Gatik has rebuilt its public website around something almost no operating company does voluntarily: a live board of its own trucks, refreshed every three hours, showing start times, end times, driving hours, stop progress and status for every driverless trip in a rolling twenty-four hour window. The line above the table reads “Everything below is live. These are real trips from our operations, refreshed every 3 hours.”
That is a striking thing for an autonomous trucking company to publish. The usual marketing site for this sector is a hero video, a safety page and a contact form. Gatik has replaced the middle of its homepage with an operations dashboard, moved its entire old marketing site to a separate archive domain, and left the news ticker at the top pointing at video coverage from Bloomberg, NBC and the Wall Street Journal rather than at its own press releases.
The timing is not accidental. On 25 August 2026 the company announced a $200 million Series D led by the Qatar Investment Authority and Koch Disruptive Technologies. A funding round is a claim. A live feed of 132 trips is a receipt. If you track how operational tooling and computer vision systems reach production, our AI models, tools and releases hub covers the same pattern across other categories, and this one is a particularly clean example of a company deciding that showing beats telling.
What follows is a full read of what the board actually publishes, what it deliberately withholds, what a single real snapshot of the underlying feed contains, how it compares to how the rest of the autonomous vehicle industry handles transparency, and what the whole exercise costs the company that attempts it.
Table of contents
- What Gatik Actually Put Live
- What The Live Operations Board Shows
- Reading One Real Snapshot Of The Feed
- What Gatik Deliberately Does Not Publish
- Why Gatik Is Publishing This Now
- How This Compares To The Rest Of The Industry
- What It Costs To Publish A Live Board
- What Businesses Should Take From This
- Your Next Steps
- Frequently Asked Questions
- References
What Gatik Actually Put Live
Start with the thing itself, because the description matters more than the announcement.
The homepage is now an operations board
Load gatik.ai and the page has four parts: a hero video with the line “The business of autonomous freight”, a news ticker of third-party coverage, a one-sentence statement of intent, and then the board. The board occupies the centre of the page. It is not a demo, not an illustration, and not a screenshot — it is a table populated from Gatik’s production systems, with a countdown timer showing how long until the next refresh.
The old marketing site moved to an archive
The clearest signal of how deliberate this was is where everything else went. Gatik’s About, Safety, Customers, News and Careers pages now resolve to archive.gatik.ai, a separate domain carrying the previous website almost intact. The main domain was cleared to make room for the live board. Companies do not casually relocate their entire content estate to an archive subdomain; that is a decision taken at the top.
What “live” means here, precisely
“Live” is doing careful work in that sentence. The board is not a real-time telemetry stream. It is a snapshot regenerated every three hours, and the page is honest about it — the header carries a countdown, and the underlying data carries an explicit refreshIntervalHours value of 3. The window covered is twenty-four hours wide, so a visitor sees a full operating day: trips already finished, trips in progress, and trips still scheduled to run.
Where the videos sit
The video half of the story is quieter than the tracking half. The hero itself is a looping muted clip of the trucks. The news ticker links out to three pieces of video and print journalism — a Bloomberg Tech segment on the $600 million contracted revenue figure, an NBC TODAY ride-along asking whether driverless freight trucks are the new way to move goods, and the Wall Street Journal’s report of a 26,000-pound box truck of Doritos leaving a distribution centre with nobody at the wheel. Gatik is pointing at other people’s cameras rather than its own.
What The Live Operations Board Shows
The table is narrow on purpose. Six columns, and each one earns its place.
| Column | Example value | What it actually tells you |
|---|---|---|
| Truck | G-009A | A display label for the row, not a fleet asset number |
| Start Time | 1:14 AM | When the trip began, shown in Pacific time |
| End Time | 11:15 AM | Actual or projected completion of the whole run |
| Driving Time | 7:18 hrs | Wheels-moving time only, excluding dwell at stops |
| Stops | Six dots, five filled | Progress through the drop sequence, as a dot per stop |
| Status | Driving / On Time | Current activity plus a schedule verdict |
The status vocabulary is operational, not promotional
The activity labels a row can show are Driving, Loading, Unloading, Parked and Completed. The schedule verdict beside them is On Time, Delayed or Completed. There is no “optimal”, no “nominal”, no invented adjective. Anyone who has watched a dispatch screen in a distribution centre will recognise every one of these words, which is the point — Gatik chose the vocabulary of the customer’s own control room rather than the vocabulary of a pitch deck.
The hover card adds four more fields
Hovering a row, or tapping it on a phone, opens a small card with four additional fields: the distance to the next stop in miles, the driving duration in decimal hours, stops completed out of stops today, and whether the load is refrigerated. That last one matters more than it looks. Refrigerated freight is the hard case for autonomous middle-mile work, because a delay is not merely late — it is spoilage.
