The platform
Five things have to happen. This is all five.
Find the business, prove the problem, win the conversation, build the fix, run the client. Every other tool in this category does one of them well and hands you the rest.
Find
Lead sourcing
Vertical, metro, Google Places.
Signal detectors
106 in the catalog. 68 run live.
Opportunity radar
Bulk scan and deep diagnostic.
Watchtower
Re-scan what you saved.
ICP
Who to hunt. Flag-gated until Autopilot is on.
AI visibility
Citation evidence, not a guess.
Market intelligence
Category × metro. Preview flag.
What is the Nerxus One platform?
Nerxus One is one product that finds a local business with a detectable problem, proves it, wins the conversation with a working sample, builds the fix, and runs the client after they pay. The internal loop is Detect, Engage, Fulfill, Manage. The site names the same motion Find, Prove, Win, Build, Run so a marketer can see where two of the steps happen after the meeting.
The buyer is the marketer. Nidrosoft is the company. Nova is the orchestrator. Ten named agents do the work. Only Relay sends outreach. Voice is inbound only. SMS needs first-party consent from Forms. Review collection has no gating branch.
The dentist, the HVAC shop, and the cafe are your customers. They may later see a portal. They never sit in the agent screen and they never get a seat that can hunt your other books. A Clerk organization is the tenant. Auth is Clerk. Data lives in Supabase. Mail goes through Resend on a domain you verified, or it does not go.
Most tools in this category stop when a spreadsheet looks full, a PDF looks branded, or a CRM row looks busy. This platform exists because a full spreadsheet with no finding is a week you cannot invoice, and a report that tells the owner to hire someone else is treated as a failed product, not a deliverable. The seam that matters is already on the finding. A mapped leak carries the fulfillment action that opens Forge in the same sitting.
How the five motions map to the modules
Find is Scout, the detector registry, Radar, Watchtower, ICP, the answer-engine probe, and market intelligence. Prove is the deep diagnostic, Atlas reports, sample-first, and branding. Win is Pulse, Relay, sequences, senders, inbox, meetings, and forms. Build is Forge, Muse, Echo, inbound voice, local SEO, conversion widgets, and compliance drafts. Run is Nova's mission control, the Autonomy Matrix, Steward, and the client portal.
A finding that maps to a fix already carries the fulfillment action. That seam is why a no-website row can open a DEMO site in the same sitting, and why a report that tells them to hire someone else is treated as a failed product, not a deliverable.
Detect is Find plus the cheap tags that ride along at search. Engage is Prove plus Win, the memo and the mail. Fulfill is Build, the DEMO and the gated DEPLOY. Manage is Run, Steward and the portal. The two naming schemes are the same week. The site names help a marketer see that two beats happen after they get hired. The internal names help the app keep one business uuid from Scout through Steward.
You can skip a module. You cannot skip the gate on a send, the lock on cold voice, the lock on cold SMS, or the public buttons on a collect page. Skipping Muse for a month is a choice. Asking Nova to text a scraped list is a refuse. The motions are a map, not a checklist you must finish before lunch.
How Find works from a vertical and a metro
Find starts when you pick a category a directory already uses and a city you can name on a map. Scout queries Google Places as the required source and the only live discovery, then adds Facebook page signals to what it finds. Yelp discovery needs a paid Fusion plan and Yellow Pages has no supported API. If a key is missing, that source reports unknown. The search does not pretend the data arrived.
You do not type a boolean about job titles. Dentists are not the same hunt as roofers. Category lives in the taxonomy. Metro is a place with a radius. ICP writes that hunt once so you stop retyping the paragraph tomorrow. Exclusions belong in the profile too. A shop you already serve should not reappear as new money.
Each business is saved once per source, keyed to the listing’s own ID. Handoffs persist first and pass a business uuid, not a provider id that dies when you change keys. LeadScore ranks the pile from cheap signals that rode along at search time: no website, no SSL, a low rating, a thin review count, an unclaimed listing. LeadScore is a sort. It is not a close-rate prophecy and it does not print a dollar.
Background scans run on a durable queue. If the queue is down, a scan can be accepted and sit. The UI is supposed to tell you that, not invent a finished list. Watchtower later re-scans what you kept and diffs what changed. Timing is a finding. A storm, a permit, a rating drop, a license drift. Freshness decays. A stale saved lead is a polite way to waste a send.
