Every module your business runs, on one record

Every module your business runs — CRM, AI softphone, coaching, campaigns and the IT Audit add-on — on one database, one timeline and one permission model.

The platform

The pillars, and what each one is for.

MADDOX is an AI-powered CRM that records every call and meeting word for word, answers questions about your business from that record, and coaches reps while they are still on the call — 200+ capabilities, all on by default. Start anywhere: every pillar below has its own hub and a set of pages under it, all listed further down this page.

Ask Maddox

Ask your CRM in plain English and get the answer with its sources — deals, calls, campaigns, coaching and reps. Nothing changes without your approval.

Sales coaching and 1 on 1s

Sales coaching software for MSP sales teams: every 1 on 1 prepped from real calls, every call graded on your rubric, and every promise tracked to a verdict.

Record Everything

Every call, meeting and touch lands in one CRM timeline, transcribed word for word and searchable by meaning — total recall for your business, at any hour.

AI Softphone

An AI softphone that reads the CRM before it dials and coaches live: when the objection lands, the rebuttal is in the rep’s hand mid-sentence.

Campaigns and marketing automation

MSP marketing automation built as a journey: lanes, steps and goals, a launch readiness check, and every campaign a draft until you approve it.

IT Audit

An add-on for MSPs: leave the client site with the IT assessment scored and a client-ready deck already built off the deal, hours of report-building done.

What they have in common

One workspace, one timeline, one permission model.

The pillars below are not eight products sharing a login. They are one database with one set of rules over it, which is the only reason the answers agree with each other.

Every record knows which customer it belongs to.

Multi-tenant isolation fails in exactly one way: somebody writes a new query and does not add the filter. So the filter is not something a query adds — it is something the model has. Every tenant-owned table carries its tenant, and the models over those tables attach the predicate to every query built through them and stamp it on every row at creation.

A developer adding the hundredth report does not opt in to isolation and cannot forget it. They would have to remove it by name, in a diff, to lose it.

The timeline is assembled at read time, not written to.

There is no timeline table anybody has to remember to write to. Open a contact or a company and the stream is composed on the spot from the six generic tables the work already produced — activities, calls, emails, meetings, notes and attribution touchpoints — newest first. A deal adds three more: its journey enrolments, its documents and its briefings.

That is why a manually logged call, an automated send and a meeting booked from a link all appear in one stream rather than in three tabs that disagree. It is also why adding a capability does not require anybody to go back and teach the timeline about it.

Machine chatter is real history and is not what a rep sees.

Fourteen activity types are what an automation or an integration writes to talk to itself: tag bookkeeping, field pokes, webhook fan-out, sync and import markers, canvas views, the enrol-and-complete churn behind a campaign milestone. They stay queryable and they are hidden by default, behind one toggle. A wall of applied tags is exactly the noise a timeline is supposed to replace.

One permission model, and it reaches the assistant too.

Access is granted per role and, where it matters, per field — a role that should not see a margin does not get the field rather than getting a blurred value. The assistant is held to the same model plus three more gates, and one of those is the identical module predicate the route gate uses, so it can never offer a capability whose screens the workspace has switched off.

What you can reach

Fifteen modules, and four things that decide whether you see one.

Every pillar below maps onto modules in one registry. Whether a given module is lit for a given workspace is answered in a fixed order, and knowing the order is the difference between a support call and a settings change.

The registry, and the one everything else depends on.

Core CRM, sales, marketing, support, telephony, chat, content, account management, data operations, AI and automation, the ecosystem, reporting, IT Audit, the Network Scanner add-on and Microsoft 365. Core CRM is the only one marked as a system module and it is the declared dependency of the other fourteen, so it cannot be taken away by anything — a workspace without it would sign in, load, and reach neither a contact nor a company nor an activity.

The order, and why silence is the interesting answer.

First the subscription ceiling: if the level explicitly denies the module, that is the answer and nothing below is consulted. Then the workspace’s own module row, which an administrator toggles and which is decisive when it exists. Then the workspace’s settings. Then the platform default.

A level that says nothing about a module changes nothing about it, and a level saying yes also changes nothing — a ceiling cannot be a floor. How buying works sets out what is actually denied where, which today is nine modules on one rung and nothing anywhere else.

What ships switched off, and stays off until you say otherwise.

IT Audit and its Network Scanner add-on are sold through the Marketplace: their module rows are created inactive, there is no settings toggle for either, and a Marketplace install is the only way they come on. Microsoft 365 is off behind two gates, one for the deployment and one for the workspace, so the code deploys dark. The PSA and accounting connectors sit behind four. Social posting, ads and SMS are mock-backed and forced off for production workspaces rather than merely defaulted off.

