A workspace per client, and no path from one to the other.

Why do agencies mix two clients up, and what breaks when they do?

An agency mixes books when a lead, a report, a sender, or a portal token is one export away from the wrong Slack channel. The owner of shop A sees a screenshot from shop B. A sequence fires from the wrong domain. A contractor pastes a diagnostic into the other brand's folder because both rows were named Joe's.

Most stacks make that easy. The list tool is a shared CSV. The auditor is a login everyone knows. The mailer has one warmup pool. The site builder has one team workspace. The reputation tool has one agency seat and a hope. None of those objects share a business uuid that Steward can inherit.

Nerxus One treats that mix as the failure you are buying insurance against. The product is the same loop a solo runs. The frame is isolation. A Clerk organization is the tenant. Saved businesses, secrets, senders, Pulse boards, and portals live there. There is no path from one client's row into another client's.

Nova routes nine specialists inside that wall. Scout never contacts from the shared pile. Relay is the only agent that sends outreach, after ComplianceGate, on a domain as tight as the contract. Steward takes the same uuid after they pay. Health stays a band. The portal stays one business. v1 does not resell a fake SaaS with your chrome. Overnight dispatch is not a production claim you can hide behind a bigger headcount.

v1 still will not sell you a white-label sub-account catalog. If your pitch to a franchise group is twenty fake products with your logo in the sidebar, wait. If your pitch is we will not leak your neighbor's reviews into your portal, that sentence is already true. The ten-clients guide is the longer method. The size page is the procurement sentence. This page stays the fear: two books touching.

What does Nerxus One do for a multi-client local agency?

It lets a team find local businesses with detectable problems, prove the finding, win the conversation with a working DEMO, fulfill the fix, and run the account without restarting the record in a second CRM. Detect, Engage, Fulfill, Manage. Find, Prove, Win, Build, Run on the site.

The buyer is still the marketer. The SMB is the customer. They may receive a magic-link portal. They do not receive Scout. They do not receive Nova. They do not receive your other pipeline.

Agencies that already sell websites, local SEO, reputation, or inbound voice can keep those offers. The change is the seam. A finding carries a fulfillment action id. Forge opens the matching builder. Echo opens the collect page that cannot gate. Muse drafts a month you still approve. Relay writes the wedge that matches the detector.

Agencies that only sell a PDF will feel over-tooled. BrightLocal already reports rank and citations. If the deliverable is the PDF, stay there. If the PDF is supposed to become a retainer, the report is the door and the page Forge builds is the product.

How does tenant isolation work before white-label sub-accounts exist?

Auth is Clerk. First sign-in provisions the organization. Integration secrets encrypt per tenant. The browser sees a masked status, never the Places key, never the Resend secret. The platform owner does not burn your quota. You do not burn a client's.

Dedupe and handoff stay inside that tenant. Scout keys a business once per source on the listing’s own ID. Handoffs persist first and pass a business uuid. Radar, Pulse, Forge, and Steward see that uuid. They do not see a provider id that dies when someone rotates a key.

Portals are one business, plain language, no internal notes, no costs, no other clients. If a portal ever showed a second shop, that is a defect, not a packaging choice.

Agency workspaces as a resale object are later. Isolation as a storage and access fact is now. The agency size page repeats this so procurement does not hear a different story than sales.

Which agents does an agency actually seat people on?

An AE lives in Pulse and the inbox. A strategist lives in Radar and Atlas. A builder lives in Forge, Muse, and Echo. A closer still approves Relay until you trust a specific action on the Autonomy Matrix. An account lead lives in Steward after the win.

Nova is shared. It routes. It does not become a fourth personality in the Slack huddle. Slash-delegate if you want a specialist. The screen still comes back as tool calls, not a group chat with mascots.

Scout stays read-only even when a hungry junior wants to "just send a few." Contact is Relay plus the gate. That split is how you keep a new seat from burning the agency domain on a Friday night.

Voice, if you sell it, is inbound only. There is no cold-call campaign hiding under an agency toggle. Synthflow is deeper if voice is the whole firm. Here, voice sits in the loop next to the site and the collect page.

