My Guide to OSINT, Noise Reduction, and Modern War Forecasting

OSINT is not just a hobby. It is a citizenship skill. It is a defense mechanism. And in a world drowning in misinformation and disinformation, it is a lifeline.

Licensed by the author from Envato Elements

If you have ever stared at a breaking headline about Ukraine, or Gaza, or Taiwan and thought, “How do analysts actually know what’s coming next?”

This is the article I wish someone had handed me ten years ago.

I didn’t learn forecasting from a crystal ball or by meditating on a mountain, although I did spend enough time in the Army infantry that I could have used a meditation break from time to time. What actually shaped my analysis were the years afterward, when I deployed with the Air Force’s E-3 AWACS to Ecuador as part of Joint Interagency Task Force South, chasing drug traffickers across the Pacific and learning how real-time intelligence is turned into real-world action.

For the record, we arrested our bad guys, we didn’t blow them up without a trial…

Later, as a veteran in the private sector, I found myself teaching these same methods to people who wanted to understand geopolitics without drowning in speculation. And, to be honest, this skill came in handy in the corporate marketing world.

Fun fact: For seven years, I worked as an adjunct professor of predictive analytics at the worst college in Michigan!

And now, as a military journalist covering the Ukraine War every day, those skills keep me sane in a digital battlefield overflowing with propaganda, hype, bots, Canva fever dreams, and the occasional Russian who insists he’s “not mobilized” while wearing brand-new Ratnik gear.

There is no magic crystal ball or obscure divination required. And no, I no longer have access to classified intelligence, although I do read a lot of books.

Good Judgement OSINT super-forecasters were over 30% more accurate than US government intelligence analysts (who had access to classified information).

Note: Good Judgement is an annual competition where people like me attempt to predict real-world events.

So how do we do it?

This guide is my complete, unfiltered playbook for how to analyze global security events, how to forecast conflict, and how to build your own OSINT (Open-Source Intelligence) workflow, without losing your mind or turning into the kind of person who peers suspiciously at every cloud because it looks “kind of missile-shaped.”

Let’s get to it.

How intel analysts see the world

Thinking in probabilities

Most people want geopolitical predictions in the same format as a fortune cookie.

“Will Russia collapse this year?”
“Will China invade Taiwan?”
“Will AI cause the heat death of humanity before I finish my student loans?”

Real forecasters don’t think in absolutes. We think in probabilities.

When I was working with the AWACS crews in Manta, Ecuador, our job wasn’t to say, “this aircraft is smuggling cocaine.” It was to say, “this aircraft has a 71% probability based on speed, altitude, track behavior, and known smuggling routes.”

That’s how intel works. You are always estimating risk, not certifying destiny.

When I predicted in 2015 that Trump had an 87% chance of winning the 2016 election, people thought I had lost my mind. Not because they had better data. But because they were trapped in a mental model where Trump losing was absolute. This is why his win hit so many Democrats so hard; they didn’t even entertain the idea that it was possible.

Replace “yes/no” with 0–100% and forecasting becomes much more honest. I know it sounds obvious, but this is the way I like to think about it: If something has a 30% chance of happening, that means if you ran the same timeline 100 times, it would occur in 30 of them. Hooray for the multiverse!

It’s not magic. It’s math plus humility.

So, let’s agree to never assign 0% or 100%

Nothing in global security is certain except two things:

1. Someone is lying.

2. Something bad will happen when nobody wants it to.

That’s why good analysts never use the word impossible. Russia invading Ukraine in 2022 seemed “extremely unlikely” right until the moment Russian tanks rolled across the border like a drunk neighbor driving across your lawn.

Leave yourself room for uncertainty. It keeps you honest.

How to filter signal from noise

Noise is the enemy of analysis. And the modern world generates more noise than a Russian mobik bus with no shocks.

When I first started forecasting, social media was still relatively sane. Today, the ratio of legitimate reporting to flaming garbage is roughly 1 to 40.

