NewSarissa v2 — Street-level precision

The visual intelligence
platform for analysts and agents

Purpose-built for geolocation from images.
Designed for the intelligence era.

SRS-2703Located

High-confidence geolocation from street-level photograph

Render location candidates before full feature extraction is complete, instead of blocking on exhaustive cross-referencing during initial analysis.

Activity
S
Sarissa created the analysis from image upload on behalf of karri · 2min ago
T
Triage Intelligence added labels ArchitectureUrban · 2min ago
karri· 4 min ago

The image shows distinctive Haussmann-style architecture and what appears to be French street signage...

jori· just now

@Sarissa can you run a deep analysis on this one?

S
Sarissa AI
$Examining image SRS-2703
$Extracting visual features...
$Detected: Haussmann architecture (96%), French signage (92%)
$Cross-referencing 48,000 visual signatures...
$Match found: Paris, 7th Arrondissement
Status
Located
Confidence
94%
Priority
High
Analyst
jori
Agent
S
Sarissa AI
Labels
ArchitectureUrbanEurope
Coordinates
48.858°N, 2.294°E
Region
Western Europe
Analysis Time
1.8s

A new species of intelligence tool. Purpose-built for visual geolocation with AI at its core, Sarissa sets a new standard for image-based reconnaissance.

FIG 0.2

Visual precision

Sarissa is shaped by the demands of real-world intelligence operations. Every feature exists because analysts need it.

FIG 0.3

Powered by AI agents

Designed for workflows shared by humans and AI. From uploading images to verifying locations with autonomous analysis.

FIG 0.4

Designed for speed

Reduces noise and restores momentum to help teams locate targets with high velocity and focus.

1.0Upload

Make intelligence workflows autonomous

Turn field images and surveillance footage into actionable intelligence that is analyzed, labeled, and geolocated for the right team.

Queued8
SRS-2085

Analyze satellite imagery of coastal...

Satellite
SRS-2094

Process drone footage from sector 7

Drone
SRS-2092

Batch upload from field operation

Field
SRS-2200

Cross-reference port surveillance...

Maritime
Analyzing3
SRS-926

Identify landmarks in urban scene

UrbanAI
SRS-2088

Feature extraction: architecture match

Architecture
SRS-924

Deep analysis with terrain model

Terrain
Located3
SRS-1487

Paris — 7th Arrondissement

94%61039
MKT-1028

Tokyo — Shibuya crossing

Verified91%
SRS-2187

São Paulo — Paulista Avenue

87%62048
Verified53
SRS-2074

London — Westminster Bridge

98%61002
SRS-1912

Cairo — Giza Plateau

96%61005
SRS-1951

New York — Times Square

99%61202
SRS-1960

Berlin — Brandenburg Gate

95%61149
didier

Has anyone been looking into the coastal surveillance batch?

lena

Anyone else noticing the architecture matching seems slower on the latest batch?

didier

Yeah, we're still running full feature extraction on every image before showing initial candidates...

andreas

We could show preliminary location candidates and refine in the background. Also worth tracking analysis timing so we know how often this happens!

@Sarissacreate urgent issues and assign to me
2.0Locate

Pinpoint any location on Earth

Plan and navigate from image to coordinates. Track analysis campaigns, regional coverage, and up-to-date visual databases.

FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
Architecture
Signage
Satellite
Elevation
Alpha
GA
3.0Analyze

AI reasoning that shows its work

Deploy AI agents that work alongside your analysts. Work on complex geolocation tasks together or delegate entire analyses end-to-end.

On it! I've received your request.
Kicked off analysis in sarissa/visual-engine environment.
Searching for architectural signatures
sarissa/visual-engine$ rg --features -g "haussmann" signatures.db
→ 847 matches across 12 regions
Extracting terrain elevation profile
Thought for 5s
Agents Command Menu
No results found.
S
Sarissa AIAgent
Steven
V
Visual EngineAgent
T
Terrain ModelAgent
F
Field AnalystAgent
Meg
4.0Verify

Confirm with ground-truth imagery

Understand visual changes at a glance with structural diffs for human and AI output. Review, compare, and verify — all within Sarissa.

sarissa/core/src/engine/geolocate.ts
sarissa/core/src/engine/geolocate.ts
1import { analyzeImage } from '@core/engine'
2import { FeatureExtractor } from '@ai/features'
3import { LocationDB } from '@data/locations'
4
5export const geolocate = async (image: Buffer) => {
6 const features = await FeatureExtractor.run(image)
7 if (!features.isComplete) {
8 return { status: 'pending' }
9 }
10 return await LocationDB.match(features)
11}
1import { analyzeImage } from '@core/engine'
2import { FeatureExtractor, ConfidenceLevel } from '@ai/features'
3import { LocationDB } from '@data/locations'
4
5export const geolocate = async (image: Buffer) => {
6 const features = await FeatureExtractor.run(image)
7 if (features.confidence === ConfidenceLevel.LOW) {
8 return { status: 'low_confidence' }
9 }
10 return await LocationDB.match(features)
11}
5.0Monitor

Understand accuracy at scale

Take the guesswork out of intelligence operations with analysis updates, accuracy metrics, and dashboards that surface what needs attention.

181410620
Feb 2025May 2025Aug 2025Nov 2025
Sarissa AI
Manual
Unassisted
Projects
Urban analysis upgradeAt risk

Architecture matching nearly complete, but terrain model updates still WIP

By romain · 1 day ago
APAC coverage expansionOn track

Japan and South Korea visual databases have been integrated

By julian · 3 hours ago

Changelog

Deep-link to AI analysis tools

Starting an analysis used to mean manually uploading to each tool. Now it means assembling the right context so Sarissa can take a first pass. We've made this much easier.

Feb 26, 2026

Advanced filters and batch exports

Refine your searches, views, and dashboards with advanced filters. Combine multiple AND/OR conditions to define exactly what you want to see.

Feb 12, 2026

Terrain model v3 for mountainous regions

We've upgraded our terrain matching engine with high-resolution elevation data. Mountain passes, ridgelines, and valley profiles now match with 3× higher fidelity.

Feb 4, 2026

Time-in-status tracking

Time spent in each analysis stage is now tracked and visible. Hover over any status indicator to see how long an image has been in queue, analysis, or review.

Jan 28, 2026

You just have to use it and you will see, you will just feel it.

S
Sarah Chen
Lead Analyst, Federal Agency

Our speed is intense and Sarissa helps us be action biased.

M
Marcus Webb
Head of Intelligence, Defense Contractor

Sarissa is excellent, just excellent. It has the right opinions for fast moving teams.

E
Elena Volkov
OSINT Director, Security Firm

Sarissa powers over 200 intelligence teams. From specialized units to national agencies.

Built for the field.
Available today.