Tech & AI Oct 18, 2026 · 7 min read

Evos raises $9m to put AI “operators” in freight back offices

Evos gets $9m to automate broker back offices. The proof so far is thin. New York startup Evos has raised $9 million in pre-seed funding for AI “operators” that do back-office work such as invoice audit, FreightWaves reports. Our view: the money follows labour budgets, not software budgets, and the claims still need checking. Evos says one operator made nearly 3,500 audit decisions in two months at 97% accuracy for broker MyFreightWorld. FreightWaves notes Evos did not explain how accuracy was measured. For forwarders, audit and status chasing are the first jobs at risk. Our call: by end of Q1 2027, buyers will demand audited, dollars-recovered numbers before they sign.

Freight’s back office is the new AI battleground, and $9m for Evos shows investors are chasing payroll Evos, a New York startup that has just come out of stealth, has raised $9 million in pre-seed funding. Pre-seed is the earliest stage of venture money, usually before a product is fully launched. Uncork Capital and Field Ventures co-led the round. Crosslink Capital, Utopia Capital Management, Sandwith Ventures and angel investors joined, FreightWaves and The Conveyor reported on 16 October. TechCrunch’s Mobility newsletter picked up the deal on 18 October.
Evos builds what it calls AI “operators”. These are software agents that learn a back-office job from experienced staff and then do it. Its first public customer is a freight broker. Its target list reads like the job descriptions of a forwarder’s operations floor.

What we think it means

Our view is simple. Investors are no longer funding freight software. They are funding freight labour.
Small brokers and forwarders do not have big IT budgets. They do have big payrolls. A product that replaces hours of invoice checking or status chasing gets paid out of the payroll line, and that line is much larger. That is the bet behind Evos, and behind a run of similar deals this year.
We think the bet is right. But the evidence Evos has published so far is company-reported, unaudited and, in one key case, easy to misread. Buyers should treat the headline numbers as a sales pitch until they see their own data.

What Evos actually does, and what it claims

The Conveyor describes the product this way. Evos operators learn how experienced staff do a job, then do it inside tools a firm already uses. That means a TMS (transportation management system, the software that books and tracks loads), SAP, Salesforce or Slack. The customer sets approval and escalation rules, so the operator knows when to hand a decision to a human.
FreightWaves adds how the hand-over works. The operators learn from employees’ decisions. Once their decisions consistently match those of experienced workers, they take on the function on their own.
The Conveyor lists the use cases as carrier sourcing, dispatch, driver compliance, freight invoice audit, customs classification and shipment-status queries. Every one of those is a job a forwarder or broker staffs by hand today.
Founder and CEO Urav Shah pitches the product around tribal knowledge. Of the long-serving staff who run small firms, he told FreightWaves that “what they know has never been written down”.
Here are the company’s numbers, all as reported by Evos:

Annual recurring revenue (ARR, the yearly value of subscription contracts) passed $1 million four weeks after launch, per FreightWaves. Evos did not give a customer count or valuation.

At broker MyFreightWorld, an operator auditing LTL (less-than-truckload, shared-trailer freight) carrier invoices “made nearly 3,500 audit decisions over two months with 97% accuracy”, per FreightWaves.

The same operator “identified billing discrepancies valued at more than $600,000 over the most recent five-week period”, FreightWaves reports.

Mike Chalfant, president of MyFreightWorld, said Evos helped overhaul its billing process “with zero upfront investment”, per The Conveyor.

Evos plans to formally launch its platform, Workhorse, “in the coming weeks”, The Conveyor reports.

Read the small print on the numbers

Two of those figures need care.
First, the 97%. FreightWaves says Evos did not explain how it calculated that accuracy rate or what sample it used. Accuracy against what? If it means “agreed with a human auditor 97% of the time”, that is useful. If it means “97% of invoices it touched had no later complaint”, that is a much weaker test. We do not know which.
Second, the $600,000. FreightWaves says this figure reflects discrepancies between invoices, not money recovered or saved. The Conveyor describes the same number as invoice value closed out. Either way, it is not cash back in the broker’s account. A flagged overcharge only becomes a saving once the carrier agrees and credits it.
Third, scale. Nearly 3,500 decisions over two months is roughly 80 a working day. That is a real job for one auditor. It is not yet proof of a whole back office.
None of this means the product does not work. It means the public proof is a single customer, measured by the vendor.