So, here’s how you tame it.

1. Build Your OSINT Dashboard Before a Crisis Hits

During my JIATF-S days, every crew shift began with setting up our information streams. If your intel feeds aren’t configured before an event, you’re already behind.

Here’s the modern civilian version:

Set Google Alerts for:

• “Ukraine counteroffensive”

• “Russian missile strike”

• “Patriot missile Ukraine”

• “Shahed production Russia”

• “NATO posture update”

• “China South China Sea”

• “Baltic Sea sabotage”

Use quotes for precision. And set them to “at most once per day,” or else your phone will vibrate like a Marine discovering Rip-It’s energy drinks exist. Also, it may be prudent to create a new Gmail account specifically for this purpose so as not to clutter up your primary inbox. After all, I don’t want all your “Wes O’Donnell” newsletter emails to get buried by Google alerts.

Collecting sources is easy. Knowing how to follow them without drowning in noise is the part that separates analysts from people screaming into Reddit threads at 3 a.m.

Below is a simple, structured method to follow each source, understand what platform they live on, and turn them into a daily OSINT radar picture.

Let’s break this down feed by feed.

Follow these OSINT sources

Defense Express

Where it lives: Website + Telegram

How to follow:

• Bookmark their English site

• Join their Telegram channel for real-time updates

Why it matters: They break technical stories early, especially missile and radar analysis.

United24

Where it lives: Website + X + YouTube

How to follow:

• Subscribe to their YouTube channel

• Turn on notifications on X

Why it matters: They get official access no one else has. When a Ukrainian pilot or commander speaks publicly, it’s usually through United24.

DeepState UA

Where it lives: Telegram + Interactive Map

How to follow:

• Follow their Telegram channel

• Check their live map for battlefield changes

Why it matters: Their geolocation and frontline updates are some of the fastest and most accurate in the war.

GeoConfirmed

Where it lives: X (formerly Twitter)

How to follow:

• Follow their X account

• Enable “See first” or notifications

Why it matters: Crowdsourced geolocation teamwork at a professional level. They verify explosions, strikes, and equipment sightings faster than most governments.

Ukraine Weapons Tracker

Where it lives: X

How to follow:

• Follow their X account

• Add them to a private “Ukraine OSINT” list

Why it matters: They track every confirmed equipment loss with photographic proof. This is data, not noise.

Oryx (archived)

Where it lives: Their website

How to follow:

• Bookmark the site for loss lists

• Understand that updates are now volunteer-run

Why it matters: Oryx pioneered visual confirmation methodology. Still valuable historically and for trend analysis.

Janes Defense

Where it lives: Website (subscription)

How to follow:

• Subscribe if you can budget it

• Use their daily briefs

Why it matters: Janes gives sober, technical breakdowns with decades of institutional memory. A perfect noise filter.

The War Zone (The Drive)

Where it lives: Website

How to follow:

• Turn on browser push notifications

• Follow their journalists on X

Why it matters: They break aerospace and stealth stories before anyone else. If a weird object is flying somewhere, they’re on it.

Euromaidan Press

Where it lives: Website + X

How to follow:

• Add their RSS feed to Feedly

• Follow them on X

Why it matters: Clean, summarized reporting. Great for situational awareness.

Bellingcat

Where it lives: Website + X + YouTube

How to follow:

• Watch their YouTube tutorials

• Follow investigative threads on X

Why it matters: They are the gold standard of OSINT methodology. Learn from their structure and skepticism.

Add specialized feeds

This is where your OSINT setup evolves from “curious observer” to “informal intelligence analyst.”

Maxar, Planet Labs

Where they live: Company blogs + X

How to follow:

• Turn on X notifications

• Subscribe to their imagery newsletters

Why it matters: They release high-res satellite pictures of strikes, bases, naval movement, and convoy losses. This is your overhead ISR without needing an actual satellite.