Where a page under this hub describes something that is off by default, it says so on the page rather than in a footnote. A capability you have to switch on is a different purchase from one that is already running.

Where to start

Pick the pillar that matches the argument you keep having.

Nobody adopts every pillar at once and nobody should. These are the four openings that actually happen, and where each one leads.

Nobody can tell you what actually happened.

The forecast is a set of opinions, the notes are three sentences a week old, and the answer to “what did we say to them” is a person’s memory. Start at Record everything for the capture, then Ask Maddox for the reading of it. That pairing is the platform’s actual thesis: the record is a by-product of the work, so the answers come from what happened rather than from what anyone recalls.

The pipeline is optimistic and the coaching is a feeling.

Every deal is committed and half of them slip. Start at Deals and pipelines for stage requirements and grades, and Coaching for the prep, commitment and evidence cycle underneath the one-on-one. The second is the one people underestimate: it is what turns a manager’s impression into something with a date and a source attached.

Marketing and sales are two systems that argue at month end.

Start at Campaigns for the journey engine and the suppression rules, and the pack catalogue for the parts you would otherwise build from scratch. Both write to the same timeline as everything else, which is the whole point — there is nothing to reconcile because there was never a second record.

You sell IT services, and the audit is a deck somebody rebuilds every time.

Start at IT Audit, and read what an MSP should look for in a CRM beside it. IT Audit is an add-on bought from the Marketplace, with a network and perimeter scanner as a second add-on on top, and it is the pillar most specific to how managed-service providers actually sell — a scored assessment against a company, the customer’s own words captured rather than paraphrased, and a client-facing deck built off the deal rather than off a template somebody edited in a hurry.

Across the record

The parts that sit between the pillars.

Some of what MADDOX does is not a pillar of its own, because it reads from three of them at once. These are the pages for those parts.

Inside a deal.

A deal collects more than a stage and an amount. Buying signals are stored as they are spotted, each with where it came from, so “is this deal moving” has evidence behind it. Path to close lays out what is left between here and a signature. Objection intelligence keeps every objection a deal has raised and whether it has been answered, so the open ones cannot hide in a call nobody re-listened to.

Around the rep.

Every recorded call and meeting can be graded against your own rubric, and the 1 on 1 that follows can itself be recorded and graded, so coaching is held to the same evidence as selling. In the field, MADDOX on mobile is an installable web app — added to a phone’s home screen from the browser, not downloaded from an app store.

Moving in, and getting delivered.

Moving from Keap is a guided import of contacts, typed custom properties and segments, described on the Keap migration page. Once you are sending, deliverability covers sender profiles and the checks that decide whether a campaign is ready to launch — because the best-written email in the world does nothing from a domain the inbox does not trust.

The limits

What this platform deliberately does not do.

These are on the hub rather than buried three pages down, because a limit found late costs more than a limit found early — and because every one of them is the sort of thing a competitor will not volunteer about themselves.

No agent acts.

An agent here is a definition you write, scope, test in a sandbox and switch off. Nothing in MADDOX sends an email, creates a record, updates a field or assigns an owner because an agent decided to. When MADDOX does change something it is because a person approved a specific proposal, and what executes is read from the database rather than from the browser that asked. The long version is written up here.

Email is the only channel MADDOX operates for you.

A campaign lane can also carry an HTTP webhook step, which posts contact fields to an address you choose, and an SMS step that sends only where you have connected your own carrier account and taken it out of dry run. Those are things you can build, not things we operate on your behalf, and the webhook in particular works on a default install — which is why it is named here rather than left for somebody to discover.

Throughput is worth stating too: campaign steps are swept in a fixed batch every few minutes across the whole platform, so a large launch delivers over hours rather than at once. A due step can also be suppressed on purpose — unsubscribe, company fit, campaign exclusivity, a fatigue cap — and the reason is recorded rather than silently dropped.

The catalogue is not a directory of other companies’ applications.

What installs from it are campaign packs and templates — content, not code — and two paid MADDOX modules, IT Audit and its Network Scanner add-on. No code anybody else wrote runs in your workspace, and anything in the catalogue was either published by us or approved by us before publication. There is no third-party application ecosystem here and this page is not a soft way of saying one is coming.

Nothing above is a category we intend to enter.

Every pillar on this page is a working part of the product with its own pages behind it, and every one of those pages is written against the code rather than against a roadmap. Where something is partly built, off by default, or a preview of scope rather than a record of an action, the page says which — and when we get that wrong, the correction gets published in the field notes rather than quietly edited in.

