Citi wants its law firms to explain their AI savings. AT&T is moving some legal work in-house.

For a professional services CFO or COO buying AI to protect margins, that is an uncomfortable combination. The client can ask for a share of the efficiency gain—or decide to do part of the work itself.

Your firm and your client may both be counting on the same savings.

Before signing the next AI contract, there is a question your leadership team needs to answer: are you investing in work your clients will still pay you to do?

Two clients. Two ways to change the deal.

The Financial Times reported that Citi had begun asking law firms to bid for work and explain their AI savings. Its global head of legal, Adam Meshel, expected lower costs per transaction when AI reduced the hours required.

That expectation belongs in a supplier's business case. A workflow can become more efficient while the buyer negotiates away part of the financial benefit.

AT&T goes further on the question of who performs the work. Reuters reported that its internal AI-assisted legal team was handling assignments traditionally given to junior lawyers at outside firms. Specific cost savings remained difficult to estimate.

On October 8, AT&T announced a collaboration with OpenAI to develop legal workflows grounded in attorney oversight. Its existing LegalEdge group, launched in January, had handled nearly 100 matters during its first six months. Those matters preceded the new collaboration.

Citi is challenging the price. AT&T is developing an alternative to buying some work externally.

These are two enterprise examples, not a forecast for every professional services firm. They do reveal a blind spot in an AI investment case built entirely around internal productivity: the buyer has a technology strategy, too.

A faster firm can still face a tougher sale

Imagine an account review with a long-standing client. Your team arrives with a productivity story: research is faster, first drafts take less time, turnaround has improved.

The client arrives with a different agenda. Which savings will reach its invoice? Which routine tasks can its own team now handle? What expertise should it still buy from you?

That is a hypothetical conversation. Citi and AT&T show why the questions deserve real preparation.

A firm can pay for new software, release employee time and still struggle to turn that capacity into profit. The people may remain on payroll. New work may not arrive. A client may seek a lower fee.

Buying more AI does not settle those questions. A stronger investment decision starts with the commercial value the technology needs to support.

Here are three moves to make before the next purchase or renewal.

1. Find the work your client still needs you for

Take one service that matters to revenue. Break a recent engagement into its components: gathering information, producing a draft, checking it, making recommendations and taking responsibility for the result.

Assess which parts a capable client team could plausibly perform itself. Then identify the expertise, context and accountability it would still need from you.

A recurring management report, for example, may combine routine assembly with recommendations informed by years of understanding the business. Making assembly faster is useful. Strengthening the recommendations may create a different reason to keep buying the service.

Ask whether the proposed investment improves the contribution the client values, reduces delivery cost, or accelerates work the client could take inside. The answer changes what deserves funding.

2. Follow the saving all the way to the business result

Citi's reported demand makes an hours-saved slide an incomplete answer.

Follow one representative engagement from intake to client acceptance. Include review, corrections, software, integration and ongoing administration. Compare the full cost of reliable delivery under the proposed workflow.

Then give released time a destination: additional profitable work with credible demand, an expense the firm can actually avoid, or a measurable improvement in service.

Keep those benefits separate. The same hours cannot fund new revenue and disappear as payroll savings simultaneously.

Test what happens if the client receives part of the gain through a lower fee. Leadership should be able to see what remains for the firm—and which assumptions still need evidence.

3. Make the vendor compete with what you already own

AT&T's example makes the choice of work important. It does not establish that your firm needs another platform.

Compare the proposed purchase with improving an existing tool and retaining the current approach. Examine three-year total cost, implementation effort, operational risk, overlapping capabilities, renewal commitments and the cost of leaving.

That comparison may support the purchase. It may reveal a workflow problem that another subscription will not solve.

A vendor can demonstrate its product. Your leadership team still needs to establish why the investment makes commercial sense for your firm.

Turn the next AI proposal into a decision you can defend

This is where TEEVA vCIO, provided by Arcway Industries LLC, fits: independent technology investment and vendor governance.

Its Technology Decision Baseline is a scoped four-week engagement covering an application and spend inventory, a technology decision memo with three-year total cost of ownership modeling, a 12-month investment roadmap and an executive readout. It gives leadership a structured comparison of technology options before committing capital.

TEEVA accepts no vendor commissions or resale margins. Its role is to help scrutinize the technology commitment, with advice independent of a license sale.

If an AI purchase or software renewal is approaching, bring that decision to the table before committing more spend.

Request a Technology Diagnostic.

Your client may already be reviewing the same economics from the other side of the invoice.