How Prove turns a listing into evidence and a modeled dollar
Prove is the deep look that turns a cheap tag into a sentence you can defend. Radar runs the detector suite on a saved list or on one business. Bulk scan and deep diagnostic are the two modes. Findings carry evidence, a confidence, and a modeled dollar with the assumptions still attached.
Dollars on Radar are modeled. They are not the SMB's register and they are not your invoice. If two screens ever disagreed about a number, that was treated as a defect in the app, not a rounding story. A finding with no evidence and no confidence degrades to a could-not-verify note. Bare dollars do not print on an Atlas report.
The forensic memo is the diagnostic page. Headline, modeled dollars, evidence, roadmap, and the DEMO that should already exist. Atlas composes branded proof from findings that already passed a credibility gate. Studio, library, schedules, and a public share token at /r/ are the surfaces. Branding profiles put your mark on the cover. The marketer owns the artifact. The prospect is not our lead.
Sample-first is the Prove subject that Win will mail. I noticed this listing has no booking button, so I already built a DEMO. Here is the live link. If you skip Radar and jump to outreach, you can still write. You will write like every other vendor. The wedge that works here is specific, and specificity is a diagnostic job.
How Win uses Pulse, Relay, and a working sample
Win is the conversation after the memo exists. Pulse parks the row on a signal-named board: No-Booking Dentists, No-SSL Plumbers, Permit-Lead Roofers. Freshness, SLA, and next best action live here. Two dollars stay separate. DEV is what the SMB recovers per month, inherited from Radar, labeled modeled. FEE is what you charge once. The forecast sums FEE over open stages.
Relay is the only agent that sends outreach, and only after ComplianceGate returns a branded pass on the tenant's verified sending domain. Templates are signal-led. Sequences stop on reply, honor quiet hours, and respect warmup caps. The inbox classifies interested, question, not now, and unsubscribe. Suggested replies stay gated in Supervised.
Calendar in outreach is your calendar. You are booking a sales meeting. The pipeline stage named Demo is that meeting. It is not the Forge working sample. Mixing those two words is how a marketer tells a prospect they already published a site when they only booked Tuesday. DEMO is a product word for the preview. Demo on Pulse is a human on a call.
Forms are the only path a compliant SMS list can be born. The consent checkbox cannot be pre-checked. A bundled marketing opt-in on a required service line is a publish block. Voice cannot go cold from this motion. If you ask Nova to dial the Austin list, Nova refuses. Email-first on a verified domain is the send that exists.
How Build fulfills from a finding without a second import
Build opens from the fulfillment action already stamped on the finding. No website opens the website tool. No booking opens the booking path. A collect-page gap opens Echo's collect page. A service gap opens the local SEO page Forge can draft from the same taxonomy that found the hole. There is no second import and no blank prompt that pretends the leak was never named.
DEMO is a Forge working sample. DEPLOY is a gated action. One Free Fix means the DEMO asset, not an unsupervised publish onto a domain you do not own yet. Compliance drafts stay labeled professional review recommended. We do not pretend this is a law firm. Muse plans and drafts the month. Publishing is approval-gated by default.
Voice in Forge is inbound only. A demo number can sit on a pitch. A live line can answer after they pay. There is no outbound cold-call path. Conversion widgets, quotes, and the instant responder are speed-to-lead you can show. They attach to the same gap Radar scored.
Echo monitors reviews, drafts replies, and owns the collect page. Public review buttons render for every visitor. There is no gating branch and no toggle that adds one. Private recovery sits alongside, never instead. Yelp and Tripadvisor stay monitor or assisted until a fuller API exists. Low-star replies stay held. Google and Facebook follow the connections you actually have.
How Run starts when Steward inherits the same row
Run starts when you mark Closed and Steward takes the same business record as a client. Health, managed services, timeline, renewals, leftover-finding upsells, Stripe invoices on your own account, and the portal invite live here. Closed is the hand-off, not a goodbye and a blank account.
Health on this site is Healthy, Watch, or At-risk. Drivers can include results from deployed assets, payment status, engagement, review velocity, and recency. The band is an internal relationship score. It is never the SMB's revenue printed as a dollar. Upsell cards come from findings that still have no deployed asset. Est. value stays labeled est.
