Chapter 08

Scale across sites and teams.

After Stand works on one page, the next question is coverage. Which sites should have chat? Which agents should answer? Which reps should be eligible? How do you keep context specific without creating a support operation?

Stand Guidebook

Chapter 08 of 09

Field guide

What to learn in this chapter

Scale Stand by separating domains, audiences, and responsibilities. One account can cover many sites and agents, but each site should still have a clear reason to exist.

Use this chapter when you manage a marketing site, docs site, app pages, client sites, campaigns, or a team where different people should answer different visitor contexts.

Register each meaningful site or audience separately.

Use site prompts for context that changes by domain, product, region, or client.

Scope agents when a generalist stand-in would give weaker answers.

Use live matching to see who would handle a visitor right now.

Site strategy

01

Create sites for meaningful context, not just URLs.

A site in Stand is a routing and context boundary. It deserves its own setup when visitors need different facts, different reps, different agents, or different measurement.

Do not split every path just because you can. Split when the visitor's job changes enough that the stand-in should behave differently.

Example

Marketing site plus docs site

Situation
The marketing site creates pricing and fit questions. The docs site creates implementation and migration questions.
Move
Use separate site prompts and either separate agents or site-specialized agents.
Insight
The same company voice can sit on top of very different visitor jobs.

Example

Agency client portfolio

Situation
An agency wants to offer Stand coverage across several client sites without per-seat overhead.
Move
Register each client site, set client-specific site prompts, and assign the right specialist or account owner.
Insight
Scale is not only more agents. It is keeping context and ownership clear.

Agent specialization

02

Use specialist agents when the visitor expects expertise.

A generalist agent can cover simple questions. A specialist agent is better when the page has its own domain language, pricing rules, product area, or audience.

Specialization should make the visitor experience clearer. If two agents would answer the same way, keep one.

Use one general agent whenThe site is small and questions are similar.
Use specialist agents whenPricing, docs, product, or client pages need different context.
Use one general agent whenOne person owns most follow-up.
Use specialist agents whenDifferent reps or specialists should own different signals.
Use one general agent whenThe prompt is still short and easy to test.
Use specialist agents whenThe prompt becomes a pile of exceptions.

Routing visibility

03

Check who would answer before traffic arrives.

The Sites page shows Matches right now for each site. This is operationally useful: it tells you whether a visitor would reach a live rep, an enabled stand-in, or nobody useful.

If multiple AI agents match the same site, Stand shows the probability split. In the common two-agent case, the next visitor may be routed 50/50 between the matching stand-ins. When a human rep is available, has I chat here checked, and is below their concurrent chat limit, the human match takes priority over AI coverage.

Stand Sites screen showing elythen.com as installed, I chat here checked, Matches right now with an AI agent at 100%, and site activity stats.
Figure 8-1. Matches right now is the routing preview. It shows installation state, human site preference, eligible agents or reps, current probability, and recent site activity.

Before adding a new site

  • The site has a clear visitor job.
  • The site prompt explains context that the page alone may not carry.
  • Matches right now shows at least one eligible rep or agent.
  • If multiple agents match, the probability split is intentional.
  • If a human should answer, I chat here and availability are set correctly.
  • The greeting fits the page and audience.
  • You know what signal you hope to learn from this site.
SituationYou are available and I chat here is checked.
Who answersYou can override matching AI agents for live visitor chats.
What to checkConfirm availability, site preference, and concurrent limit.
SituationYou are unavailable or at your limit.
Who answersEligible AI agents cover the visitor if agents are active.
What to checkConfirm the right agents appear in Matches right now.
SituationTwo agents match the site.
Who answersStand can randomly select between them according to the shown probabilities.
What to checkUse specialization if one agent should own the site.
SituationNo rep or agent matches.
Who answersThe site is installed but uncovered.
What to checkCreate an agent for the site or adjust rep site preferences.

Screenshot reading

What Matches right now tells you

Green light + Installed
The snippet has been detected for the site and the site is enabled.
I chat here
You personally answer this site when you are available and below your concurrent chat limit.
AI badge
The match is an enabled stand-in rather than a human rep.
Probability
One eligible agent may show 100%. Two eligible AI agents can share traffic evenly at 50/50.
No matches
Visitors will not have useful coverage until a rep matches the site or an agent is created for it.

Questions

Common reader notes

Can one account run Stand on multiple sites?

Yes. Stand supports unlimited sites across plans, each with its own embed snippet, prompt, and usage stats.

Can an agent specialize in one site?

Yes. Agents can be general or scoped to a specific site.

What does Matches right now mean?

It shows the reps and agents that would currently be eligible if a visitor arrived on that site, including the probability split when multiple AI agents match.

Try the guide on one real page.