The short answer
Intelligent content is content that adapts to the person reading it, and to the AI agents that now research on that person’s behalf. It draws on data about the buyer’s industry, company, and stage of buying, then changes to fit. Intercept, a B2B marketing agency for global technology companies, plans and produces intelligent content for brands including Microsoft and SAP.
What intelligent content is
Static B2B content is finished the day it is published. One version goes out to every reader. It stays the same until someone rewrites it. Intelligent content combines human creativity, AI, and data. The asset can change with the reader and improve as buyers engage with it.
Phyllis Davidson, vice president and principal analyst at Forrester, laid out this shift at SAP Concur’s Influence by Design conference in October 2026.1 In her framing, intelligent content serves two audiences at once, human buyers and AI agents.
What changes when content becomes intelligent
We see four differences between today’s content and intelligent content.
Today’s content and intelligent content
| Today’s content | Intelligent content |
|---|---|
| One versionEvery reader gets the same asset. | Many versionsEach industry, account, or role gets its own version, built from the same source. |
| Finished at launchThe asset stays the same until someone rewrites it. | Improves after launchThe asset can be updated as buyers engage and their needs change. |
| Read by peopleThe asset is written for the buyer alone. | Read by people and AIThe asset is also clear and structured enough for an AI agent to cite accurately. |
| Checked by handA person reviews each asset one at a time. | Checked by rulesPeople set the sources, rules, and tests. Every version passes them before it is shown. |
Why buyers ignore vendor content
Our agentic audience intelligence platform, Watchtower, analyzed 16,077 public conversations in which decision makers in finance and in travel and expense discussed vendor content, from October 2025 to October 2026. The five most common reasons they gave for dismissing it were close to equally common. Each appeared in 17% to 22% of conversations.2
Why vendor content misses the mark with finance and travel and expense decision makers
In a separate set of 32,167 conversations about the kind of personalization these buyers value, 43% pointed to content fully tailored to their organization. Far fewer pointed to content that reflects the constraints of their sector (18%) or matches where they are in the buying process (16%).2 The full study is in our B2B buyer research. The argument is in our article “Your buyers don’t see 89% of your content.”
The Content Personalization Spectrum
Personalization at scale stalls when teams treat it as an all-or-nothing choice for their whole content program. We treat it as a spectrum of five levels. Each level changes more of the asset and needs more information about the buyer to do it.
- Generic. One version serves everyone.
- Surface. A name, logo, or industry label is swapped in.
- Segment. The asset is rewritten for one industry, role, or market.
- Account. One account’s priorities and evidence shape the asset.
- Living. A new version is assembled the moment someone asks, whether that is a person or an AI agent.
Content that understands a buyer’s industry or role needs the segment level. Content fully tailored to the buyer’s organization was the form of personalization these decision makers valued most (43%).2 In our spectrum, that starts at the account level. Not every asset needs to reach the top. The spectrum helps a team decide how far each asset should go.
Real-world examples
Segment by industry: SAP Concur
SAP Concur asked us to create versions of eight general-audience assets for finance leaders in state and local government. We built a custom AI setup grounded in SAP Concur’s brand and messaging and tuned it alongside SAP Concur’s own reviewers. We estimated the regular workflow at 24 days per asset. Once the setup was tuned, each asset took seven days in total. The biggest gain came at the start. Time to a first draft fell by 80%. Revised drafts were ready 60% faster after each round of feedback.3
Segment by role: Microsoft
For Microsoft, we turned its Frontier Firm research into eight interactive web guides. Each guide is written for a different executive, including the CFO, the chief information security officer, and the chief revenue officer.
Account: HP
With HP’s ABM team, we are setting up a process in which AI agents tailor campaign content to individual target accounts. It draws on HP’s content library, data on each account’s interests, and what HP’s sellers know about each account.
Living: AMD
For AMD, we built a living e-book that writes a new version of itself for each reader’s company. In our demo, it produced a company-specific version in about 90 seconds, drawing on public information such as earnings calls and annual reports.4
The hard part is quality control
AI can now draft a version for every industry, role, or account quickly and cheaply. Making sure every version is accurate and on brand is far harder. The same request to an AI model can produce a different answer each time. Quality at scale depends on fixed checks that give the same pass or fail result every time they run.
On the AMD e-book, no person reviews a version before it reaches the buyer. People approve the sources, rules, and tests up front instead. We built a test suite of 108 cases and tested the system on 30 companies. Every version must pass six checks before it is shown.4 On the SAP Concur and Microsoft work, a person approves each asset before it goes out. The HP process is still being set up and is designed the same way.
How Intercept gets you there
We start by placing your content on the spectrum and deciding which level each asset has earned. Then we write down what the AI needs to know, including your brand, your products, and the rules every version must follow. We empower copywriters as content engineers. They design that knowledge and those checks. They still approve the work. The result is a managed service that produces intelligent content and keeps improving it.
These programs are steps toward what Forrester calls an agentic content system.1 We have not yet seen anyone run the full system end to end. We have already built parts of it with clients looking to operate at the frontier. Content is one part of a wider agentic marketing system.
You can see how that work is going on our agentic content system page.
Who we work with
We work with mid-market and enterprise technology companies, in the US and around the world, that want a managed agency partner for content across many audiences and markets. Our programs are built for teams with procurement, brand, and legal review. We are not the right fit for very small companies or one-off projects.
Sources
- 1. Forrester, presentation by Phyllis Davidson, vice president and principal analyst, at SAP Concur Influence by Design, Bellevue, Washington, October 20, 2026.
- 2. Intercept Watchtower, shares of 16,077 and 32,167 conversations among finance and travel and expense decision makers, global, October 2025 to October 2026. Watchtower studies public posts on X, Reddit, Quora, Bluesky, and Amazon reviews. It draws its sample with a random sampling method built on a statistical physics technique published in peer-reviewed research. It then classifies the audience and weights the sample against census data. The data reflects what people say in public. It does not measure beliefs they keep to themselves.
- 3. Intercept, SAP Concur state and local government program, 2026. The 24-day figure is Intercept’s estimate of the regular workflow, from brief to final packaged asset. The 80% and 60% figures measure production time, not in-market results, which are not yet reported.
- 4. Intercept, AMD living e-book program, 2026. The test suite has 108 cases and was run on 30 companies. The 90-second figure comes from a demo. Each version must pass six automated checks before it is shown.