The portal is magic-link, one business, plain language, no internal notes, no costs, no sibling shops. The SMB is read-only. They do not receive Scout. They do not receive Nova. They do not receive your other pipeline. If a token ever opened two shops, that is a defect.
Steward drafts. Relay still sends. Forge still builds the next service. Atlas still writes the recap. Echo still keeps reputation. If you ask Nova how this client is, Nova routes to Steward. If you ask Nova to email the recap, Nova still ends at Relay and the gate. Client mail after the sale is still a send. It does not grow a second rail.
What Nova does as the orchestrator
Nova classifies the sentence, picks the specialist, enforces guardrails before dispatch, and returns one reply after the tools finish. Route, don't do is the rule. Nova does not scrape a listing, does not score a finding, does not hit Resend, and does not write a DEMO.
Classification is two-tier. Keyword overlap against each tool's keyword list runs first, at no token cost. A high score routes immediately. An ambiguous sentence falls to an LLM fallback. A weak score asks one clarifying question instead of guessing a send.
Multi-part asks become a task graph. Find dentists in Austin with no booking, diagnose the three I would take, and draft the mail is Scout, then Radar, then Relay. Nova stays on the edges. The activity row still names which specialist ran, because an orchestrator that hides the worker is how you lose the audit.
You type into one screen. That screen is Nova. The difference is what happens after the sentence. A tool call hits a real module, writes a real row, and shows up in activity with a gate decision. A chatbot that cannot persist a business uuid is a demo of tone.
What Scout, Radar, and Pulse own after a save
Scout searches directories, attaches cheap detectors at search time, scores the pile with LeadScore, and saves a business. Scout never emails, never texts, and never calls. The Autonomy Matrix keeps Scout off the send rows. There is no hidden toggle that turns a search into a blast. Google Places is required per tenant. Other sources report unknown until you add a key.
Radar diagnoses. 106 catalog detectors tag website, SSL, ADA, GBP, hiring, permits, service-gap, and AEO families. Scout rides the cheap subset at search. Radar runs deeper families at Quick, Standard, or Deep. Adapters without a key stay unknown. Present, absent, unknown, or conflicting, plus evidence and a checked-at time. A failed fetch is not a clean no.
The detection engine fetches once, stores an artifact, then runs pure detectors over that artifact. Neither Scout nor Radar should hammer a provider once per checkbox. Unknown is allowed because a green check with no artifact is how marketers send a No-SSL email to a site that has had HTTPS for a year.
Pulse orders the funnel after the save. A saved business becomes a deal on a signal-named board. Pulse does not get its own dollar. Radar models opportunity. The pipeline stores fee. Steward later shows retainers you actually entered. Inventing a fourth number so a homepage tile looks busy is how marketers stop trusting the roster. Neither Radar nor Pulse sends.
What Relay, Atlas, and Forge produce
Relay drafts and sends email through ComplianceGate on the tenant's verified domain. Unverified is a blocking gate, not a warning. Stop-on-reply is the inbox job. Atlas can share a report through a public token or ask Relay to mail it. Steward can ask Relay to send an invoice or a portal magic link. If a screen looks like it sent, the activity row should name Relay.
Atlas composes branded, dollar-quantified reports from findings that already passed a credibility gate. Twenty-six proof types, compose once. Atlas does not invent a customer ROI for this site. Share analytics stay in your library. Opens are your signal. We will not mine /r/ traffic into platform outbound.
Forge builds the fix. Website, inbound voice, instant responder, local SEO pages, conversion widgets, compliance drafts, and the rest of the fulfillment registry share one pattern. DEMO first, DEPLOY gated. DEMO is a working preview, not a sales meeting. Forge does not mail. Muse does not mail. Echo's review requests that leave the building still pass the same gate and the same consent rules.
v1 white-label is the artifact and the portal wearing your identity. Reselling the whole platform as a chrome-stamped console is not shipping. A public template marketplace is not shipping. Client self-serve report ordering is not shipping. Print that before a franchise lead asks for twenty fake products with your logo in the sidebar.
What Muse, Echo, and Steward keep after the meeting
Muse calendars the month across channels and drafts the posts. Publish waits for a yes. Nothing posts because a model felt confident. Instagram and paid ads as outreach channels are on the not-doing list. A draft you paste into a tool you already pay for is your business. Using Nerxus as a stealth ads desk is not a supported path.