One record, read many ways

The same call answers four different questions.

A call arrives from the softphone and becomes one row. From there it is a transcript on the contact, an activity on the deal timeline, evidence under a coaching commitment, and a scored behavior on the rep’s card — without anybody filing it four times, and without four systems holding four versions of it.

  • Captured once, at the point the work happens, rather than logged afterwards
  • Read by whichever pillar is asking, from the same row
  • Visible only where the role and the field permissions allow it
A speaker-attributed call transcript, agent lines in blue and customer lines in green, thirteen utterances long, with the most recent agent line highlighted and a talk-share bar beneath it reading agent 55 percent, customer 45 percent.

Features

Ask Maddox

Ask your CRM in plain English and get the answer with its sources — deals, calls, campaigns, coaching and reps. Nothing changes without your approval.

Ask Maddox overviewAsk your CRM in plain English and get the answer with its sources — deals, calls, campaigns, coaching and reps. Nothing changes without your approval. Ask about coachingAsk what was agreed at the last one on one, which commitments are still open, and what is on someone's development plan. It reports; it does not judge. Ask about contacts and companiesAsk about one person or one company and get their record read back: history, deals, emails, consent, buying committee and health. Plus the shape of your base. Ask about deals and pipelineAsk what happened on a deal, what it is signalling, which of yours need you today, or how the pipeline looks. Maddox answers from the deal record, not memory. Ask about performanceAsk how you are tracking, what your calls actually sounded like, and where a number moved. Withheld figures are reported as withheld, never as zero. Draft a campaign by describing itDescribe the campaign you want and Maddox builds a draft from your real lists and step types, ready in the builder. It cannot enrol anyone or send anything. Follow-ups, reminders and meetingsSay what you promised and Maddox proposes the task, or checks a calendar and proposes a meeting time. Nothing is saved or booked until you approve the card. How Maddox makes changesMaddox proposes, you approve, the server applies. What executes is read from the database, never from the request, and re-checked against the live record. The morning briefingWhat you owe, which deals need you and how you are pacing to target, assembled before you ask and emailed at your own local start of day if you want it. What Maddox will not doSome capabilities are unreachable by the assistant for every user, administrators included. Not a permission an admin passes, but absent tools.

Features

Sales coaching and 1 on 1s

Sales coaching software for MSP sales teams: every 1 on 1 prepped from real calls, every call graded on your rubric, and every promise tracked to a verdict.

Sales coaching and 1 on 1s overviewSales coaching software for MSP sales teams: every 1 on 1 prepped from real calls, every call graded on your rubric, and every promise tracked to a verdict. Coaching for AEsAE coaching on the deal itself: qualification, open objections, follow-up, risky language and close execution, with every meeting graded against your playbook. Coaching for BDRsBDR coaching for MSP outbound: weekly call targets judged day by day, same-day pace nudges, one definition of a DM reached, and calls graded on your script. Coaching for ownersFor MSP owners: one accountability score per sales manager, a weekly digest of what managers owe, waivers with reasons, and the record behind every number. Coaching for sales managersCoaching sales managers on how they coach: a different 1 on 1 document, graded on what they did in the room rather than on their team results for the quarter. Development plansSales rep development plans built from the record: structured items with evidence and exercises, measured against a baseline frozen when each item was made. Every call graded against your own rubricEvery sales call graded against your own rubric: gatekeeper, decision maker, objections, QA and script adherence for BDRs, and playbook steps for AE meetings. Evidence, not opinionEvery coaching finding cites real records, quoted verbatim. The evidence run is append-only, deterministic and verifiable back to its source, not opinion. Individualized from their own historyEvery 1 on 1 is built from that rep history: the trend across their last five cycles, frozen baselines and follow-through, so coaching continues, not restarts. Manager accountabilityOne accountability score per sales manager, built from nine weights the owner controls, plus a ledger of what is owed, promise verdicts and a weekly digest. Progress and commitmentsCoaching commitments and manager promises in one ledger: each gets a due horizon and a verdict from evidence, and a given answer cannot be quietly rewritten. Record the 1 on 1, get the grade and the summaryRecord or upload a 1 on 1 and MADDOX grades it against your coaching rubric with verbatim quotes, emails the rep a summary, and grounds the next prep in it. ScorecardsRole-aware sales scorecards for MSP teams: role-specific metrics, weights you control, call targets judged day by day, and not measurable in place of a zero. Self-assessmentRep self-assessment before the 1 on 1: three questions only the rep can answer, written before the meeting, so the talk starts from a gap you can both see. Since your last one on oneWhat changed since the last 1 on 1: the window is worked out from the record, and every coaching session inside it is read back item by item, basis stated. The coaching libraryA sales coaching library your managers actually reuse: your own advice by role, versioned, so editing a play never rewrites what a rep was told earlier. The one on one1 on 1 prep for sales managers: prior numbers, what was agreed, coaching since and current numbers, assembled from the record before you sit down with a rep. What will actually move the needleOut of everything a rep could work on, MADDOX ranks the levers by the gap to your own goals, so each coaching cycle targets the one thing most likely to help.