Every row carries a public trip identifier
Each row in the page markup carries a nine-character public trip ID such as EEACGPL8N or WY9HEWPWU. These are opaque handles, not sequential numbers, and they are stable within a snapshot. Their presence is what turns the board from a picture into a record: two people looking at the same trip can name it identically, and a claim made about a specific run can be checked against a specific row.
Reading One Real Snapshot Of The Feed
The board is paginated with a “Load more” button, and the whole set is available as structured data. Here is one complete snapshot, generated at 17:56 UTC on 25 August 2026, read in full.
| Measure | Value in this snapshot | Note |
|---|---|---|
| Trips in the window | 132 | Rolling 24 hours, not a calendar day |
| Refresh interval | 3 hours | Stated explicitly in the data, not just the copy |
| Time zones represented | 2 | 99 trips Central, 33 Arizona |
| Trips flagged delayed | 1 | The other 131 read on time, arrived or scheduled |
| Total driving time | 15,912 minutes | 265.2 hours across the window |
| Median trip driving time | 55 minutes | Mean is 120.5 minutes — the tail is long |
| Longest single trip | 468 minutes | 7.8 hours of driving on one run |
| Total stops scheduled | 607 | Averaging 4.6 stops per trip |
| Published inter-stop legs | 475 | 4,846 miles of leg distance in total |
| Loads marked refrigerated | 0 | In this snapshot only; the field exists for all rows |
A hundred and thirty-two trips is a real operation
The headline number is the one that resists spin. A hundred and thirty-two trips in twenty-four hours is not a pilot. It is also not yet a fleet of hundreds, which is what the company has said it is building toward. Both readings are available from the same table, and that symmetry is the tell that this is data rather than marketing — a promotional dashboard would not let you count the ceiling as easily as the floor.
Where the trips are running
Ninety-nine of the 132 trips carry a Central time zone and 33 carry Arizona time. That maps cleanly onto the states Gatik has said it operates in — Texas and Arkansas on Central, Arizona on its own clock — without the company having to name a single customer site. It is a good example of publishing the shape of an operation without publishing its coordinates.
The one delayed row is the whole argument
Exactly one trip in 132 carried a delayed status. That is 0.8%, which sits comfortably behind the “99% on-time delivery” figure in the Series D announcement and the “more than 98%” figure quoted in the PepsiCo release ten weeks earlier. The important part is not that the number is good. It is that the delayed row is displayed at all, in the same typeface as the rest, on the company’s own homepage.
Trips are short, and the distribution is lumpy
The median trip in this snapshot involves 55 minutes of driving; the mean is 120.5 minutes. When the mean is more than double the median, a small number of long runs is doing the pulling — and indeed the longest single trip carries 468 minutes of driving, nearly eight hours. Middle-mile freight is not one uniform job. It is a pile of short shuttles with a handful of long hauls threaded through it.
How the operational day breaks down
Grouping the 132 trips by phase gives 65 completed, 42 still scheduled, 16 in progress, 7 unloading and 2 loading.
The legs are shorter than the word “trucking” suggests
Across the 475 published inter-stop legs, total distance comes to 4,846 miles, giving a mean leg of 10.2 miles and a median of 7.5. Sorting those legs into distance bands gives 172 under five miles, 174 between five and ten, 77 between ten and twenty, 41 between twenty and forty, and 11 above forty miles — with the longest single leg at 77.5 miles.
Seventy-three per cent of legs are under ten miles. That is the middle mile as an engineering problem: not long-haul cruising, but dense, repetitive, surface-street work between a distribution centre and the stores around it.
What Gatik Deliberately Does Not Publish
The omissions are as considered as the inclusions, and reading them is the fastest way to understand the design.
No customer names and no addresses
Not one row names a retailer, a grocer, a distribution centre or a store. The company’s separate announcements name PepsiCo and describe Fortune 50 retail and grocery customers, but the operational board keeps every commercial partner anonymous. A customer’s replenishment cadence is competitively sensitive information about that customer, and publishing it would make the board unshippable.
No coordinates and no route geometry
There is no map. There are no latitudes, no longitudes and no polylines. The only spatial information published is the distance from the previous stop to the next one, in metres, per leg. You can reconstruct how far a truck travels between drops. You cannot reconstruct where it went, which is precisely the line a company can publish behind without exposing either its customers or its trucks.
Truck numbers are labels, not identities
The G-001A style identifiers look like fleet asset tags, but they are generated by row position in the table rather than read from the vehicle. Two rows numbered G-004A in different snapshots are not necessarily the same physical truck. This is easy to misread, and it is worth naming plainly: the board discloses trips, not vehicles.
Refrigerated is the only cargo attribute
Each trip carries a single cargo flag for whether the load is refrigerated. Nothing else about the freight is exposed — no weight, no commodity, no case count, no value. One boolean is enough to signal that Gatik is running temperature-controlled work, and small enough that it reveals nothing about any individual shipment.
