Is it an AI add-on or an AI-native system? Five ways to tell
Most systems a forwarder will be shown are older systems with AI features added. A few are built AI-first. This page does not say who is who. It shows you how to tell.
Five tells, what each one looks like in a demo, and how to check all five on one real enquiry. Ask every vendor, including us.
Most systems a forwarder is shown are older systems with AI features added. A few are built AI-first. Five tells separate them: where the work happens, who fetches, whether it acts or only answers, who approves, and what happens between desks. Ask every vendor to show each tell on one real enquiry, including us.
By the end of this lesson you can
- Name the five tells.
- Run them in any demo on one real enquiry.
- Say honestly when an add-on is enough.
- Explain the difference to a colleague in a minute.
Why does the difference matter to a forwarder?
In a slide deck the two look the same. On a Tuesday morning they are not the same at all.
An add-on puts an AI feature next to your old screens. The old way of working stays. Your team still opens the portals. They still copy the rate across. They still forward the mail to the next desk.
An AI-native system is built the other way round. The software does the reading, the fetching and the typing. Your people read the result and say yes or no.
Here is the plain version. A chat box answers a question. An agent does the work and brings you the result to approve.
Why should you care? Because your hours do not go on judgement. They go on hunting a rate, chasing an agent and typing the same job twice. An add-on helps you type. It does not go and fetch.
At Infinity Logistics, average response time fell from more than 60 minutes to about 15 minutes. At Infinity Logistics, the win rate rose from 35% to 73%. That is our own customer, one company, one year, one signed reference. Read it as proof that the way of working matters. It is not a promise for your firm.
So the question is not which vendor is clever. It is where the work happens once the mail lands.
What are the five tells?
You do not need a list of vendors, and this page will not give you one. Five tells are enough.
Each one is something you can see. Not something you have to be told.
| The tell | An add-on looks like | AI-native looks like |
|---|---|---|
| Where the work happens | A chat box beside the same old screens. Your team still lives in the inbox. | One task screen instead of the inbox. Watches every company mailbox, pricing, operations, tracking and accounts, and follows each conversation from start to finish. |
| Who does the fetching | You still open the portals. The AI writes up what you found. | Pulls your rate cards, checks the carrier, looks up the customer’s history, reads the shipment status, and talks to your TMS or accounting system. Nobody hops between websites to put an answer together. |
| Does it act, or only answer | It answers the question you typed. Then it waits for you. | Works out what each email is and what it needs, then takes the action that message calls for. Chasing, tracking checks, quote expiry and follow-ups run on their own. Nobody has to remember. |
| Who approves | Nobody, or everybody. The reply appears and no name sits behind it. | Your team approves every reply before it sends. Your team approves. They do not compose, forward, file or chase. The reply goes out in their name. |
| What happens between desks | Forward the mail. Copy the details. Type them in again. | Work passes from pricing to operations to accounts with the email and everything it found attached. Nobody forwards it, re-keys it or asks for an update. |
- Where the work happens
- An add-on looks like
- A chat box beside the same old screens. Your team still lives in the inbox.
- AI-native looks like
- One task screen instead of the inbox. Watches every company mailbox, pricing, operations, tracking and accounts, and follows each conversation from start to finish.
- Who does the fetching
- An add-on looks like
- You still open the portals. The AI writes up what you found.
- AI-native looks like
- Pulls your rate cards, checks the carrier, looks up the customer’s history, reads the shipment status, and talks to your TMS or accounting system. Nobody hops between websites to put an answer together.
- Does it act, or only answer
- An add-on looks like
- It answers the question you typed. Then it waits for you.
- AI-native looks like
- Works out what each email is and what it needs, then takes the action that message calls for. Chasing, tracking checks, quote expiry and follow-ups run on their own. Nobody has to remember.
- Who approves
- An add-on looks like
- Nobody, or everybody. The reply appears and no name sits behind it.
- AI-native looks like
- Your team approves every reply before it sends. Your team approves. They do not compose, forward, file or chase. The reply goes out in their name.
- What happens between desks
- An add-on looks like
- Forward the mail. Copy the details. Type them in again.
- AI-native looks like
- Work passes from pricing to operations to accounts with the email and everything it found attached. Nobody forwards it, re-keys it or asks for an update.
We build one of these systems, so here is our own claim in our own words. “Your people work in FreighAI, not in the inbox.” Hold that sentence to the same five tells you hold everyone else to.
Ask every vendor to show each tell on one real enquiry, start to finish. Including us. A vendor who is happy to do it will say yes in one sentence.
A tell is not a score. Three out of five is not a pass mark. Weigh tell two the most. That is the one that gives you hours back.
How do I run the five tells in a demo?
Use one enquiry. Your own. Start where the mail lands and watch the whole thing.
- 01Pick one real enquiry from last week. Pick a messy one.
- 02Send it the way the customer sent it. Do not tidy it up first.
- 03Ask them to begin at the mailbox, not on a slide.
- 04Tell one. Watch where their person works. Count the screens they open.
- 05Tell two. Watch who fetches the rate. Count the websites.
- 06Tell three. Watch whether it moves on its own, or waits for you to type.
