Investor Advisor Media Docs
AvGuard · Correspondence
· An Invitation ·
Private · For the Named Recipient
Strategic Advisor Invitation
To —

Petter
Hörnfeldt.
Mentour Pilot.

A letter, a prototype, and an invitation to help ensure a new safety system remains grounded in the realities of the cockpit.

// From
Grzegorz Founder, AvGuard · Student pilot, Bliss Aviation · Bournemouth, UK
// Regarding
A Strategic Advisor role An offline AI safety system for general aviation analogue cockpits
From the desk of Grzegorz · Founder, AvGuard
April · 2026
Hi Petter,

I hope you've been well.

Over the past months I've been working on a safety-focused aviation project inspired by the patterns that repeatedly appear in accident investigations — particularly in general aviation, where pilots often operate legacy aircraft with limited real-time situational support.

I now have early real-world flight recordings captured directly in-aircraft, including realistic vibration conditions that are very difficult to replicate in lab environments. I also have cooperation from my local flight school, which allows me to test concepts in authentic cockpit settings, although access is understandably limited due to cost and aircraft availability.

My goal is to build an offline-capable system that observes cockpit instrumentation and context, correlates it with known risk patterns derived from real accident scenarios, and provides early situational awareness cues before a chain of events develops.

Your work has strongly influenced how I think about the problem — especially the importance of recognising subtle precursors rather than reacting to already critical situations.

Because of this alignment, I wanted to ask whether you would consider acting as a Strategic Advisor to the project, contributing a small amount of time occasionally to sanity-check direction and ensure the safety narrative remains grounded in real operational reality.

In return, I would allocate an equity portion reflecting the value of your insight and credibility in the aviation safety domain. Your involvement would also help communicate clearly to early-stage investors that the project is guided by real aviation expertise.

There would be no heavy time commitment expected — mainly occasional feedback on direction, assumptions, and prioritisation of safety-relevant features.

I believe this project has the potential to meaningfully improve situational awareness in GA cockpits, particularly for aircraft that do not benefit from modern glass avionics.

If this sounds interesting, I would be very grateful for a short call to show you the current prototype approach and flight footage.

Regardless, I want to thank you — your work has already contributed significantly to shaping the thinking behind this system.

Kind regards,
Grzegorz
Founder · AvGuard
Bournemouth · United Kingdom
Av
Confidential
01 · Alignment

Why your voice — and
why this system.

Recognise the precursors — before they
become the emergency. — AvGuard's design philosophy, shaped by years of your work
// Your Teaching

Subtle precursors
over headline failures.

Your channel has, for years, patiently walked pilots through the quiet moments before the loud ones — the small deviations, the drifting trends, the misread cues. That's exactly the interval AvGuard is built to monitor.

The system doesn't wait for a stall warning. It watches the degradation of total energy, the correlation of airspeed with bank angle, the context of the phase of flight. This is not an accident — it is the direct translation of a philosophy your work has championed.

// The Gap We're Filling

Legacy cockpits
deserve modern minds.

Roughly 211,000 aircraft in the US alone still fly with 1960s analogue gauges. Modern glass retrofits cost $50,000–$100,000 — out of reach for most training schools and private owners. These are the very aircraft that most need better situational support, and the very pilots your audience is composed of.

AvGuard is a non-invasive, clamp-on edge unit: a camera reads the six-pack, a local reasoning engine correlates live data against a vectorised NTSB/ASN corpus, and alerts are issued within the aircraft — no cloud, no connectivity, no subscription to a service that might not work at 8,000 feet.

02 · The Patterns

The accident chains
you've taught a generation
of pilots to see.