Why invoice audit goes first

Evos chose a good first target. LTL invoices are messy, and the errors follow patterns a machine can learn.
FreightWaves cites a 2024 AFS Logistics analysis that found 11,881 errors in a review of nearly 250,000 LTL invoices. That is just under 5%. Inaccurate accessorial charges (extra fees for things like liftgates or residential delivery) made up 42.8% of errors. Fuel surcharges made up 14.4%. Freight misclassification made up 12.3%.
That mix matters. Accessorials and fuel are rule-based checks against a rate sheet. A trained agent can do them all day without getting bored. Misclassification is harder, but still pattern-based.
Evos is also aiming at small firms on purpose. FreightWaves reports Evos citing that 99.3% of U.S. motor carriers operate fewer than 100 trucks, and that 93% of U.S. manufacturers have fewer than 100 employees. Those firms cannot afford a dedicated audit team or a big software project. Paying for an outcome is easier to approve.

Evos is part of a wider wave

This is not a one-off. 2026 has seen a steady run of funding for AI that does freight desk work, rather than software that helps people do it.

Freight Hero raised a $5 million seed round in July, led by Field Ventures, FreightWaves reported on 30 July. It runs brokers’ back offices for a flat fee per load, from rate confirmation to proof of delivery. It says AI agents handle over 90% of customer load touches, with human “Heroes” taking exceptions. Field Ventures partner Jillian Williams put the thesis bluntly: “Traditional industries like freight brokerage don’t have large software budgets, but instead have enormous labor budgets”.

Cargofy raised $11 million in a Series A in June, IT Brief US reported on 23 June. That was $6 million of new capital and $5 million of secondary share sales. It says it has annual revenue above $10 million and 2,000 paying customers. CEO Stakh Vozniak said: “One person can now do the work of ten, and revenue per employee grows.”

Opereit, based in Barcelona, raised $2.5 million in June for AI freight-bill auditing and carrier claims, The Conveyor reports. It competes head-on with Evos’s first use case.

Note that Field Ventures co-led Evos and led Freight Hero. One investor backing two firms in the same niche within three months tells you how strongly it believes in the category.
The caution comes from the big end of the market. FreightWaves cites an August 2026 Boston Consulting Group report: only 13% of executives surveyed at 30 leading global logistics companies said AI was delivering measurable financial results. Money is flowing faster than proof.

What it means for forwarders and shippers

Our read for forwarders is practical.
The first jobs to move will be the ones with clear rules and lots of volume. Freight invoice audit, carrier compliance checks and “where is my shipment” queries fit that. Customs classification is on Evos’s list too. That is higher-risk: a wrong tariff code can mean penalties, so expect tight human sign-off there for a long time.
Pricing will change. These vendors sell outcomes: a flat fee per load, or “zero upfront investment” deals. That makes them easy to trial. It also means you should negotiate on verified results, not flagged ones. Ask for savings measured as carrier credits actually received.
Your data is the product. These agents learn from your best people. Before signing, settle who owns the trained model and the decision logs if you leave. That knowledge “has never been written down”, in Shah’s words. Once it is, you want to keep it.
Headcount will flatten before it falls. Freight Hero customer Ally Logistics told FreightWaves it grew revenue 82.4% year-on-year without meaningfully increasing operations headcount. That is the likely pattern: firms grow volume without hiring, rather than cutting staff outright.
For shippers, the effect cuts both ways. Brokers that audit carrier bills more tightly may hold margin. Shippers should expect their own invoices to be checked by machines too, and disputes to arrive faster.

Our prediction

Evos says Workhorse launches formally in the coming weeks. Our call, and it is ours: by the end of Q1 2027, at least one AI back-office vendor in freight, Evos or a rival, will publish dollars actually recovered for a named customer, not just discrepancies flagged. Buyers who have read the small print on this round will not sign larger contracts without it. We also expect at least two more funding rounds for AI freight-audit or back-office agents to be announced before the end of 2026, given how quickly this year’s deals have stacked up.

Sources