.

Post-race analyst AI

Produces a detailed report of every decision, manoeuvre and condition — so the debrief is structured and built on observable facts rather than memory.

Sensor-watch AI

Together the AIs follow everything that happens on the boat — gusts, tacks, speed breakdowns — through the onboard sensor suite. Nothing measurable goes unnoticed.

Watch AI

Reads the scientific literature and the specialist racing press on the web, then commits what matters to memory — so the crew can be given genuine expert advice. It issues a daily briefing and contributes to the post-race reports.

And together, they correct themselves.

The set of AIs learns, finds its own mistakes and corrects them at every training cycle on the water and after every race. What was wrong last Sunday is measured, named and fixed.

250+ adjustable parameters, used to refine the AI models and improve what they deliver.

Everything measurable, measured.

The AIs share one sensor suite and follow every action on board, second by second.

GustsRise predicted before it hits, with its lead time.
Tacks & gybesDetected, timed, and costed in seconds lost.
Speed breakdownsA drop the average hides — caught and dated.
True windDirection, strength, shift and oscillation period.
Heel & attitudeCompass, gyroscope, calibrated heading.
PositionDual GPS, ground track, distance to the line.

The debrief writes itself.

Below: the opening of a real report, generated automatically after a race on Lac Tremblant. Nothing was rewritten for this website — the scores and the numbers are those of that day.

Race report — 30 August 2026 · Lac Tremblant · CAN157 · 44.2 min · wind 9.2 kt from 290.3°

START5.7/10

SPEED4.7/10

MANOEUVRES1.7/10

TACTICS—/10

AI COACHING5.4/10

1 · Synthesis

Race of 30 August 2026 (44.2 min) in a westerly reaching wind (TWD ≈ 290°, mean TWS 9.17 kt, trend –13.4°). The start cost 14.5 s of delay (21.2 s late crossing). Upwind, mean TWA 51.4°, SOG 4.83 kt, 88.5 % of polar, mean heel 14.9°. Downwind, TWA 150.7°, SOG 5.46 kt, 95.0 % of polar, mean heel 2.6°. Six tacks (mean loss 3.6 s) and no gybes, total loss 48.3 s (1.8 % of the time). A speed breakdown at 15:50 caused a 0.65 kt drop over 515 m for 243 s. Fourteen shifts were detected, only one played (rate 7.1 %). The AI coaching flags 52 anomalies across 18 checks, dominated by family B1 (45 cases of impossible TWA angles). One crew error held a 46° header for 347 s without tacking.

Speed breakdown(s) detected — a step in the speed that an average hides:

15:50 upwind: speed falls from 94.8 % to 82 % of polar — 243 s lost (515 m)

And the analysis that comes with it.

**The track, read metre by metre** Race 1 — wind 310°, tacking angle 68°, shift +11°. The trace is coloured by how well the boat was sailing: green in the groove (TWA ≈ 41°), red pinched and slow, orange bearing away. The six tacks are circled; the blue line is the downwind return.

**The wind across the whole course** Dominant wind west ≈ 288°, 8.9 kt — 43 measured, 40 filled in from the learned atlas, 110 interpolated. Each arrow points where the wind is going, coloured by strength. Green dots are lifts, red dots headers: 6 and 8 on this race.

The report runs 12 pages: conditions, start, speed, manoeuvres, wind, tactics, AI-coaching audit, fleet, progression, crew decisions and an action plan.

Autonomous on board. No cloud, no tokens.

Navigatek builds specialised race-strategy algorithms, operated by AIs running on dedicated infrastructure — and, ultimately, on the boat itself. No per-token bill, no subscription that grows with every race.

The reasoning is simple: running race operations on cloud AI would be far too expensive, and the network to reach it may simply not be there. A lake, a bay, an offshore leg — coverage is never a promise. So the system has to stand on its own: to cut the cost of a race, and to depend on no network at all.

The problems it is built to solve

Spot an anomaly nobody sees

A rope trailing in the water, weed on the keel — a drag that costs speed the whole leg while the crew looks for the fault everywhere else.

Measure every manoeuvre

From the plain tack to the harder ones — absorbing a gust without losing the groove — detected, timed and scored against a perfect execution.

Remember the wind across the course

Learn how the breeze behaves as the boat moves through it, so a bold tactical plan can be offered — the side that pays, argued with measurements.

Read any sensor worth reading

Heel, drift, heading, wind — through sensors of every kind, including ones never meant for boats. If a sensor exists and can be useful, Navigatek can read it and put it to work.

