From sandbox to first live client in two weeks.
A two-week activation plan that turns the first signed agency client into a live, governed assistant — not a tour through every feature.
Guide
The path is simple: pick the client, pick the use case, ground the assistant in a small approved source set, test the top 20 questions, publish the widget, monitor usage. Most agencies get this done inside two weeks if they hold the scope.
What does the first client launch actually look like?
Pick one client. One assistant. One use case (support, lead capture, or internal knowledge). Five to ten approved sources. Top 20 questions tested. Publish the widget. Look at usage at the end of week two.
What gets harder without it
Doing everything for the first client kills the timeline
Adding workspaces, Content Studio, and three assistants in the first client makes a four-week launch out of a two-week plan.
Sources without tests blow up on day one
Every assistant has a top 20 that has to work. Publishing without running them means the first real visitor finds the bugs.
No usage review means no upsell story
If you do not look at week-two analytics with the client, you miss the conversation about more assistants, Content Studio, or knowledge maintenance.
What the solution includes
Week 1, days 1 to 3: scope and create the tenant
The decisions made here shape everything that follows.
- Pick the client. Pick the use case. Pick the assistant template.
- Create the client tenant in AgentticAI. Apply the package that matches their plan.
- Configure branding: domain, logo, colors.
- Invite the client admin.
Week 1, days 4 to 7: ground and test
The work that makes the assistant actually useful.
- Collect the top 20 questions the assistant has to answer.
- Upload three to ten approved documents.
- Crawl one or two approved web sources.
- Add Q&A pairs for anything that has to be exact.
- Test all 20 in the playground. Fix what is off.
Week 2, days 8 to 10: publish and watch
The transition from build to operate.
- Publish the widget on the client website (or the standalone page).
- Confirm the embed works on mobile and desktop.
- Set the usage usage warnings.
- Watch the first real conversations on day one.
Week 2, days 11 to 14: review and expand
The conversation that turns a launch into a service.
- Look at analytics with the client.
- Identify the top three new questions the assistant could not answer.
- Decide on the next assistant or content workflow.
- Book the monthly review.
What to hold firm
The scope decisions that protect the timeline.
One client. One assistant. One use case.
Five to ten sources, not the whole library.
Top 20 questions tested before publish.
No new features added during launch week.
What to defer to month two
Capabilities to introduce only after the first client is live.
Additional assistants for the same client.
Content Studio and Media Library workflows.
client-owned AI keys (platform default first, then switch).
Custom actions or webhook integrations.
When to delay launch
Signs the two-week target is not safe to hit.
Top 20 questions still failing in playground after three test passes.
Client admin not yet onboarded by end of week one.
Branding requirements that need custom domain setup beyond the package allowance.
Sources that require legal review the client has not started.
How teams get to value
Day 1: scope conversation with the client. Lock the use case and the top 20 questions.
Day 7: top 20 all passing in the playground.
Day 10: widget live on a staging URL.
Day 14: widget live in production, usage review booked.
What you can measure
The widget is on the public site by day 14 with passing top 20.
The launch process documented for the next client.
Week-two analytics review opens the door to a second assistant, content services, or knowledge maintenance.
Questions teams ask
What if the client wants three assistants at launch?
Push back. Launch one well, then add the others. Three-at-launch usually means four-week timeline and a bumpy first conversation.
Should the client see the agency dashboard?
No. Client admins see their own client portal. The agency tier is yours.
What if the client provides their own provider key?
You can switch to client-owned AI keys after the first week of live traffic. Setting client-owned AI keys up before launch is one more variable that can slip the date.
What if the top 20 questions list is not ready?
Build it with the client during the day-1 scope call. Without it, there is nothing concrete to test against.
Explore adjacent solution paths
Want a launch checklist for your first client?
Request the two-week activation checklist and we will tailor it to the use case you have in mind.