What does Monday look like inside an agency that refuses to mix books?

Monday each pod runs its ICP, not a shared scrap pile. Dental in one metro for the book that can fulfill booking and reviews. Home services in another for the book that can fulfill service-gap and permits. Scout tags cheap detectors at search time. LeadScore ranks sellability, not prophecy.

Saves belong to the tenant that hunted. If you run multiple legal entities, do not share a Clerk org to save a seat. The isolation story dies the moment two brands share a sender and a lead table.

Watchtower is how Monday stays cheap. Re-scan what you already saved instead of paying for a fresh dump that forgets last month's rating drop. Timing is a finding you can staff. A storm, a permit, a license drift. Freshness has a clock. A stale row is a wasted AE hour.

Market intelligence, when preview is on, rolls category by metro with provenance. Until preview is on, do not sell a pricing map you cannot open. Hidden until preview is a real state on that module.

What does Wednesday look like when every pod attaches a DEMO?

Wednesday the strategist deep-diagnoses the rows the AE would take. Radar stages evidence, confidence, modeled dollars, and the fulfillment action. Atlas brands the memo with the agency mark, or with a client-facing cover if that is the motion. Credibility gate refuses bare dollars. Assumptions stay attached.

The builder opens Forge from the action id in the same sitting. DEMO first. The AE gets a link they can put in the mail. DEPLOY waits for a paid client and an approval. One Free Fix means that DEMO, not an unsupervised publish.

Relay drafts the wedge. ComplianceGate brands it. Supervised means a human still says yes. Sequences stop on reply. Inbox classes are interested, question, not now, unsubscribe. SLA clocks sit on the AE, not on a fictional SDR.

If two pods need two domains, verify two domains. Unverified is a blocking gate. Sharing one warmup personality across three brands is how agencies inherit a reputation problem they then blame on "AI email."

What does Friday look like when Steward takes the win?

Friday the record becomes a client without a CSV export. Health is Healthy, Watch, or At-risk. Upsells come from leftover findings, not from a brainstorm. Renewals are dates someone owns.

The portal is the client-facing surface. Magic-link. One business. Results they can read. If you need a full agency CRM with twenty funnel recipes, GoHighLevel still wins after the account exists. Steward is the continuation of the finding, not a clone of a decade of snapshots.

Echo and Muse keep running inside the same tenant. Collection cannot gate. Publish stays gated. Yelp and Tripadvisor, where connected, are monitor or assisted. Do not tell a client you are live-replying there until those credentials are exercised.

Invoices, when the commercial path is on, are you charging them. We will not print your fee, and we will not print ours.

Which compliance lock matters most to a multi-client shop?

Isolation plus the sending domain. A gating complaint is close behind. An agency that hides public review buttons on a collect page is one screenshot from a platform ban and a client lawsuit story. Echo has no such branch.

Cold voice and cold SMS stay locked even when a client asks you to "just have the robot call the list." You can sell inbound voice as a missed-call catch. You cannot sell a dialer. Forms are the only birthplace of a compliant SMS list.

Compliance fixers stay drafts. Professional review recommended. If you staff a lawyer, they review. If you do not, you say so. Selling ADA as if the model is counsel is how agencies get in the paper.

The Autonomy Matrix is per agent and per action. A principal can leave Scout autonomous on search and keep Relay on Supervised. Overnight playbook dispatch is honest about the open half. Governance is live. Autonomous sending while you sleep is not a claim this page will make.

How do agencies keep DEV and FEE from lying in the forecast?

DEV is modeled recovery for the SMB, per month, with assumptions. FEE is what you charge. Pulse forecasts FEE. A board that sums DEV and calls it revenue is a lie you will repeat on a Monday standup.

LeadScore is not FEE. Opportunity score is not FEE. Atlas will not let a bare dollar through the credibility gate. If two screens disagree about a number, that is a defect in the app, not a rounding story.

Retainers should map to leftover findings. A shop that bought a site still has a thin GBP, a collect-page gap, or a service they never listed. That is month two. It is not a new hunt in a new tool.