Why Gatik Is Publishing This Now
Three things converged in the same fortnight, and the board is the visible product of all three.
The Series D changed the audience
On 25 August 2026 Gatik announced $200 million in Series D financing led by the Qatar Investment Authority and Koch Disruptive Technologies, with Millennium Management, ARK Invest and Intact Private Capital participating. Announcements of that size bring institutional scrutiny. A live board answers the first question those investors’ analysts will ask — is any of this actually running — before it is asked.
The claims on the site and in the release agree
The Series D release states more than $600 million in contracted revenue, 85,000 fully driverless orders completed and 99% on-time delivery. The live board is a continuous, independently checkable sample of the third of those. Nobody has to take the on-time claim on trust when a rolling day of trips is sitting on the homepage with a delayed column that is capable of filling up.
The 2026 credibility gap in autonomy is real
Autonomous freight has spent a decade producing demonstration videos, and buyers have learned to discount them heavily. Gautam Narang, Gatik’s chief executive and co-founder, put the counter-position bluntly in January: “Autonomous trucking is no longer a promise. It’s a business.” A homepage that reads like a control room rather than a brochure is that sentence rendered as a design decision.
Video does the work a statistic cannot
The board proves consistency. It cannot prove that a 26,000-pound truck with an empty cab is an ordinary sight on a public road. That is what the video coverage is for, and it is notable that Gatik surfaces third-party footage — Bloomberg Tech, NBC’s TODAY, the Wall Street Journal — instead of its own edit. Journalism a company did not control is worth more than a showreel it did.
How This Compares To The Rest Of The Industry
Autonomous vehicle transparency already exists. What Gatik has done is pick a different axis of it.
| Approach | What gets published | Update cadence |
|---|---|---|
| Gatik live operations board | Every trip in a rolling 24 hours, with status and stop progress | Every three hours |
| Waymo Safety Impact hub | Crash rates per million rider-only miles against human benchmarks | Aligned to regulatory reporting cycles |
| NHTSA Standing General Order | Incident reports for crashes involving automated driving systems | Periodic CSV releases |
| Third-party safety assessment | Independent review of a published safety case and framework | One-off, per milestone |
| Conventional AV marketing | Curated video, cumulative mileage totals, customer logos | Whenever it flatters |
Safety data and operations data answer different questions
Waymo’s hub reports that its driver was involved in 92% fewer serious-injury crashes than human drivers over 220.6 million rider-only miles through March 2026. That answers “is it safe?” It does not answer “does it turn up?” Gatik’s board answers the second question and says almost nothing about the first. A freight buyer needs both, and until now the second had no public source at all.
The regulator already collects the bad news
Under NHTSA’s Standing General Order, operators of automated driving systems must report crashes on public roads, and the agency publishes the resulting incident data. That means the failure side of the ledger is already, to a degree, a matter of public record. Publishing the routine side voluntarily is what changes the ratio — otherwise the only public data about an autonomous fleet is the data about its worst days.
A live feed proves what a PDF cannot
A safety report is a document about the past, assembled by the company describing itself. A live board is a standing invitation to check. Anyone can open it tomorrow, or next month, or on a bad weather day in Texas, and see what the fleet did. That difference in kind is worth more than any single figure printed inside it.
What It Costs To Publish A Live Board
This is the part most organisations underestimate, and it is not the engineering.
The build is the easy half
Behind the page is an ordinary, well-behaved data path: a snapshot regenerated on a three-hour schedule, served as structured records, with a version stamp and generation timestamp attached so a stale copy can be detected. The page renders the first rows server-side and pages the rest. Any competent team with access to its own dispatch system could ship the equivalent in weeks. Nobody is blocked by the technology.
Publishing a number you cannot retract
The hard half is governance. Once a delayed count is on the homepage it is on the homepage on the bad days too, and the internal pressure to add a filter, extend the window or quietly redefine “on time” arrives immediately. Surviving that pressure requires a decision, made in advance and at a senior level, that the definition does not move. Everything else is a rendering detail.
The failure mode is silence, and it is visible
A live board fails loudly. If the snapshot stops refreshing, the countdown reveals it. If a bad day happens, the delayed rows show it. There is no graceful degradation into an old screenshot, because the timestamp keeps advancing whether the data does or not. That is a genuine operational risk, and accepting it is the actual commitment being made here.
Three hours is a considered compromise
The refresh interval is worth reading as a design answer rather than a limitation. Real-time streaming would expose fleet positions and invite scraping of customer patterns. Daily updates would be indistinguishable from a report. Three hours is frequent enough that a visitor sees the day change, and coarse enough that no individual truck can be followed in anything close to real time.
