AI Systems

Axial AI systems, in full.

Not a tool recommendation list, and not a subscription. These are built systems — deployed in your environment, wired into the tools you already run, and maintained by us. Each one is made of workers you can see, and each carries a written boundary: what it does, what it prepares for you, and what it refuses. Together they are the systems that install behind the flagship website. The diagnosis decides which of them you actually need first.

Tools don’t run themselves

A new platform appears every week promising to handle your calls, your follow-up, your content. Most businesses buy several, use none consistently, and are left with exactly the problems they started with.

The failure is not the tool. It is the absence of a system — a designed, integrated workflow that does a specific commercial job whether or not anyone remembers to use it.

What we install instead

We choose the right underlying technology for each job, configure it to your business, wire it into your CRM, phone and booking, train your team, and hand it over running.

What you get is a system that does the work — answering, qualifying, following up, asking for the review — without your daily intervention. Usually installed alongside the flagship website. Two surfaces, one diagnosis.

The systems

Engage one, or several. Each is a service line in its own right; which ones you install, and in what order, is what the diagnosis decides — not a pre-priced bundle.

01 / 10

AI Lead Intelligence System

Lead generation, research and prioritisation

Sources, cleans, enriches and scores prospects into one pipeline that ranks where sales attention earns most.

15
workers: 2
human gates: 6
steps

9 hrs → 4 min
first reply: 11% → 15%
conversion: +£9.4k
a month, from 150 leads

Modelled for a typical business. Change the inputs in the calculator ↓

What it is

Finds, de-duplicates, researches and ranks prospects, so the people who can close business spend their hours on the ones worth closing. Most lead tools optimise for volume; this one allocates the scarcest thing you have: sales attention.

Does unaided: Harvesting, resolution, enrichment and scoring, unaided.
Prepares for you: Every brief, draft and promotion to contact, for a person to approve.
Refuses: Contacting a suppressed, duplicated or already-owned record; merging two companies on a name alone; harvesting an unregistered source.

The workers

Source register keeper
Source harvester
Entity resolver
Decision-maker locator
Tiered enrichment runner
Account value scorer
Conversion-likelihood scorer
Evidence-strength grader
Promotion and suppression gate
Contact-route verifier
Account brief writer
Outreach drafter
Outreach sequencer
Pipeline and source reporter
Score recalibrator

How it works

01Harvest signalsdirectories, hiring posts, funding notices, competitor lists
02Resolve one recordfour sightings become one company, evidence stacked against it
03Enrich by tierresearch spend follows promise, not curiosity
04Score three waysvalue, likelihood to convert, strength of the evidence
05Promote to contacta ranked queue, with the reason stated
06Draft outreachaccount brief and first touch, ready to approve

Feedback: replies, meetings, margin and losses re-fit the scoring

What it changes

Todaya list worked from the top
With ita queue with a stated reason behind every position
Over timeranking runs on your own sales history, not our assumptions

What you receive

One canonical record per prospectbranches, aliases and decision-makers linked
Tiered enrichment and scoringconfidence and missing-data flags carried on every record
Promotion, watchlist and suppression controlscontact stays governed and never duplicated
Briefs and outreach draftsfor every promoted opportunity
Source and pipeline dashboardscost per promoted lead, and why each lead was ranked

Sales capacity, not lead count, is the binding constraint: the return is a higher expected value for every hour of outreach, and a cost per qualified opportunity that management can actually manage.

See the lead generation demo Build a more intelligent pipeline

AI Lead Intelligence System, drawn: 6 stages from signal to decision, 2 human gates, one shared ledger and 15 named workers.

Open this map at full size ↗

02 / 10

AI Web Concierge and Intake Agent

Website concierge and intake

An informed first point of contact that answers, qualifies and routes each enquiry to the right next step.

12
workers: 2
human gates: 6
steps

Over 1 hr → 60 sec
response time: 14 bookings
recovered a month: +£2.9k
a month, at £350 a booking

Modelled for a typical business. Change the inputs in the calculator ↓

What it is

An informed first point of contact on your site: it answers what it is permitted to, asks only what is relevant, and hands over the moment judgement is required. A conventional site gives everyone the same form and loses most of them without learning where.

Does unaided: Answers from approved knowledge, source attached, and hands over.
Prepares for you: Every booking request and every callback, for a person.
Refuses: Clinical, diagnostic and treatment content, and any question no approved source covers.

