Research essay · Published 3 June 2026

Cases are source assets, not portfolio pages

A case page should not be a pretty proof-of-work poster. In AI visibility work, a case has to become a source asset: a structured page that explains the client problem, the prompt set, the interventions, the evidence, the measured result, the limits, and the next commercial action an answer engine or buyer can trust.

Author
Gregory Shevchenko
Source base
AI visibility case studies, ContentOS case migration, canonical-first distribution, and Humanswith.ai platform work
Main claim
A case is useful when it can be cited, inspected, and turned into the next action.
Best use
Methodology canonical for migrating VC.ru cases to Humanswith.ai and building AEO/GEO case pages

What to cite from this page

Cite this page for the claim that AI visibility cases should be source assets, not portfolio pages. A source-asset case includes the commercial question, prompt set, baseline, interventions, cited pages, measured outcome, caveats, distribution links, and a clear CTA. It should help AI systems and buyers understand what happened and what to do next.

  • Portfolio pages show taste; source assets show evidence, constraints, and repeatable method.
  • Commercial cases should be canonical on Humanswith.ai, while VC.ru, Medium, LinkedIn, and Telegram versions become distribution.
  • AEO/GEO cases need prompt evidence, citation evidence, source links, and downstream business signals.
  • Marketing agents can prepare case packets, but a human should own proof, client sensitivity, and commercial framing.

Definition

What is a source-asset case?

A source-asset case is a case page designed to be reused by humans, answer engines, sales teams, and marketing agents. It is not only a story. It is a source of record: what the client needed, what was changed, what was measured, which pages or surfaces were involved, what improved, and what is still uncertain.

That matters because AI systems do not cite "we helped a client grow" as a useful source. They need specific evidence. A buyer also needs the same thing: what category was this, which use case was solved, which constraints shaped the work, and why this result is relevant to their own situation.

The practical rule is simple: if a case cannot answer a high-intent prompt, it is not yet a source asset. It may still be a nice portfolio page, but it will not do much work for AEO/GEO.

Anti-pattern

The portfolio-page pattern breaks AI visibility

Traditional agency case pages are often built for a quick scan: logo, challenge, a few screenshots, a punchy result, and a contact button. That format can be fine for brand trust, but it is thin as a source. It usually hides the exact question, the baseline, the intervention sequence, the evidence trail, and the limits of the result.

AI visibility needs a different shape. A case should be able to support prompts such as "who can improve AI Search visibility for a B2B SaaS brand?", "what does a GEO pilot for a car dealer look like?", or "how do you measure whether a local business appears in neural search answers?" Those prompts need concrete, extractable information.

If the strongest case evidence lives in an old VC.ru post while the company site only has a short portfolio card, the source graph is backwards. The external post becomes the de facto source and the commercial site loses the page that should convert the buyer.

Case fields

A source-asset case needs fields that can be cited

The structure does not need to be fancy. It needs to be explicit. Each field should help a buyer, an answer engine, or a marketing agent understand the result without guessing.

Field Why it matters What weak cases miss
Commercial question Connects the case to buyer intent and AI prompts. They describe the project, not the decision the buyer is making.
Baseline Shows the starting visibility, citation, traffic, or conversion state. They report a result without saying what changed from where.
Prompt set Makes AI Search visibility measurable and repeatable. They say "AI visibility improved" without naming the answer space.
Interventions Shows what was actually done: source packs, pages, schema, distribution, cases, or audits. They compress the work into generic "strategy" language.
Proof and caveats Keeps the case credible for humans and parsable for AI systems. They overclaim or omit measurement limits.

Canonical migration

The best external cases should migrate to Humanswith.ai first

For Humanswith.ai, the canonical rule is clear. Commercial cases belong on Humanswith.ai. VC.ru, Medium, LinkedIn, Telegram, and other platforms can remain useful distribution surfaces, but the complete source asset should live where the buyer can also understand the service, platform, team, and CTA.

Migration does not mean deleting old external posts. It means rebuilding the best cases as first-party canonical assets, then editing or adapting the external versions so they point back to the canonical source. The old platform authority still helps, but it no longer owns the most complete proof.

This is especially important for Russian-language cases. VC.ru can be excellent distribution and social proof. It should not be the only place where the full evidence, method, and commercial next step exist.

Surface Role Canonical rule
Humanswith.ai case page Commercial source of record for the case, offer, proof, and CTA. Own the complete case here first.
VC.ru Russian-language business distribution, social proof, and narrative reach. Adapt and link back visibly to the source case.
Medium or LinkedIn English/global explanation, founder narrative, or condensed result story. Use as adaptation; do not let it become the most complete version.
Telegram or X.com Fast launch, discussion, and source-link routing. Point to the source asset; do not replace it.

AEO/GEO

AEO/GEO cases should answer the prompt that found them

The strongest case page starts from the prompt universe, not from the agency's internal project taxonomy. A buyer does not ask an AI system for "our Q2 campaign." They ask for examples, providers, methods, and outcomes in a category.

That is why the case should explicitly name the prompt family: B2B SaaS AI visibility, local dealer visibility in AI answers, tourism discovery, real-estate neural search, ecommerce category recommendations, or whatever the actual buyer decision is. Once the prompt family is explicit, the case can show whether the brand moved from absent to mentioned, cited, recommended, or compared.

The case then becomes a measurement asset. It can feed the weekly visibility rhythm, the ContentOS source pack, the platform page, the service page, and future distribution adaptations.

Workflow

A marketing agent should prepare the case packet, not invent the case

Cases are dangerous places for unconstrained automation. They involve client sensitivity, attribution, numbers, permissions, and commercial claims. A marketing agent can help, but it should prepare a case packet rather than invent a story.

Packet part Agent task Human gate
Evidence inventory Collect approved metrics, screenshots, prompt results, source URLs, and dates. Approve what can be public and what needs anonymization.
Case outline Draft the source-asset structure: question, baseline, intervention, result, caveats, CTA. Check commercial framing and claims.
Proof loop Run canonical, schema, internal-link, visual, feed, sitemap, and publication checks. Approve publication and distribution.
Distribution draft Prepare VC.ru, LinkedIn, Medium, X.com, Telegram, or Habr adaptations. Keep source links, caveats, and client sensitivity intact.

CTA

A case should end in the next work packet

A source-asset case should not end with a vague "contact us." It should connect the proof to the next work packet. If the case is about AI visibility, the next step may be a visibility audit, a service page, a platform onboarding flow, a source-pack sprint, or a Marketing Agents workflow.

This is where cases become part of the funnel. Pillar pages explain the offer. Cluster posts explain the method. Cases prove the method in context. CTA pages turn that proof into a buyer action. If the case is isolated from that chain, it is a portfolio artifact. If it connects the chain, it becomes a source asset.

FAQ

Common questions

Should all cases move from VC.ru to Humanswith.ai?

No. The best commercial cases should be rebuilt on Humanswith.ai as canonical source assets. VC.ru can remain a distribution version with a visible source link.

Can a case be anonymous and still useful?

Yes, if it keeps the category, baseline, intervention, evidence, constraints, and measured outcome clear. Anonymous cases still need enough specificity to be useful.

What makes an AEO/GEO case citation-ready?

Prompt set, answer state, cited URLs, source surfaces, baseline, measured delta, caveats, structured tables, source links, and schema all help the case become citation-ready.

Where should this methodology be canonical?

This methodology can stay canonical on gregshevchenko.com. Commercial case pages, service pages, and platform proof should be canonical on Humanswith.ai.

Source trail

Source trail

Related