- 07Tell four. Ask who approves. Ask whose name goes on the reply.
- 08Tell five. Watch the work reach operations. A forward or a re-type is your answer.
- 09Ask for a second run on a lane that an agent has to price.
- 10Write the five answers down the same day. One line each.
- 01Pick one enquiryReal, from last week, and messy.
- 02Send it as it camePhoto, forwarded mail, third attachment and all.
- 03Start at the mailboxWhere the mail lands, not on a slide.
- 04Watch the fetchingCount the websites a person opens.
- 05Watch it actDoes it move on its own, or wait?
- 06Watch the approvalOne named person, and whose name sends it.
- 07Watch the hand-overA forward or a re-type is your answer.
- GATEYou decideFive answers, written down the same day.
Give every vendor the same enquiry. Including us. Same mail, same five tells, same questions.
Two answers matter more than the rest. Who opened a website, and whose name went on the reply. Write those two down word for word.
What does an add-on do well, honestly?
More than this page has said so far. An add-on can be a fine first step.
Here is what it is genuinely good at.
- It switches on quickly. Your screens do not change.
- Nobody has to learn a new way of working.
- It writes a decent first draft of a reply.
- It finds an old mail faster than you can.
- It reads a long document and gives you the short version.
- Your data stays where it already sits.
So when is an add-on the right call? When one of these is true.
- You renewed your system last year and cannot move now.
- Quoting is not where your day goes.
- You are three people and already quick.
- You want to see what AI feels like before you spend real effort.
- Your rates sit in one place, and finding a rate takes minutes.
And when is it not enough? Look at where your hours go. If they go on fetching and chasing, an add-on will not give them back. It helps you write. It does not fetch, chase or hand over.
If you already have an add-on, run the five tells on it first. You may find it does more than you thought. You may also find your day has not changed at all. Either answer is worth knowing before you shop.
What does a chat box on an old screen cost me over a year?
Time. Here is the sum in round numbers, with the working shown.
Say a desk of four handles 20 enquiries a day. By hand, one quote takes one to two hours. That is 20 to 40 hours of quoting every day. This is an example, not a market number. Put your own count in and the shape holds.
Now the other way of working. Our own claim, in the words we can stand behind:
Where the rate can be fetched without waiting on a person, a priced quote goes out in 10–15 minutes instead of 1–2 hours. That holds for ocean and air just as it does for express; the mode does not change it.
Not every lane works like that, and we will not pretend it does. Where a lane has to be priced by another agent or over the phone, the wait belongs to whoever is answering. No software fixes a person who has not replied yet.
Ten minutes saved on one quote sounds small. Twenty enquiries a day, five days a week, for a year is not small. That is the gap you are choosing between.
Infinity Logistics quoted 70% more shipments with the same five-person team, comparing Q4 2024 with Q4 2025. Again: one company, one signed reference, not a market average.
A chat box on an old screen does not close that gap. The typing gets easier. The fetching and the waiting stay exactly where they were.
There is a quieter cost too. The first credible quote usually wins. Every hour spent hunting a rate is an hour the enquiry sits still.
Where do I go next?
Four short steps, in this order.
- 01Read what an agent actually does. It takes five minutes.
- 02Take the twelve questions into your next vendor call.
- 03Sit in the free monthly talk. Nothing to prepare.
- 04Bring your own enquiry to the session for operations leaders.
One last thing. Run the five tells on the system you already have. The answers tell you what you are really shopping for.
Common questions
Is an add-on always the wrong choice?
No. An add-on can be a fine first step. It switches on quickly and nothing changes for your team. It stops being enough when your hours go on fetching rates and chasing people. Run the five tells on what you have and see which of the two you own.
How do I ask a vendor about this without being rude?
Do not ask. Ask them to show it. Send one real enquiry and ask them to run it from the mailbox to the quote. Then watch for the five tells yourself. A vendor who is happy to run it says yes in one sentence.
The vendor says AI-native and the demo is a chat box. Who is right?
The demo. Words are cheap and screens are not. If a person still opens the portals in front of you, the fetching is still yours. Write down what you saw, not what you were told.
Can an older system become AI-native later?
It can change over time, and some do. You cannot buy the version that does not exist yet. Judge what runs today, on your own enquiry, in front of you.
Do the five tells replace the twelve questions?
No. They work together. The five tells are what you watch in a demo. The twelve questions are what you ask about the rest: data, exit, support and proof. Use both in the same meeting.
- FreighAI Playground: the Playground narrative, section 5, “Add-on or AI-native: how to tell” (internal document, 7 September 2026)verified 2026-09-07
- FreighAI product claims: every capability sentence quoted on this page, taken word for word (mailbox coverage, fetching, acting on its own, approval, hand-overs between desks, quotation turnaround both ways)verified 2026-09-02
- Infinity Logistics: signed customer reference, 15 May 2026 (response time, win rate, and 70% more shipments quoted by the same five-person team)verified 2026-08-31
Run the five tells on your next demo
Pick one real enquiry from last week. Ask every vendor to run it from the mailbox to the quote while you watch. Ask us to do the same. Five written answers are worth more than any slide.