NOTAM Trackers (Notice to Airmen)

Where they live:

• ICAO NOTAM search

• NotamMap.com

• Telegram aggregators

How to follow:

• Check daily NOTAM clusters in Russia, Belarus, and Crimea

Why it matters: Large new NOTAM zones often mean missile tests, air defense drills, or temporary flight restrictions before strikes.

Maritime AIS Trackers

Where they live: MarineTraffic, VesselFinder

How to follow:

• Track Black Sea movements

• Watch for Russian naval logistics patterns

Why it matters: Russia’s fleet movements give away missile launch timings, resupply cycles, and evacuation patterns.

FlightRadar24

Where it lives: App + Web

How to follow:

• Watch NATO ISR aircraft loiter patterns

• Create alerts for E-3, E-8, RC-135, RQ-4 tags

Why it matters: When NATO ISR jets start circling aggressively, something big is usually happening. Old AWACS guy here, trust me.

ACLED

Where it lives: Website (subscription optional)

How to follow:

• Use their conflict map to visualize event clusters

Why it matters: Lets you track global escalation patterns, not just Ukraine.

Global Terrorism Tracker

Where it lives: Database

How to follow:

• Download the latest reports

Why it matters: Useful for forecasting instability spillover and proxy activity.

How to actually use all of this

Most people follow these sources like they follow gym memberships: enthusiastically at first, then not at all.

Here’s how to build a sustainable OSINT workflow:

Step 1: Build a Feedly or Inoreader dashboard

Create categories:

• Ukraine War

• Missile and Air Defense

• Naval Movement

• Russia Internal

• NATO Air Activity

• Satellite Imagery

• Defense Industry

Add RSS feeds where possible. This becomes your “Intel Summary” page.

Step 2: Build a Twitter/X List

Call it:

“Ukraine OSINT — Primary Feeds”

Add:

• GeoConfirmed

• UATracker

• Defence Express

• Visegrád 24

• Malinka_110

• War Mapper

• Frontelligence Insight

• Analysts like Rob Lee, Dara Massicot, Trent Telenko, John Ridge

Then view only that list, not your chaotic main feed.

Step 3: Set Google Alerts

Use Boolean logic: (examples)

• “Ukraine missile strike” AND “satellite”

• “Black Sea Fleet” AND “Sevastopol”

• “Iskander” AND “Patriot”

• “Shahed” AND “air defense”

• “Crimea explosion”

• “Kursk airfield”

Set them to “deliver as it happens.”

Step 4: Pipe everything into an AI summarizer

At the end of each day, feed your OSINT inputs into an AI tool and ask:

• “Summarize patterns.”

• “What is new compared to yesterday?”

• “What indicators moved?”

AI keeps you from drowning. You provide the interpretation. More on AI in a bit.

Step 5: Track the signal, ignore the noise

After a week, you’ll start noticing patterns:

• Missile strikes cluster on foggy nights

• Russian rail activity spikes before major offensives

• Changes in NOTAM zones precede missile launches

• NATO ISR aircraft shift prior to big events

• Satellite imagery drops hint at destroyed depots

Your brain begins building a radar picture, just like sitting at an AWACS console, minus the bad coffee and the smell of hydraulic fluid.

This is how modern OSINT works. It is structured. Disciplined. Repeatable. And when done correctly, frighteningly accurate.

This is the part they never tell you: You do not need to be a government analyst to understand the world. You just need a workflow. And now you have one.

2. Clean Up Your Algorithm (Before It Ruins You)

Your algorithm follows your behavior like a golden retriever. Whatever you click, it feeds you more of, whether that’s quality intel or unhinged theories cooked in a basement by someone wearing night-vision goggles indoors.

Curate your inputs:

• Unfollow emotional, partisan, or rage-inducing accounts

• Replace viral junk with OSINT analysts and defense reporters

• Add multilingual sources (Ukrainian, Polish, Baltic, Nordic)

One of the biggest advantages I had forecasting the 2016 election was that I listened to what Midwestern factory workers in Michigan were actually saying, not what coastal pundits imagined they were saying. Noise reduction is about removing filters, not adding them. For the record, I am not a Trump supporter.