Features

AI Softphone

An AI softphone that reads the CRM before it dials and coaches live: when the objection lands, the rebuttal is in the rep’s hand mid-sentence.

Features

AI everywhere

Get AI on every job in the CRM without the runaway invoice: each feature runs on its own model, spend is metered per feature, and you set the cap.

Features

An objection library that learns from your calls

An objection handling library built from your own calls: plays your reps used to beat real objections, reviewed by managers and sent to the AI softphone.

Features

Buying signals and close opportunities

MADDOX reads your sales calls and email threads for what the buyer said that signals a purchase, quotes it, and shows whether your rep took the close.

Features

Contacts and companies

Contacts and companies for MSPs: your own objects, typed properties you can filter and segment on, field-level permissions, and the incumbent IT provider.

Features

Deals and pipelines

MSP sales pipeline software that asks why: a stage gate on required evidence, per-step qualification in your own wording, and lead sources you can report on.

Features

Email deliverability and sender protection for MSPs

Sender profiles with sending caps, a warm-up ramp, nightly health checks and DNS graded on six checks, so one bad send cannot burn the domain you bill from.

Features

MADDOX on your phone

MADDOX installs on your phone as a web app shaped to your role: the deal brief, your day, a canvassing map and push alerts, signed in for up to 30 days.

Features

Meetings and booking

Meeting scheduling for MSP sales teams: a themed public booking page, real availability re-checked at booking, round robins, and a record of every reschedule.

Features

Path to Close: budget, pain and the close

Path to Close tracks five closing steps on every MSP deal — pain, budget, decision maker, offer, open objections — with one budget record and a pain ladder.

Features

Quotes and proposals

Quotes and proposals for MSPs in your brand: your own numbering, a layout you preview live, letterhead built from your details, priced from your price books.

Features

Reports and dashboards

CRM reports and dashboards that read the same database as your records: no sync to fail, a drilldown from any number to its rows, and saved views on a schedule.

Features

Support and tickets

Helpdesk ticketing inside your CRM, beside your existing tools: live presence so two agents never answer twice, merge and split, routing by rule, an open SLA.

Features

Switch from Keap to an MSP CRM

Migrate from Keap to a CRM built for MSPs: a guided six-step mapping wizard, your history back-dated onto the timeline, and TurboDial calls transcribed.

Questions

The things people actually ask.

Do we have to adopt all of this at once?

No, and nobody does. Core CRM runs on its own and everything else is a module you switch on when you want it. The order that works is usually capture first, then the reading of it, then the parts that write outward – because a campaign built on a record nobody trusts is a faster way of being wrong.

What decides whether we can see a module?

Four things, in order: whether your subscription level explicitly denies it, then your workspace own module row, then your workspace settings, then the platform default. A level that says nothing changes nothing, and a level cannot switch on something you have switched off.

What happens to the systems we already run?

Where a system is the source of truth it stays the source of truth and MADDOX reads from it – that is how the PSA connectors are built, and they ship off behind four gates. The API and webhooks are open. What MADDOX will not do is quietly become a second copy of a record you already own somewhere else.

Is any of this AI making decisions on its own?

No. An agent is configured rather than dispatched: you write it, scope it, test it in a sandbox and switch it off, and no agent sends, creates, updates or assigns anything. The assistant reads; when something needs changing it proposes the change, you approve it, and what runs is read from the database rather than from the request.

Who can see what?

Permissions are role and field level, every record carries its workspace and the filter is applied by the model rather than by each query, and changes are written to an audit trail you can query. Four domains – erasure, API keys, user and role administration, and marketplace mutation – are refused to the assistant and to a support operator regardless of role.

Where do we find what something costs?

You ask us, or you ask the partner who already runs your systems. There is no self-serve checkout and no published price list, because the right level depends on what you are already running. The pricing page explains what a level actually changes rather than what it costs, which is the part you can check before you talk to anybody, and the partner programme is the other route in – same platform, different person holding the relationship.

See it against your own pipeline.

MADDOX is opening to new teams by invitation. Ask for one and we set you up on your own pipeline, or bring us the one you already argue about and we will walk it with you.