Echo collects. The collect page cannot hide public review buttons. Private recovery may sit beside a low-star path. It cannot replace the public path. Cold review SMS is locked. Forms consent only. Do not tell a client you are posting live on Yelp or Tripadvisor from this product until those credentials are exercised.
Steward keeps the book after they pay. The same uuid. The same leftover findings. The same modeled assumptions if you still show a diagnostic number. Health stays a band. Invoices go through your own Stripe account via a gate when that connection exists. We have not described a platform take rate. Do not read a percentage into the silence.
Older product notes sometimes folded post-sale work into Atlas. The site and the app split that job. Atlas owns branded reports and proof artifacts. Steward owns the book of business, renewals, invoices, and the portal invite. If you ask Nova for a Win the Pitch pack, Atlas works. If you ask who is due this month, Steward works.
Why Supervised is the default on the Autonomy Matrix
Supervised is the default because a send, a publish, a deploy, or a spend is how you get sued, suspended, or fired. The Autonomy Matrix is a per-agent, per-action grid with three levels: Autonomous, Supervised, or Off. Read-only work such as search, scan, draft, score, report, and sweep may default to Autonomous. Anything that sends, posts, publishes, or spends defaults to Supervised until you opt that cell up.
Four cells cannot move. Cold voice stays Off. Cold SMS stays Off. Review gating stays Off. Email without a verified domain stays blocked at the gate. Those rows show a lock icon. A tenant admin cannot enable them. A playbook cannot smuggle them in, because the gate reads the tool's tier at dispatch, not the playbook's wish.
Off as a global Agent Mode stages work and runs nothing until you switch. Supervised drafts and queues. Autonomous permits enabled playbooks inside caps, windows, and scope. The matrix is the fine grain. The mode is the master switch. Most restrictive wins. Governance is live. The overnight playbook runner is the open half. We will not claim overnight playbooks have run in production.
Memory follows the same ten names. Episodic rows record what an agent did to a business. Semantic facts stay on that business as it moves from saved lead to deal to client. Procedural patterns that you reject in Approvals get quieter the next time. None of that gives Scout a mouth.
How a Clerk organization becomes a tenant
A Clerk organization is the tenant. First sign-in provisions identity. Isolation is enforced in Postgres with tenant_id and row-level security. Super-admin is a separate plane. A marketer cannot wander into another book by guessing a URL. Background jobs that bypass RLS still go through a tenant-guard.
Secrets are per-tenant and encrypted. The browser sees masked status. The platform owner does not burn your Places quota. You do not burn a client's. An LLM key is per-tenant too. A shared platform brain that quietly spends your neighbor's budget is out of bounds.
Public tokens (/r, /b, /f) are how strangers see a report, a booking page, or a form without a seat. Those routes have to be public in the proxy. Forgetting that once turned every share link into a sign-in wall. Viewers and respondents are not waitlist rows.
v1 does not resell white-label sub-accounts. Two legal brands should assume two tenants. Sharing a tenant to share a Places key is how two books touch. The client portal is not a fake product with your chrome. If a prospect asks for that resale console, the honest answer is later, not a toggle we hid.
Who the platform is for, and who should use something else
The platform is for a solo marketer or a small team that sells websites, reputation, local SEO, inbound voice, or a retainer built from those jobs to local SMBs. If your week is dentists in one metro and roofers in another, you want one set of names that survive the close.
If you sell to VPs of Sales on LinkedIn, use a people graph. Apollo is better at that corpus. If you only need a raw local list with a couple of flags and you will fulfill in five other tools, a lighter scraper will feel faster. LeadsGorilla wins that job. If you already have the client and only need a CRM, GoHighLevel is the desk you grow into.
If reviews are the only service you sell, Climbo or Birdeye go deeper on that pillar. If you need inbox rotation at volume, Instantly is the sending shop. If the deliverable is a rank PDF, BrightLocal already does it. If you already have the client and only need a site, Lovable is faster in isolation. If inbound voice is the whole offer, Synthflow is the specialist. If you need a broad SEO suite, Semrush is wider.
We close the loop they leave open. We do not pretend to beat a specialist at the one job they spent a decade on. Agencies running many books care about isolation more than a new persona. Agents run inside a tenant. There is no path from one client's saved leads into another client's.
