Affordable AI takes more than a lower price

Affordable AI takes more than a lower price

Elizabeth Field DiGiovanni runs Parcel Air, which makes travel activity boxes designed to keep young children occupied on flights. Everything that goes into the Los Angeles solopreneur’s boxes has to come from somewhere: the toys, the crayons, the packaging. Most of the factories she works with are in Asia, where the workday begins around the time Elizabeth is putting her 4-year-old and her 2-year-old to bed.

She was sourcing—finding factories that could make what she needed and getting the job quoted. Those questions run in sequence; each answer sets up the next. A minimum order quantity changes which price break to ask about. A certification answer decides whether lead time matters at all. With the time difference, every link in that chain cost her a day. One vendor negotiation took more than three weeks to produce a quote.

AI agents for e-commerce could have done that work for her. Elizabeth assumed reaching that point meant a week of setup she could not spare, choosing among tools and connecting her catalog and supplier list. Now one of our customers, Elizabeth told us when discussing a case study that while running a company she was also minding two small children at home, and there is rarely a free minute.

AI AFFORDABILITY

Affordability, for a business like hers, includes the total cost of adoption.

A Goldman Sachs survey conducted in August found that 86% of surveyed alumni of its 10,000 Small Businesses program were using AI, but only 15% had fully integrated it into their core operations. Those owners completed a business education program, so they generally sit ahead of small firms. Separately, Census Bureau data from December 2025 to May 2026 showed AI use below 20% among U.S. employer businesses with four or fewer employees, compared with 37% among those with at least 250.

Meanwhile, Stanford’s 2025 AI Index tracked the cost of querying a model at roughly GPT-3.5-level performance from $20 per million tokens in November 2022 to $0.07 by October 2024. We have watched the same curve inside our own product, Accio, where token cost fell by half in four months while consumption grew tenfold. Lowering the cost of intelligence still leaves the work required to put it to use.

THE COSTS BEYOND THE SUBSCRIPTION

Someone still has to configure the tools, understand the business questions, and check the results. Larger companies can spread that work across teams. Parcel Air has no employees, so Elizabeth performs all of it. Those demands belong in any affordability calculation. The companies building these systems, ours included, can lower them at three levels.

1. Lower the technical barrier

An owner looking for suppliers should be able to start with that job, without first learning model selection, prompt engineering, or software integration. Without dedicated IT teams, small businesses have little capacity to assemble and maintain AI systems.

Pre-configured expert agents and sourcing plugins do that work in advance, reducing the configuration, custom development, and integration an owner would otherwise take on.

2. Close the expertise gap

For AI agents for e-commerce to deliver value, they must combine technical simplicity with an understanding of how businesses source, trade, and operate.

A tariff changes landed cost. A missing certification can prevent goods from entering a market. Supplier screening and international logistics require knowledge that owners without procurement teams may not have.

Developers can embed that knowledge inside the owner’s workflow. Owners should still be able to see what needs checking, with the system flagging missing information and claims requiring documentation.

3. Make autonomy easier to trust

An agent that requires constant supervision can create as much work as it removes. Unlimited authority, however, introduces risks an owner may not be able to afford.

Owners need clear permissions, a visible record of actions, and the ability to inspect recommendations and intervene. Routine supplier research and organizing quotes can run independently once the workflow is proven. Orders, payments, and contractual commitments should require explicit human approval.

Those controls belong in the product itself. The goal is less routine supervision while the owner keeps control over consequential decisions.

WHAT CHANGED FOR PARCEL AIR

Elizabeth sources through Alibaba.com and runs this work on Accio, our AI agent designed for small businesses and solo founders. She briefs it the way she would brief a colleague. Before bed, she writes out what she needs: the product, her target cost, the safety certifications, and the quantity. The agents work with suppliers through the business day in Asia while she sleeps. In the morning, before the children are up, she reads what came back and makes the call. She described it to us as having an entire APAC team on the night shift.

The first time she ran it properly, she says, the AI went through 68 suppliers in 3 minutes and 17 seconds. A quote that used to take three weeks came back in 12 hours.

She still sets the target cost. She still decides which supplier to trust and which certification claim needs a document behind it. What changed was the waiting. She stopped losing a day to every question, and she got her evenings back.

MEASURE THE FULL COST OF ADOPTION

Those of us who build AI tools for small businesses should be able to answer two questions. How much time does it take an owner to get a useful result, including setup and learning? And how much time does the tool save on recurring tasks once you factor in setup, review, and upkeep?

For a small business owner, the real test comes when the tool becomes part of a busy workweek. Small businesses everywhere should be able to afford to put AI to work. As Elizabeth puts it, “You don’t need a bigger team. You need the night shift covered.”

Kuo Zhang is president of Global B2B Business at Alibaba Ecommerce Group.