Vijayanta Gupta

Essay

The Board Memo Every B2B Software CEO Needs to Write in 2026

. 7 min read

Most B2B software boards in 2026 are reading two kinds of AI material. The first is the product update: features shipped, models integrated, copilots launched. The second is the macro essay: a category-level claim about how AI is changing software. Neither answers the question that should sit on the next agenda. What does AI do to this specific company?

The CEO is the only person who can write that memo. Not the chief product officer, who will frame it around the roadmap. Not the CFO, who will frame it around margin. Not the head of AI, who will frame it around capability. The board needs the integrated view from the operator carrying the P&L and the strategy together. Five questions, five paragraphs. Signed by the CEO. Distributed to the board ahead of strategy.

1. What workflow do we monetise today

State the workflow in one sentence. Not the product. The workflow. The job the user opens the software to do.

This question forces clarity on what is actually being paid for. Most software contracts are tied to a workflow that has a clear edge, a defined input, and a defined output. The seat price is a proxy for that workflow being executed by a user. When the workflow holds, the contract holds. When the workflow erodes, the contract erodes whether or not the product is still being used.

The honest version of this paragraph names the workflow precisely and names the share of revenue tied to it. Vague answers are themselves the finding.

2. Could an AI agent remove that workflow

State the time horizon and the trigger conditions.

Workflow obsolescence risk is the risk that the product boundary the company protects is now too narrow, the part of the value chain the software owns is precisely the part agents do best, and the customer no longer needs to enter the application to get the outcome. The diligence question is not whether competitors are adding AI features. It is whether AI is making the workflow itself disappear.

A useful version of this paragraph lists three concrete agent capabilities that already exist, three more that will exist within twelve months, and the customer behaviour that would have to change for the workflow to be removed. If the board cannot picture the path, it has not yet been imagined seriously enough.

3. Does AI raise or lower our switching costs

State the answer in revenue terms.

This is the question with the most contested answer inside the boardroom. The optimistic case is that AI raises switching costs because models trained on customer data, agents tuned to customer workflows, and integrations into customer systems compound stickiness. The pessimistic case is that AI lowers switching costs because the new model orchestration layer abstracts the underlying application and lets the customer swap providers more easily than before.

Both can be true inside the same company. The relevant test is whether the switching cost mechanisms the company is investing in compound with usage. A switching cost that comes from proprietary workflow data the customer cannot recreate elsewhere strengthens with use. A switching cost that comes from integration friction the customer is already paying to remove is depreciating.

4. Can we price for AI value without destroying margin

State the gross margin range expected at full AI deployment and name the inference assumption.

Inference is the new cost line. ICONIQ Capital’s January 2026 State of AI report puts inference at 23 percent of total revenue at scaling-stage AI B2B companies, with surveyed AI product builders expecting average gross margins of about 52 percent in 2026 SaaS Mag, 2026. Public SaaS issuers in Q1 2026 began separately disclosing inference-cost ratios in the four to nine percent of revenue band at the COGS line SaaS Mag, 2026. Across the public software index, a corridor of 60 to 70 percent gross margin has established itself for any company shipping meaningful AI capability.

Pricing is the lever that closes the gap. Seat pricing carries some of it. Outcome pricing carries more. Mixed contracts now sit at 92 percent of AI software companies in 2025, which is a signal that the industry has already moved trendingtopics.eu, 2025. Salesforce’s roughly 500-dollar AI-augmented seat against the 175-dollar non-AI seat is the visible benchmark from a category leader Tomasz Tunguz, 2025.

The honest paragraph here states a gross margin target at full AI deployment, an inference-cost-to-revenue ratio the operating plan can hold, and one specific pricing experiment the company is prepared to run this year.

5. What part of our GTM motion becomes obsolete

Name the parts and the timeline.

The buyer is already doing AI-assisted research before sales sees the deal. The demo is no longer where the prospect learns the product. The first sales conversation is no longer where the prospect learns the category. The motion was designed for a buyer who arrived earlier in the journey and bought from a seller; the motion now meets a buyer who has done half of the work with an AI agent and needs validation, proof, and clarity, not education.

The honest version of this paragraph names which sales motions stop compounding (outbound to undifferentiated personas, top-of-funnel content as a brand-builder, demo as conversion), which need to be rebuilt (proof-of-value as the new evaluation step, technical reference architecture as a buying artifact, pricing simplicity as a competitive weapon), and the rough quarter in which the rebuild has to be funded.

What to do with the memo

Send the memo to the chair before strategy. Use it to set the agenda for the next three board meetings, not the next three quarters. The point is not the document. The point is forcing the answers out of the company’s head and onto a page that has the CEO’s name at the top.

Three signs the memo is working. First, the company starts disagreeing with itself in useful ways: the chief product officer and the CFO will not give the same answer to question four, and the gap is the agenda. Second, the next product review changes shape: the team starts asking which workflow each feature defends or extends, not which feature parity gap it closes. Third, the next board pack lands shorter and harder.

Three signs the memo is not working. First, the paragraphs read like a strategy presentation summary, with no specific gross margin numbers, no named workflow, and no quarter committed. Second, the language drifts into roadmap theatre, where every AI feature is treated as a strategic move that signals motion to markets while leaving pricing, margin, and the location of value capture untouched. Third, the board accepts it without asking which paragraph was the hardest to write.

The first growth curve was won by leaders who could ship and sell faster than peers. The second growth curve will be won by leaders who can write the paragraph above harder than peers. The market is already pricing two different business models inside the same sector, with median public SaaS at 3.3 times revenue in March 2026 against AI-native platforms at 16 to 18 times Aventis Advisors, March 2026; SaaSRise, 2026. The memo is the first deliverable that decides which of the two the company is building toward.

Sources

Back to writing