3. Don’t chase the sensational. Chase the repeatable.

Here’s a rule from my AWACS days:

If it happens once, it’s interesting.
If it happens twice, it’s a pattern worth paying attention to.
If it happens three times, it’s someone’s doctrine.

Apply that to Russian missile strikes, Chinese naval movements, Iranian proxy attacks, or even Ukrainian drone production. Patterns shape reality. Individual events are usually just noise.

Example:
Russia hitting Ukraine’s civilian apartment buildings in 2022 was news.
Russia hitting them again in 2023 was a trend (Oh, I guess they are doing it on purpose).
Russia ramping up glide-bomb strikes on civilians in 2024 was just their doctrine.

Track doctrine, not drama.

The art and science of bias removal

Bias isn’t a flaw. It’s a feature of being human. But in forecasting, you treat it the same way you treat a malfunctioning compass: you don’t throw it away, you calibrate it.

My own bias: Ukraine deserves to win.

That’s not subtle. I write about Ukraine because I care about their survival. But when I forecast outcomes, I deliberately shift tone from passionate to clinical.

Forecasting mode Wes: Measured, numeric, emotionless.

Commentary mode Wes: “These Soviet ass hats couldn’t hit a stationary barn even if the barn was yelling coordinates.”

That’s intentional compartmentalization. And you need your own version of it.

How do you find your bias?

Ask:

• What outcome do I want to be true?

• Am I giving too much weight to sources that agree with me?

• Is my social feed an echo chamber?

Bias isn’t eliminated; it’s mapped and controlled. The goal is to be aware of it.

Think of it like the AWACS radar: you have blind spots whether you admit them or not. Better to know where they are.

Building the next-level OSINT workflow

A practical, step-by-step training plan for beginners and aspiring analysts.

This is the part I wish someone had taught me when I was learning the ropes, before I was sitting in a radar aircraft over the Pacific trying to decide whether a blip at 40,000 feet was a civilian jet or something more interesting, like a UAP!

Here is a full OSINT training pipeline you can start today.

STEP 1: Set up your monitoring architecture

Channels to build:

Daily intel stack

• Google Alerts (global)

• Twitter/X lists (curated OSINT analysts)

• RSS feeds (Defense Express, Janes, EuroIntegration)

• YouTube live cams (Black Sea, Baltic ports, border crossings)

• Telegram channels (curated through aggregators to avoid malware)

Weekly deep-dive

• The Economist defense section

• RAND Corporation reports

• CSIS briefings

• Ukrainian General Staff operational summaries

• Polish MOD updates

• Baltic security research publications

Monthly SITREP

• ACLED conflict maps

• IMF/World Bank macro indicators for high-risk states

• Energy and commodity flows

• Global sanctions tracker

• Military balance changes

Once you install these habits, the world stops feeling random. You start recognizing the difference between normal turbulence and actual historical change.

STEP 2: Learn source verification

You only need three levels:

Level 1: Visual verification

Is the image new? Recycled? AI?
Use:

• Reverse image search

• Sun position check

• Weather check

• Geolocation landmarks

• Metadata (if available)

Level 2: Source pedigree

Is this account reliable?
Unknown?
Known propagandist?
Repeat offender?

I keep a mental “reliability score” for every source I follow, a habit I picked up in the military.

Level 3: Cross-checking

One source is noise. Two similar sources are a coincidence. Three independent sources without coordination are a fact.

This rule will save you from 90% of online nonsense.

STEP 3: Learn to spot manipulation

Red flags:

• Emotional language (“genocide,” “obliterated,” “crushed,” “humiliated”)

• Claims without images despite modern ubiquity

• Perfectly framed footage (suspicious for combat)

• Video that appears slightly too clean or too chaotic

• Reports that break only in one language

• Claims from countries with strict media control

If something violates physics, logistics, or basic military doctrine, it is either false or Russian.