If a client wants a rank PDF every Friday, BrightLocal may still be in the stack. We will not pretend twenty-six report types are a Semrush replacement. The report here exists to sell the fix.

What should an agency sell in month one, and what should wait?

Sell the detector pack you can fulfill with the people you already pay. Web studios sell the site DEMO. SEO shops sell the service page and the AEO gap. Reputation shops sell collection that will survive a complaint. Voice-capable shops sell the inbound line. Mixed agencies sell the finding that matches the pod.

Wait on white-label resale. Wait on invented packaging that implies the SMB is the user. Wait on Facebook, Trustpilot, or GBP actions you have not connected. Unknown is a first-class state.

Wait on flipping every action to autonomous because a competitor's homepage said zero approval clicks. Our default is the opposite on purpose. Supervised is how you keep a junior off a client's domain.

Wait on medical promises this product cannot make. The medical industry page says we are not an EHR and we do not claim HIPAA. If a healthcare book needs that posture, read that page before the kickoff.

How does an agency run ten local accounts without a shared scrap pile?

Write ICPs per offer, not per mood. Staff Watchtower on the saved set so Monday is diffs, not a new scrape. Keep senders as tight as the contract. Keep portals to one business. Keep Pulse boards signal-named so a new AE can read them.

The ten-clients guide is the longer method. This page is the fear: mixing two books. The size page is the procurement sentence: isolation, not sub-account resale.

Client eleven is an economics question and a seat question. It is also an isolation question. If the eleventh account requires a second legal entity, it requires a second tenant. Sharing an org to dodge a seat is how the fear comes true.

Nova can be the shared front door. It cannot be the shared brain that casually retrieves the other client's diagnostic because someone typed a similar name. Route, do not do. Specialists stay inside the tenant's rows.

When should an agency keep another tool and only add Nerxus One for the loop?

Keep a specialist when the specialist is the product you sold. Reputation-only firms can stay on Climbo or Birdeye for depth and use us when they want find, prove, and fulfill in the same sitting. Voice-only firms can stay on Synthflow. Site-only shops can stay on a builder.

Keep GoHighLevel if the book is already there and the funnels are load-bearing. Adding us in front of that CRM is a detect-and-sample layer, not a midnight migration speech.

Keep Semrush or BrightLocal if the buyer bought research or rank tracking. We sell the local finding and the page that fixes it. We do not sell a full SEO suite.

Drop the shared CSV when the csv is how leads cross books. That is the replacement this page actually cares about.

How do you keep Pulse boards from becoming a second shared scrap pile?

Name every board after a signal and a book, not after a mood. No-Booking Dentists Austin for the dental pod is a campaign. Q3 Pipeline is how two AEs work the same owner. Pulse will let you make the junk drawer. Isolation does not rename your boards for you.

One row, one owner, one next action. When a strategist deep-diagnoses, the uuid stays on that board until it closes or dies. Exporting the board to a spreadsheet so a contractor can "clean it up" is how a second copy of the book is born. The contractor then mails from a personal account and the gate never sees it.

DEV and FEE stay split on every board, including the ones that feel like retainers. A pod that sums modeled chair-hour recovery and calls it booked revenue will lie in the Friday standup. Forecast FEE. Atlas still refuses bare dollars on the memo that leaves the building.

When a row wins, it leaves the hunt board for Steward. Size does not change that handoff. A won column that still holds the diagnostic is how next week's hunter wastes a DEMO on a client you already have.

What does a client portal owe an agency book that a shared Google Drive does not?

A portal owes one business, plain language, no costs, and no path to the neighbor. A Drive folder owes whoever has the link. Agencies leak reviews, NAP drafts, and the other client's rate card because someone forwarded a folder named Clients.

Magic-link is the door. The SMB reads results. They do not read detector names, LeadScore, or your Pulse comments. If you need them to approve a Muse draft, that approval is still your job unless a later product puts a thin approve button on their side. v1 is read-only on purpose.

