Fragmented truth
Each vendor weighs its own sensor quality. The same scene reads differently on every console in the room.
Defense AI SolutionsFrom scattered sensors
From scattered sensors
to one picture that decides.
Tiktaalik builds the AI layer for command and control. It fuses any vendor’s sensor data into a single correlated picture and cues the right response, with a human in the loop and an audit trail on every decision.
Built for the interfaces you already run: SAPIENT, Cursor on Target, Remote ID, and vendor SDKs.
Sensor to cueConcept view
IN
Sensors
EO/IR, radar, RF, acoustic. Any vendor.
CORE
Fusion
Ingest, weight, correlate.
PIC
Picture
One object, one track.
OUT
Cue
Right effector, right second.
Human in the loopAudit trailOperator overrideRules of engagement
Sensors argue. Seconds cost.
The space between is where the mission is won or lost.
The space between is where the mission is won or lost.
The operating floor
Every console tells a different story. The operators in between carry the cost.
Photo — ANGIE BAONGOC, UnsplashEvery block is world class. The layer between them is missing.
Command and control has excellent parts. What it lacks is the intelligence that connects them into one automated loop.
Ghost tracks
One sensor's false alarm becomes a track. The same object over three sensors becomes three threats.
The human bottleneck
Under a fast, many object threat, decision making stays manual. Operators become integrators, and hesitation costs the mission.
Why it matters. In counter drone defence the distance from detection to defeat is measured in seconds. Operators do not need more feeds. They need a picture that has already decided.
One loop, from sensor to shooter.
Tiktaalik adds the connective layer: a pipeline that ingests everything, fuses it honestly, and cues the right effector at the right second.
Plug in, do not replace
Runs on the command and control you already operate, over open standards. Nothing gets thrown away.
Fusion with provenance
Sensor quality metadata flows into every decision, so real signals separate from noise.
Ranked by confidence
Operators see the highest value target first, not the loudest alarm.
Automated, never autonomous
Every cue carries an audit trail and an operator override, under rules of engagement you configure.
Proof in weeks
A fixed scope pilot with agreed success definitions, on an indicative eight week path.
Ingest
Any vendor's sensors arrive over SAPIENT, Cursor on Target, Remote ID, and vendor SDKs.
Fuse and correlate
One object, one track, weighted by each sensor's real quality.
Decide and cue
Ranked threats become effector cues, with a human gate on every action.
A toolkit that closes the loop.
Engineered to drop into an existing ecosystem. Each service stands alone. Together they form the automated loop.
Sensor fusion
Vendor neutral ingestion into open command and control standards, so every sensor in your estate arrives in one dialect.
Digital twin
Simulate sensor placement, coverage, and threats before you commit hardware.
Provenance metadata
Quality weighting that keeps clutter and false inputs out of the picture.
Track correlation
One object, one track, even across overlapping sensors.
Automated cueing
From detection to defeat, orchestrated across your effectors.
Rules of engagement
Cueing gates that keep automation inside your doctrine.
Swarm scenario injection
Train operators against realistic many object threats.
After action debrief
Replay and forensics for every decision, including the ones not taken.
A fixed scope, then proof.
We agree success definitions up front, then demonstrate them on your picture. An indicative path to a working prototype:
Align and scope
MVP scope, interfaces, and success definitions agreed together, against your picture.
Build
The preliminary MVP, end to end, on your data.
Integrate and test
Integration and testing inside your environment.
Demo
A working prototype demonstration of the loop.
You keep your command and control. The first engagement is a fixed scope pilot with agreed success definitions, so you see exactly what you are buying before it starts. The timeline shortens or lengthens with the scope we agree.
Proof policy: no invented logos or quotes on this page. When customers allow us to name them, proof will replace promises.
The questions we get before the first meeting.
Q01Are the existing C2 blocks not enough?
The blocks are excellent. What is missing is the layer that connects them: fusion, correlation, prioritization, and cueing. We do not compete with your command and control. We make it work as one system.
Q02Which interfaces do you support?
Open standard ingest over SAPIENT, Cursor on Target (ATAK), and Remote ID, plus vendor SDKs when a source needs one. The scoping phase locks the list for your estate.
Q03Where does the human sit in the loop?
In the middle. Every automated decision is auditable and overridable. Cueing runs under rules of engagement you configure. The layer is automated by design, and autonomous only where your doctrine allows.
Q04How do you handle security and data?
The layer deploys inside your environment, where your C2 already runs. We follow your security review process, and specifics are handled under NDA.
Q05How fast can we see something real?
We agree success definitions in the first two weeks, then run a pilot toward a working demo on an indicative eight week path. The timeline tracks the scope we agree.
Q06Why the name Tiktaalik?
Tiktaalik roseae is the transitional fossil, a fish with limbs that crossed from water to land. The name is the job: helping sensor networks cross from separate feeds to one autonomous loop.
The end state
Fused, ranked, and ready. A floor that acts as one system.
Photo — Tetiana Sapon, UnsplashFrom scattered sensors
to one picture that decides.
Tell us about your picture. We will scope the layer that makes it decide.
Replies from the founding team at team@tiktaalik.ai. For pilots and support, see the contact page.