STEP 4: Practice micro-forecasting

Forecast small events first:

• “Will Russia launch a glide-bomb attack this week?”

• “Will Ukraine strike Crimea again before the end of the month?”

• “Will China send military jets across the Taiwan ADIZ threshold tomorrow?”

Assign probabilities, write them down, check your accuracy.

This is how superforecasters train.
It works.

STEP 5: Engage in community analysis

Teams outperform individuals.

When I was working with joint intel crews, we always cross-checked interpretations. One analyst catches the pattern, another catches the anomaly. No one sees everything.

Find a community:

• My YouTube channel comment section (which, by the way, is surprisingly supportive)

• OSINT Discords

• Substack discussions

• X/Twitter analyst circles

• Bellingcat volunteer groups

Share data. Compare notes. Compete. Improve.

The difference between OSINT and espionage

OSINT is powerful because:

1. More data exists in public than behind classified walls

2. Classified intel is often slower, because it must be protected

3. OSINT is adaptable and crowd-correcting

4. Governments lie, but physics doesn’t

5. A satellite image does not care who votes for whom

During the early days of Russia’s 2022 invasion, OSINT analysts found tank convoys, border buildups, unit rotations, abandoned gear, and missile trajectories days before government spokespeople addressed them. Not because the Pentagon was clueless. But because OSINT has no clearance process.

Your advantage? You are fast, flexible, and largely unburdened by bureaucracy.

Using OSINT to forecast the Ukraine War

This is where we combine everything:

probability + noise reduction + bias management + disciplined data flow.

Let’s model it.

How Those Inputs Actually Produce the Forecast (The Analyst’s Behind-the-Scenes)

Forecast percentages are not vibes. They’re not “I have a feeling” analytics. They are the product of structured, layered inputs processed the same way military intelligence shops, think tanks, and national forecasting teams do it.

Here is how the sausage gets made.

1. Manpower Shortages (Russia)

What it tells us:
Russia can sustain offensive pressure only if it can replace bodies. Current losses plus demographic collapse plus covert mobilization limits mean Russia is increasingly scraping the bottom of the barrel.

How it influences the model:

• Insufficient Russian manpower raises Ukraine’s recapture probability.

• Successful Russian mobilization (forced or otherwise) lowers it.

Current weight in model: +8% toward Ukrainian gains.

2. Ukrainian Long-Range Strike Success

This includes ATACMS, Storm Shadow, SCALP, homegrown kamikaze drones, and the new hit-rate against Russian logistics nodes.

What it tells us:
Russian supply hubs, fuel depots, airbases, bridges, and radars are getting hit deeper, more often, with growing accuracy.

How it influences the model:

• If strike success continues or expands, Russia’s ability to hold territory collapses proportionally.

• If Western restrictions tighten, Ukraine’s probability drops.

Current weight: +7% toward Ukrainian gains.

3. Western Aid Timelines

This is the single largest “swing variable” in the forecast.

What it tells us:
Ukraine’s combat power is chained to Western industrial output and political will.

How it influences the model:

• Stable aid = stable momentum

• Accelerated aid = increased probability of Ukrainian offensive success

• Disrupted aid = sharp probability drop

Current weight: –5% uncertainty penalty due to U.S. volatility but +10% upside from renewed EU ramp-ups and the PURL initiative.

4. Russian Domestic Political Stability

This includes Kremlin elite fractures, economic strain, regional dissent, and military mutinies.

What it tells us:
Russia is stable, but not secure. The system holds only as long as Putin’s patronage networks hold.

How it influences the model:

• Any shock to stability weakens Russia’s ability to sustain war.

• A stable Kremlin lowers the probability of Ukrainian breakthroughs.

Current weight: +2% (small but growing)

5. Ukrainian Drone Asymmetry

Ukraine is out-innovating Russia in drone warfare faster than Russia can adapt.