Two locations for one legal client can still be two portal tokens if they are two businesses in the tenant. Collapsing them to save a click is how a downtown shop sees the suburb's one-star week.

If the client demands a login that looks like software they own, you are in the resale conversation. v1 will not give you that chrome. Tell them the truth in the kickoff so month two is not a surprise ticket.

How should an agency onboard a new pod without cloning the wrong brand kit?

Give the pod an ICP, a Pulse board, a sender they are allowed to use, and an Atlas profile that already says the right legal name. A shared default cover that still carries last year's brand is an isolation bug in human form. Nova can be the shared door. The cover cannot be the shared brain.

Run one hunt through DEMO before they get Relay. A pod that sends first and brands later will mail a finding with the other client's voice. Supervised is how a principal catches that before the domain pays.

Steward templates, if you keep any, should not include internal notes from the last book. Portals stay one business. A new AE who copies last week's portal paragraph and forgets the shop name is why Friday exists as a check, not a vibe.

How do design-partner agencies evaluate without a public price table?

Use the demo path if you are evaluating at volume. Twenty minutes with the loop, not a slide about agents. Join the waitlist if you want a batch seat. Access opens in batches.

Bring your own Places key conversation. Per-tenant keys are the architecture. We will not burn a shared platform quota to make a trial look generous.

Ask about the four locks, the gate, and the portal. If a seller promises white-label sub-accounts in v1, they are off script. If they promise HIPAA, SOC 2, or overnight playbooks already running in production, they are off script.

The product does not change between this page and Agency by size. The fear is mixing books. The size frame is how many seats share Nova. Read both if procurement wants both sentences.

What is the first mix-up that usually happens in week two?

The first mix-up that usually happens in week two is a screenshot or a draft that names the wrong shop because two rows were named Joe's and someone exported a CSV into Slack. The owner of shop A sees shop B's rating drop. A contractor pastes a diagnostic into the other brand's folder. A sequence fires from the domain that was warming for a different book.

Most stacks make that easy. The list tool is a shared CSV. The auditor is a login everyone knows. The mailer has one warmup pool. The site builder has one team workspace. The reputation tool has one agency seat and a hope. None of those objects share a business uuid that Steward can inherit.

Nerxus One treats that mix as the failure you are buying insurance against. Saved businesses, secrets, senders, Pulse boards, and portals live in a Clerk organization. There is no path from one client's row into another client's. A portal that ever showed a second shop is a defect, not a packaging choice.

v1 still will not sell you a white-label sub-account catalog. If your pitch to a franchise group is twenty fake products with your logo in the sidebar, wait. If your pitch is we will not leak your neighbor's reviews into your portal, that sentence is already true. Say it in the kickoff so month two is not a surprise ticket.

How do you keep a contractor from becoming a second book?

You keep a contractor from becoming a second book by giving them a seat and a role inside the tenant, not a forwarded password and a CSV. Give them Forge and the uuid. Do not give them Relay. Do not give them DEPLOY until they have shipped DEMOs you would send. The fulfillment action id is their brief.

A contractor who copies listings into their own scraper is off the team. A contractor who mails from a personal account is a send the gate never saw. A contractor who wants to resell the chrome as their own AI studio is asking for the v1 Not doing. They can sell the work. They cannot sell a sub-account factory we do not ship.

Secrets stay in the tenant. Their Places key is not your Places key. Pay them on FEE you collected, or on a craft rate you set. Do not pay them on DEV. Modeled recovery is not a contractor bonus pool.

When they leave, take them off Relay and DEPLOY first. Pulse ownership moves to a named human. If they take a CSV, treat it as a leak. Isolation does not follow the laptop. The product's job was to make the daily export unnecessary. Humans still download things.

What does a sender incident look like after it already mailed?

A sender incident after it already mailed looks like a complaint that lands on a domain two books were sharing, and then every sequence on that domain pauses. Sharing one warmup personality across three brands is how agencies inherit a reputation problem they then blame on AI email.