How marketers use the five motions in a week
Monday you pick dentists in Austin with no online booking. You ask Nova. Nova routes to Scout. You save the ones with a basic tag you can fulfill. You do not email them yet. ICP is the sentence you will run again next week without retyping.
Tuesday Radar deep-diagnoses the three you would actually take. Atlas can brand the memo. Forge builds a DEMO site or a booking block from the fulfillment action on the finding. Pulse parks the rest on a No-Booking Dentists board so they do not rot in a notes app.
Wednesday Relay drafts from the finding. You approve. The sequence stops if they reply. The inbox classifies. Calendar is the book link in the mail, and it is your calendar, not the dentist's patient book. Thursday you can ask Nova for a Win the Pitch pack. Friday you ask who is waiting and the approval queue names Relay drafts, not a mystery assistant.
None of that requires you to believe an overnight playbook already ran in production. Supervised is how a careful marketer spends the first month. Echo stays quiet until reputation is on the contract. Steward stays quiet until Closed. Get started on this site is the waitlist until access opens in batches.
What unknown means on this platform
Unknown means the tenant did not attach the key or the fetch did not complete. It does not mean the business is clean. It does not mean the business is dirty. It means you cannot sell that sentence yet.
Marketers hate unknown because it looks unfinished. The alternative is a lie. We picked unknown. Settings, Connections is where unknown goes away, one provider at a time. Feature gates that need a provider key fail closed and say unknown. They do not draw a green check from missing data.
Google Places is the only required key because search cannot start without it. Yelp, job boards, permit feeds, and license sources unlock their own detectors. You do not owe us a full keychain on day one. Intelligence rollups stay hidden until preview is on. A rollup that cannot name provenance does not ship as a fact.
When a number appears on two screens, it must resolve in one function. DEV on Pulse is Radar's modeled dollar. FEE is what you typed. Health is a band. LeadScore is a rank. Mixing those four into one hero metric is how a marketing site gets demoed into a hole.
What the platform will not do
It will not send cold voice. It will not send cold SMS. It will not hide public review buttons. It will not print a plan price until commercial policy exists. It will not claim overnight playbooks have run in production while the runner is still the open half. It will not invent customers so this page looks busier.
It will not claim SOC 2. It will not claim HIPAA. It will not call Yelp or Tripadvisor a live reply network. It will not invent a close rate from Scout. Scout does not close. Relay does not close. The marketer closes. LeadScore is a sort.
It will not resell white-label sub-accounts in v1. It will not add an eleventh agent that cold-calls. It will not merge Scout into Relay so a search blasts. It will not treat the pipeline Demo stage as a Forge DEMO. It will not print a number of leads per hour. Volume depends on the metro, the vertical, the keys you attached, and whether the queue is running.
Roadmap Not doing also includes Instagram or paid ads as outreach, invented plan prices, and EU-first residency until we decide. Those refusals are product locks, not a mood. An about page that softened them to sound friendlier would be a different company. We would rather lose a buyer who wanted a dialer.
Questions
Is Nerxus One an AI SDR?
No. It is an operating system for marketers who sell growth work to local SMBs. Scout never contacts anyone. Relay is the only agent that sends outreach, and only after ComplianceGate on a verified domain. Voice is inbound only. Cold SMS is locked.
What do I read first?
Lead sourcing, opportunity radar, outreach, websites, and the agent roster. Those five are the column leaders. Then read guardrails so the four locks are visible before you ask for a seat.
What is the difference between DEMO and the pipeline stage named Demo?
DEMO is a Forge working sample, a live preview you can put in a sentence. DEPLOY is gated. The Pulse stage named Demo is a sales meeting on your calendar. Mixing those words is how a preview gets described as a publish.
Who is the user, and who is the customer?
The marketer is the user. The local SMB is their customer. The SMB may get a read-only portal for one business. They do not get Scout, Nova, or your other pipeline.
Do overnight playbooks already run in production?
No. Governance is live. The overnight runner is the open half. Supervised is the default. We will not date a fantasy or claim autonomous dispatch has already spent a night.
How do I get started?
Get started is the waitlist at /join. Access opens in batches. There is no public price table. Login will point at the app when that URL is public. The marketing site is not the product.
Find the leak. Ship the work. Keep the account.
Tell us your vertical, your metro, and what you can fulfill. Supervised is the default. Plans start at $99/mo.
14-day trial. Card required, converts on day 15. Full table on /pricing.