These are the recurring patterns AvGuard's correlation engine is designed to surface early — each drawn directly from the kinds of cases your audience already understands intuitively.

i. Base-to-final stall/spin Low altitude · uncoordinated turn · decaying airspeed Energy · AoA
ii. VFR into IMC Visibility degradation · spatial disorientation · graveyard spiral Attitude · Bank
iii. Fuel mismanagement Tank selector missed · imbalance · unexpected starvation Trend · POH
iv. Engine parameter drift Rising EGT · falling RPM · mixture misconfiguration Correlation
v. Unstable approach Crossing gates high/fast · go-around threshold missed Phase · Energy
vi. Gradual altitude loss VSI/altimeter divergence · pilot workload saturation Z-score · Rate
vii. Pitot / static anomalies Airspeed inconsistency · altitude–IAS decoupling Cross-val.
viii. Controlled flight toward terrain Descent rate vs. terrain model · pattern match Context
03 · The Role

What the advisor role
actually looks like.

A low-commitment, high-signal role — designed to respect your time while giving the project the benefit of a genuinely experienced aviator's sanity check.

// Time Commitment
~2h / mo

Roughly a monthly check-in plus occasional written feedback on safety-relevant design choices. Asynchronous wherever possible.

// Compensation
Equity

An advisor equity allocation (vesting over 24 months) proportional to the value of your insight and credibility. Specific percentage negotiated on the call.

// Term
Flexible

Initial commitment of 12 months, renewable by mutual agreement. Exitable at any time — no hard commitment, no reputational obligation if the direction ever feels wrong.

// What the Role Is

  • Occasional sanity-check on direction and priorities
  • Written feedback on safety-relevant features before release
  • Optional mention in investor materials and public-facing page
  • A sounding board when an ethical or operational question arises
  • Light introductions where naturally relevant — never asked-for

// What the Role Is Not

  • A full-time or even part-time engagement
  • A requirement to publicly endorse the product
  • Responsibility for operational outcomes or certification
  • Code review, engineering, or day-to-day involvement
  • Any obligation to promote AvGuard on your channel
04 · On the Call

What you'd see
in the first 20 minutes.

If we find half an hour, here is what I'd show you. No slides. Just the system, the footage, and the rationale.

// Clip 01
Real-world cockpit footage.
In-aircraft recording at Bliss Aviation — six-pack under engine vibration.
Runtime · ~2 min
// Clip 02
The correlation engine.
Live telemetry vector-searched against 151,183 NTSB & ASN records.
Runtime · ~3 min
// Clip 03
Base-to-final, worked live.
Reconstructed NTSB scenario in MSFS — alert issued 30 s before the stall.
Runtime · ~2 min
// 01

Real-world flight footage.

In-cockpit video captured at Bliss Aviation showing the six-pack under actual engine vibration — the exact environment that defeats most academic computer-vision work. This is the footage the gauge-reading models are trained and validated on.

Global shutter · 60 fps Cessna 152/172 VFR + pattern work
// 02

The correlation engine.

A live walkthrough of how live telemetry is vector-searched against 27,684 NTSB accident reports and 3,062 ASN incidents — and how the physics-aware filter prevents the LLM from hallucinating advice outside the aircraft's actual operating envelope.

151,183 vectors Offline RAG POH-bounded
// 03

A worked base-to-final example.

A reconstructed scenario from real NTSB data. Airspeed trending down, bank angle creeping up, AoA rising — and the thirty-second window in which the system raises a specific, POH-derived alert before the stall.

Simulated · MSFS NTSB-grounded 30s lead time
// 04

Your honest questions.

The most valuable twenty minutes, honestly. Where the aviation reality does not match the engineer's assumptions — which is every project like this one, without exception, unless someone like you challenges it early.

Open discussion Candid feedback No agenda
· A Single Request ·

Would you give
thirty minutes
to the footage?

If the idea resonates after that call, we talk about the advisor role properly. If it doesn't, no harm done — and you've helped shape the thinking either way.

No slides. No formal pitch. Just the footage, the prototype, and an aviator's honest reaction to it.

Regardless of your answer — thank you.
Your work has already contributed
significantly to how this system thinks.

Kind regards,
Grzegorz

Narration Offline AI Co-pilots for Analog Planes