What Businesses Should Take From This
You are almost certainly not running driverless trucks. The transferable part is the method, and it applies to any organisation whose service quality is currently a claim.
Pick the metric you would rather hide
The board works because the delayed column exists. A dashboard that can only look good is decoration, and readers detect it instantly. The useful exercise is to identify the single operational number your customers ask about most and you disclose least — resolution time, uptime, delivery accuracy, backlog age — and publish that one, with its bad values intact. For most teams that number already sits inside a reporting stack built for internal eyes, which is where data analytics work usually has to start.
Publish the shape, not the secrets
Gatik’s design shows that the choice is not binary. Distances without coordinates, statuses without customers, trip identifiers without vehicle identities: each is a deliberate reduction that keeps the signal and drops the sensitivity. Most organisations that decide against transparency have compared full disclosure to none, and never considered the shape in between. Approaches like this pair naturally with workflow automation work, because the same pipeline that moves the data internally can feed the public view.
Set a cadence you can survive on your worst week
A refresh interval is a promise. Choose one you can meet during an incident, a holiday period and a staffing gap, because those are the weeks when a stalled dashboard does more reputational harm than never having built it. Three hours works for Gatik because a three-hour gap in a twenty-four hour window is tolerable. Your number will be different, and it should be chosen pessimistically.
Make it machine-readable from day one
The Gatik board is backed by structured records with stable identifiers, not by a rendered image. That means the data can be read, checked, compared over time and cited precisely. If you publish a picture of your numbers, you have published a claim. If you publish the records, you have published evidence — and the difference matters increasingly as data visualization becomes something machines consume as often as people do.
Your Next Steps
If you are evaluating autonomous freight, open the board and watch it across several days rather than once. A single snapshot tells you the fleet exists. A week of snapshots tells you whether the delayed count is stable, whether trip volume holds through a weekend, and whether the refresh promise is kept — which is far more informative than any figure in a press release.
If you are not in freight, run the exercise on your own organisation. Name the operational metric you would least like to publish, work out what a reduced, non-sensitive version of it would look like, and estimate what it would take to refresh it on a fixed schedule. Most teams discover the data already exists internally and the blocker is entirely a matter of nerve.
Either way, the thing to watch next is not whether Gatik’s fleet grows. It is whether the board survives contact with a genuinely bad week — and whether any competitor decides that matching it is now the cost of being believed.
Frequently Asked Questions
What is Gatik’s live tracking page?
It is a table on the gatik.ai homepage listing every driverless trip in a rolling twenty-four hour window, with the truck label, start and end times, driving time, stop progress and current status for each one. It is populated from Gatik’s production operations and refreshed on a three-hour cycle.
Does Gatik show live video of its trucks?
Not of individual trips. The homepage carries a looping hero clip, and the news ticker links to third-party video coverage including a Bloomberg Tech segment and an NBC TODAY ride-along. The per-trip data is tabular, not visual — there is no camera feed attached to a row.
How often does the board refresh?
Every three hours. The interval is stated in the underlying data as well as in the page copy, and the header shows a countdown to the next update alongside a Pacific time zone label.
What does Gatik publish for each trip?
Start and end times, driving duration in minutes, the total and completed stop counts, an operational status and phase, the time zone, a refrigerated cargo flag, a public trip identifier, and the distance in metres between consecutive stops. No customer, location, coordinate or route data is included.
Where do Gatik’s driverless trucks operate?
Texas, Arizona, Arkansas and Canada, moving freight between distribution centres and stores for Fortune 50 retailers, grocers and consumer goods companies. In the snapshot examined here, 99 of 132 trips ran on Central time and 33 on Arizona time.
How much driverless work has Gatik completed?
The company reported 85,000 fully driverless orders completed alongside its August 2026 Series D, up from 60,000 orders, 2,000 operating hours and 10,000 driverless miles reported in January 2026, with more than $600 million in contracted revenue and a stated on-time delivery rate of 99%.
References
Gatik — The Business of Autonomous Freight (live operations board)
Gatik Raises $200 Million; Series D Led by QIA and KDT — 25 August 2026
First U.S. Company to Operate Fully Driverless Trucks at Scale — 27 January 2026
PepsiCo and Gatik Announce Multi-Year Agreement — 8 June 2026
Industry-First Third-Party Safety Assessment Framework — Gatik
Are Driverless Freight Trucks the New Way to Transport Goods? — NBC TODAY
Driverless Trucking Firm Signs Major $600M Deal — Bloomberg Tech
Driverless Trucks Are Here — And They’re Delivering Bags of Doritos — Wall Street Journal
Gatik launches fully driverless commercial trucking operations — FreightWaves
New $200 Million Investment For Its Driverless Truck Tech — Forbes
More AI coverage: explore Progressive Robot's AI Models, Tools & Releases hub — hands-on reviews, setup guides and benchmarks in one place.