What it tells us:
The battlefield is now a drone-first environment, and Ukraine’s FPV, EW, and naval unmanned systems are reshaping the battlespace.

How it influences the model:

• Superior drone asymmetry gives Ukraine maneuver options even without tanks.

• If Russia closes the gap, probability decreases.

Current weight: +6% toward Ukrainian gains.

6. Weather Cycles

Mud, winter freeze, foliage cycles, snowpack, visibility windows. Classic “operational seasonality.”

What it tells us:
Ukraine can make territorial gains only when the ground, visibility, and mobility align.

How it influences the model:

• Favorable weather window increases offensive potential.

• Mud season and freeze/thaw cycles reduce it.

Current weight: –3% during heavy mud periods but +3% during hard freeze.

7. Attrition Rates

This includes equipment losses, artillery duels, drone consumption, pilot losses, and vehicle replacement rates.

What it tells us:
War is math. Whoever burns slower wins.

How it influences the model:

• Russia: high losses but deep reserves

• Ukraine: lower losses but shallow reserves

Current weight: Net –2% due to Ukraine’s limited replacement pool unless Western production ramps further.

8. Satellite Imagery of Fortifications

Public and classified-adjacent imagery shows Russia continuing to reinforce Zaporozhia, Donetsk, and Luhansk.

What it tells us:
Fortification density correlates with breakthrough difficulty.

How it influences the model:

• Dense Russian defenses reduce Ukrainian breakthrough probability.

• Successful deep strikes weaken the lines but do not erase them.

Current weight: –4% due to extensive Russian defensive layers.

9. Ammunition Production Curves

Artillery shells, air-defense missiles, FPV drones, and precision munitions.

What it tells us:
2026 is the first year Western production meaningfully outpaces Russian production in multiple categories.

How it influences the model:

• If production meets planned levels by Q2 2026, Ukraine gains a firepower edge.

• If supply chains lag, probability drops significantly.

Current weight: +8% upside potential but locked behind manufacturing timelines.

Putting It All Together

Each input feeds into a weighted probabilistic model that calculates:

Probabilities

The actual percentage chance of Ukraine regaining meaningful terrain in 2026.

Drivers

The strongest upward pressures on Ukrainian success:

• long-range strike effectiveness

• drone innovation

• Western ammunition output

Constraints

The downward forces:

• Russian defensive layering

• Western political volatility

• Ukraine’s limited manpower and airframes

Black Swans

Events that could swing the war dramatically:

• sudden Russian political instability

• a missile strike triggering NATO involvement

• catastrophic failure of Russian ammunition supplies

• unexpected collapse of a defensive sector

• U.S. election aftermath shifting aid

These do not get baked into the core model, but remain “scenario multipliers.”

The Current Forecast Explained

Baseline Probability (52%)

Given current aid levels, fortification density, winter cycles, and Ukrainian strike capabilities, Ukraine has:

a 52% chance of recapturing significant territory in 2026.

This is essentially a coin flip with a slight tilt toward Ukrainian success.

It assumes:

• slow but steady Western aid

• 18–20 operational F-16s

• stable Ukrainian drone production

• no collapse on the Russian side

Optimistic Scenario (68%)

If:

• F-16 operations scale to 24+ jets

• Gripens with Meteor missiles arrive by summer 2026

• PURL shells hit projected volumes

• ATACMS and Storm Shadow remain unrestricted

Then Ukraine’s ability to contest the air, strike logistics, and paralyze Russian staging points radically improves.

Result:
68% chance of meaningful territorial recovery.

This is still not guaranteed, but far more favorable.

Why This Matters

Forecasting is not prophecy. It is disciplined uncertainty.

You’re not “guessing,” you’re processing inputs, weighting variables, and updating as conditions change, exactly how:

• Allied intelligence cells

• NATO operational planners

• think-tank war-gamers

• financial risk analysts

do it every day.

It reinforces credibility and shows that you bring real analytic tradecraft to your analysis.