If two pods need two domains, verify two domains. Unverified is a blocking gate. Relay remains the only mail sender. ComplianceGate brands every send. Supervised remains the default even if you have an ops seat. Volume is Instantly's win, not a reason to lift our gate.

Activity should name Relay and the pass or the block. A cheerful we already reached out without a row is a defect. Approve re-checks the gate. A domain that expired between draft and click still fails. Kill switch is on Mission Control when a feed looks wrong. Resume one playbook, not all of them.

Retirement is the other half of the incident. Close the board. Pause Relay. Take the domain off that book if the contract ends. A retired shop that still receives a sequence from last quarter is how you inherit a complaint after the invoice stops. Steward can keep the closed record so you do not re-hunt them. Exclusions in the ICP should name them.

How should two AEs split a metro without splitting the uuid?

Two AEs should split a metro without splitting the uuid by writing two ICPs and two signal-named boards when they sell two offers, then letting the listing ID save the shop once. Ownership decides who speaks. A shared we will both try is the collision.

One row, one owner, one next action. When a strategist deep-diagnoses, the uuid stays on that board until it closes or dies. Exporting the board so a contractor can clean it up is how a second copy of the book is born. Pulse will let you make a junk drawer named Q3 Pipeline. Isolation does not rename your boards for you.

Stop-on-reply is team-wide. An AE who keeps a personal sequence running in another mailer after the inbox classified not now is the collision. Forecast FEE, not DEV. Two AEs stacking modeled recovery will lie to the principal. Atlas credibility gate exists because that lie is common.

When a row wins, it leaves the hunt board for Steward. A won column that still holds the diagnostic is how next week's hunter wastes a DEMO on a client you already have. Size does not change that handoff. Client eleven is an economics question and an isolation question. If the eleventh account requires a second legal entity, it requires a second tenant.

What portal sentence do you tell a franchise lead in the kickoff?

The portal sentence you tell a franchise lead in the kickoff is one business, plain language, no costs, no path to the neighbor, and no chrome-resale console in v1. Magic-link is the door. They read results. They do not read detector names, LeadScore, or your Pulse comments.

Two locations for one legal client can still be two portal tokens if they are two businesses in the tenant. Collapsing them to save a click is how a downtown shop sees the suburb's one-star week. A Drive folder owes whoever has the link. Agencies leak reviews, NAP drafts, and the other client's rate card because someone forwarded a folder named Clients.

If they demand a login that looks like software they own, you are in the resale conversation. v1 will not give you that chrome. Tell them the truth so month two is not a surprise ticket. If they want to approve a Muse draft, that approval is still your job unless a later product puts a thin button on their side. v1 is read-only on purpose.

The SMB is not the SaaS user. Making them one is how you inherit their clicks as your incidents. Branding the glass with your logo is in scope. Selling them a sub-tenant that can run Scout is out of v1. Isolation still holds. They cannot walk sideways into another shop you manage.

How does Watchtower stay cheap without becoming a shared scrap pile?

Watchtower stays cheap without becoming a shared scrap pile by re-scanning the businesses you already saved, landing diffs on the board that owns the uuid, and naming a human. A rating drop, a permit, a hours change. Five Slack screenshots is how two AEs both mail, or how nobody does.

Monday each pod runs its ICP, not a shared scrap pile. Dental in one metro for the book that can fulfill booking and reviews. Home services in another for the book that can fulfill service-gap and permits. Saves belong to the tenant that hunted. If you run multiple legal entities, do not share a Clerk org to save a seat.

Freshness clocks match the vertical. Roofers after hail are not a law firm's practice-area page. A hunter who wants a daily metro scrape because the channel felt quiet is paying for theater. Unknown sources stay unlabeled as live. A Watchtower tile that implies Yelp depth you do not have will be screenshotted into a QBR.

Market intelligence, when preview is on, rolls category by metro with provenance. Until preview is on, do not sell a pricing map you cannot open. Hidden until preview is a real state. Timing is a finding you can staff. A stale row is a wasted AE hour. Nova can be the shared front door. It cannot casually retrieve the other client's diagnostic because someone typed a similar name.