Who it is for.

Navigatek is not a product for sale. It is a system that exists and already races, together with the expertise that built it — of which you integrate all or part. Three ways to use it.

Racing teams

Professional and amateur teams who build — or commission — their own race tools, and who are looking for deep expertise in the method rather than one more product.

Advanced expertise to fold into your development methods, for the technology you use during a race as much as for the technology used to debrief it.

What you already haveA review of your tools, your data, and how decisions are actually made on board today.
Where the measurement is missingWe name what is decided on impression for want of a number. That is where the gain sits — not in one more sensor.
The method, not the recipeHow to instrument, how to verify, and how to refuse to conclude when the data does not allow it — the part that cannot be copied.
Handover to your teamYour people leave with the approach and can run it again without us. Expertise that creates a dependency is not expertise.

Your involvement: your technical team and your race data. Expertise is handed over, not rented indefinitely.

Sailing schools & academies

Institutions teaching race sailing that want a structured, fact-based debrief without having to build the technical chain themselves.

A complete post-race coaching kit, integrated at your school: sensors, recording and report, ready to serve as teaching material.

What the session has to produceWe start from the lesson you teach, not from the sensor we could fit.
The kit, fitted to your boatsSensors and recording installed, ready to use. Nothing to operate during the session.
The debrief becomes the lessonEvery manoeuvre timed and costed, every decision dated: the report is the material, and it no longer has to be prepared.
Your instructors own itThey read the report and use it without us — the condition for the kit to still be useful two seasons from now.

Your involvement: your boats and one typical session to calibrate the kit. During the lesson, nothing: the instructor keeps their eyes on the students.

Makers of racing technology

Manufacturers of racing instruments who want to put artificial intelligence into their products, and to know where to start.

Two questions, one answer: how to integrate AI into your products, and which capabilities from this project to take over directly — all of them or some.

What your instruments already knowAn inventory of the data you capture, of what is missing, and of what AI can genuinely draw from it.
What can be taken as it standsAmong what this project already runs in a race, what integrates into your products without being reinvented — and what does not lend itself to it.
Measure before promisingWe establish on real data what is actually reachable, and we say so when it is not. A capability announced and never delivered costs more than one refused.
Integration, then proof on the waterNothing ships on a bench result alone: it races first, then goes into your instruments.

Your involvement: your data and your product constraints. The measuring and the proving happen on our side.

Built on the water, not in a lab.

Navigatek was born on Lac Tremblant, aboard a J/70, over a full season of racing. Every function was tested in a real race before it was added to the product — not the other way round.

Frequently asked questions

Does it work on any body of water?

Yes — with or without network coverage. That independence is a design choice, not a side effect: nothing on board waits for a signal.

What equipment do I need?

Almost anything can work. The products are built from sensors of every type, and the readings go through a normalisation step so that they become compatible with the rest of the system.

Is it only for racing boats — and only certain sizes?

It is a high-end solution made for racers, but it serves just as well for learning performance sailing.

Can I use it on a catamaran?

Some of the algorithms carry over.

How do I receive the coaching in real time?

By audio — through a speaker in the cockpit, spoken aloud while you sail.

What happens if a sensor fails?

The system degrades rather than serving something false: it drops back to a source it can still trust, and it says so out loud. A wrong number is worse than a missing one.

Is there a subscription, or per-token billing?

No. The AIs run on dedicated infrastructure and, ultimately, on the boat itself — so the cost of a race does not grow with how much you use it.

Does the race data stay mine?

Yes. Because everything runs on board or on your own infrastructure, your logs, your tracks and your reports stay where you put them.

The watch: what sailing publishes, read for you.

Every morning, Navigatek reads the sailing press and the blogs, keeps only what a racing crew can actually use, and says why. Each item carries its source link — so you can always go back to the article.

Example — watch of 11 September 2026

J/70 Worlds at Cascais, Portugal Day 3

Source: Sail-World — sail-world.com

The article covers the third day of the J/70 World Championship in Cascais, with the standings changes and the wind conditions.

What to take from it: On a course closer to shore, and therefore shiftier, a bad start can be made up by working the shifts and the puffs.

For a J/70: Using shifts to regain places applies directly to club racing on a lake like Tremblant, where shore effects are common.

Get in touch.

Whether you race, teach or build instruments — if any of this speaks to how you work, write to us. We are happy to go into the detail: what is measured, how it runs on board, and what it would take on your boat.

Born on a J/70. Other classes will follow.