The ethics of modern OSINT

This field is powerful, but there are rules:

• Do not reveal Ukrainian troop movements

• Do not publish coordinates of active Ukrainian units

• Do not circulate unverified atrocity footage

• Do not amplify Russian propaganda you cannot debunk

• Do not enable targeting by hostile states

There’s a fine line between analysis and harm. I err on the side of protecting Ukrainian soldiers, even if it means letting some stories sit until they’re no longer sensitive.

Analysis without ethics is just voyeurism.

How AI fits in modern OSINT (and how to use it without getting played)

Artificial intelligence is transforming OSINT faster than Russia can lose a drone. And yes, I rely on AI tools every day; but not in the “AI will replace analysts” way. More like the “AI is a tireless intern who can sift through mountains of junk so I don’t have to” way.

But to use AI correctly, you need to understand its strengths, its blind spots, and the ways it can be manipulated.

This is where your human analysis and the machine’s brute-force horsepower merge into something that looks suspiciously like the future of intelligence work.

Let’s break it down.

1. AI is unbeatable at volume, terrible at judgment

Humans are good at nuance like pattern recognition, gut checks, “that’s weird” moments… but terrible at doing repetitive analysis for hours.

AI is the opposite. Give it a thousand Telegram posts, fifty PDFs, twenty satellite images, and a timestamp, and it will digest all of it without complaining or asking for caffeine.

What it cannot do is tell you whether a Russian source is trolling, whether a photo “feels” off, or whether a story is circulating because it’s true or because a bot farm is having a productive day.

So, here’s the rule:

AI should accelerate your analysis, not replace it.

2. Use AI to build your OSINT radar, not your OSINT conclusions

These tools are extremely powerful for:

• Keyword monitoring

• Pattern detection

• Language translation

• Timeline reconstruction

• Image sorting

• Document extraction

• Identifying missing data

AI is a force multiplier for search and triage, not a crystal ball.

Here are some specific, practical uses:

AI Keyword Sweeping

You can train AI tools to continuously monitor phrases like:

• “ПВО працює” (“air defense working”)

• “обстріл” (“shelling”)

• “новый удар” (“new strike”)

• “Севастополь взрыв” (“Sevastopol explosion”)

This essentially gives you a multilingual early-warning system. It helps that I speak Russian (poorly), read it better than I speak it, and understand it better than I read it.

AI Language Decoding

Forget Google Translate’s clunky guesses. Modern LLMs can parse Russian military slang, Ukrainian operational phrasing, and even the weird naval jargon from Baltic fishermen who keep spotting “mysterious objects” near pipelines.

AI Image Analysis

You can ask AI to:

• Compare images for geolocation clues

• Check whether two photos show the same vehicle

• Highlight heat signatures

• Flag inconsistencies that suggest manipulation

The catch? Never trust AI’s answer on image veracity without your own verification. AI hallucinates harder than a Marine who found a crate of expired MREs.

3. AI is a massive edge in pattern detection

If you feed AI:

• Daily missile strike logs

• Satellite imagery

• Naval movements

• Fuel depot fires

• Russian railway movement data

It will spit out correlations you might miss, such as:

• Surge in Russian rail logistics two days before glide bomb waves

• Iranian flights landing in Russia before drone factory spikes

• Weather patterns influencing Shahed launch timing

• Increased Russian naval activity before Kalibr launches

This is the analytic equivalent of adding another AWACS radar crew to your brain, minus the smell of desiccants and cold turkey sandwiches.

4. The danger: AI is extremely easy to manipulate

AI is also a megaphone for disinformation.

Bad actors can tailor content to exploit:

• Algorithm biases (which is why I never use AI for conclusions)

• Language gaps

• AI’s inability to detect cultural nuance

• Deepfake imagery

• Synthetic or hallucinated “evidence”

An image of a “Patriot launcher destroyed” can be generated in under 10 seconds and spread to 300,000 viewers before it hits your feed.

Which is why analysts must use AI like a sword, not a shield.

Your mental filters matter more now, not less.

5. The best OSINT analysts use AI like a crew chief uses tools

In the Air Force, crew chiefs are gods. They know every bolt on the jet. They use power tools, but the judgment is theirs.

That’s how AI should fit into OSINT:

AI = the toolbox
Human = the expert

Use the tools, keep the wisdom.

6. My current OSINT AI stack (you can steal this)

Here’s what I personally use:

Text & Document Analysis

• ChatGPT (pattern analysis, summaries, cross-language synthesis)

• Perplexity (fast discovery, linking sources)

• Claude (excellent for long-form contextual analysis)

Image & Video Tools

• Hive Moderation (deepfake detection)

• ExifTool (metadata extraction)

• SunCalc (sun angle verification)

• Sentinel Hub EO Browser (satellite comparison)

Monitoring Tools

• Google Alerts fed into AI summarizers

• Twitter/X list feeds parsed through AI for sentiment and clustering

• RSS feeds dumped into LLMs for daily global briefs

Predictive Modeling

• Simple Bayesian updating frameworks implemented through LLM prompts

• AI-generated probability grids for “if X happens, likelihood of Y increases”

• Monte Carlo-style event simulations using structured LLM prompting

It sounds complicated. It isn’t. It’s building a personal intelligence cell with a laptop.

7. AI forecasting: the hybrid model

This is where AI shines:

You give it general variables:

• Weapons production

• Weather

• Political stability

• Force posture

• Industrial output

• Previous attack timings

It generates pattern baselines.

You then adjust with human judgment: “This Russian unit is low morale, this politician is lying, this source is planted, this timeline is unrealistic.”

AI gives you the map. You determine the terrain.

Together, that creates the closest thing to real government intelligence forecasting civilians have ever had access to.

Where this all leads

Here’s the future analysts need to prepare for:

1. AI-generated propaganda floods the zone

2. AI-powered OSINT becomes mandatory

3. Human judgment becomes the limiting factor

4. Small teams outperform governments

5. The line between analyst and citizen disappears

The Ukraine War is the first conflict in history where civilians with laptops, satellite imagery, and AI analysis tools can outperform legacy intelligence bureaucracies on speed and accuracy.

The next war will push that even further.

And the analysts who thrive will be the ones who combine:

• disciplined workflows

• probability-based reasoning

• noise reduction

• bias control

• AI augmentation

• ethical guardrails

• community collaboration

That is the new tradecraft.

AI doesn’t replace human analysts. AI replaces bad analysts; the ones who rely on gut feelings, vibes, and whichever meme hit their feed this morning.

Good analysts? AI turns them into monsters. And I say that with admiration.

Putting it all together

Here is what you can do today to become a competent, disciplined global security analyst:

1. Build your OSINT dashboard

2. Curate your algorithm

3. Learn verification

4. Track patterns, not anomalies

5. Think in probabilities

6. Map your biases

7. Forecast small, then scale

8. Join analytical communities

9. Stay ethical

10. Keep learning

If you follow these steps for even three months, you will start predicting global events with shocking accuracy.

You will see the world differently. And the world will seem far less mysterious.

Today, the mission is to help people understand the most consequential conflict of our time.

OSINT is not just a hobby. It is a citizenship skill. It is a defense mechanism. And in a world drowning in misinformation and disinformation, it is a lifeline.

The goal is not perfection. The goal is clarity.

One final note… After I forecast the future, I add back in my own opinion, bias, sarcasm, humor, and conversational tone into all of my writing and videos. That’s the real secret sauce and why (I believe) people enjoy subscribing.

If you found this guide useful, please share it. And if you want more long-form, unfiltered analysis like this, consider joining my paid Substack community at Eyes Only. Paid subscribers receive two additional, exclusive articles every Wednesday and Sunday.

Because the war is not slowing down. The world is not getting quieter. And the future belongs to those who can separate signal from noise.

Stay frosty.

Слава Україні!