The workers

Safety-policy screen
Intent classifier
Approved-answer retriever
Progressive intake collector
Enquiry router
Handoff packager
Booking request raiser
Callback request raiser
CRM record writer
Question and gap logger
Journey analytics reporter
Operator correction capture

How it works

01Screen and classifysafety first, then what the visitor actually wants
02Answer from approvalreturned exactly as approved, source attached
03Ask what mattersintake adapts to urgency and value as it goes
04Route the enquiryby service, geography, urgency, value or language
05Hand overbooking or callback raised; the transcript travels with it
06Record and learnCRM record, unanswered questions, drop-off points

Feedback: the questions it could not answer become the next approved answers

What it changes

Todayout of hours is a closed door; the form loses the visitor
With itthe visitor leaves as a booking or a qualified record
Over timethe demand you are not serving becomes a list to act on

What you receive

A branded concierge interfacegoverned, versioned knowledge behind it
Progressive intake and routingby service, geography, urgency, value or language
Human handoff with contextthe full conversation, to the right member of staff
Booking, CRM and follow-up integrationsincluding out-of-hours capture
Conversation analyticsquestion themes, drop-off points, conversion by path

It compresses the interval between interest and action — the stretch of the funnel where enquiries are cheapest to win and most easily lost.

Talk to the concierge Turn website visits into guided journeys

AI Web Concierge and Intake Agent, drawn: 6 stages from signal to decision, 1 human gate, one shared ledger and 12 named workers.

Open this map at full size ↗

03 / 10

AI Reception and Call Intelligence System

Reception, intake and call analysis

Answers, routes and recovers calls, then converts the conversations into searchable operational intelligence.

14
workers: 3
human gates: 6
steps

39 bookings
recovered a month: 16 hrs back
reception time a month: +£2.5k
a month, from 120 calls a week

Modelled for a typical business. Change the inputs in the calculator ↓

What it is

Answers, routes and recovers telephone calls, and — separately, only if you want it — turns those conversations into something searchable. The phone is usually the highest-intent channel and the least instrumented; a missed call leaves no record.

Does unaided: Approved questions, availability search, callback and record.
Prepares for you: Every booking, cancellation, message and transfer, for a person.
Refuses: Clinical advice, diagnosis, fee and insurance decisions, and retention beyond your window; the clinical refusals cannot be switched on later.

The workers

Reception

Call router
Approved-answer lookup
Availability search
Booking and cancellation handler
Callback task raiser
Escalation router
Warm-transfer briefer
Follow-up message sender
Consent and retention guard
Call insight reporter

Intelligence, elected separately

Transcriber and summariser
Call-outcome classifier
Objection and theme miner
Quality-review sampler

How it works

01Answer and verifyfirst-touch calls, approved questions only
02Intake and routeavailability searched; anything complex or clinical stays human-led
03Book or callbacka slot held, or a callback task raised
04Chase missed callsby message or link, before they settle elsewhere
05Transfer warma briefing travels with the caller
06Transcribe, if electedsummary, outcome and objection, inside consent and retention rules

Feedback: call outcomes and themes feed the approved answers

What it changes

Todaycalls ring out at half four and leave no trace
With itanswered, or chased before that person settles on whoever did answer
Over timeevery conversation ends with a record, whether or not anyone was free

What you receive

Configured voice receptionapproved call flows, routing and escalation rules
Missed-call recoveryand out-of-hours enquiry capture
Telephony, calendar and CRM integrationwith warm-transfer briefings for staff
Transcription and outcome classificationwhere elected
Consent, retention and quality reviewaccess governance and a sampling process

Recovered calls are incremental revenue at close to zero marginal cost, and the analysis layer converts a standing operating expense into a continuous read on demand, objections and service quality.

Capture and understand every call

AI Reception and Call Intelligence System, drawn: 6 stages from signal to decision, 3 human gates, one shared ledger and 14 named workers.

Open this map at full size ↗

04 / 10

AI Content and Campaign System

Marketing, ad and content production

A production system that turns genuine expertise into campaigns, ads and multimedia without diluting one strategic voice.

14
workers: 3
human gates: 6
steps

4 hrs → 1.7 hrs
per piece: 46 hrs back
a month: 47 in the time of 20
pieces, if the hours stay

Modelled for a typical business. Change the inputs in the calculator ↓

What it is

Turns real expertise into campaigns, ads and multimedia at volume, without the output drifting from one strategic voice. The constraint in most marketing is throughput, not ideas: how many experiments a small team can run before coherence gives way.

Does unaided: Format resolution, variant fan-out, export of approved work, and the record.
Prepares for you: Every brief, creative direction, draft and media plan, for sign-off.
Refuses: Publishing anything, any unapproved claim, and any number or urgency phrase without an approved source.

The workers

Campaign brief writer
Creative director
Media plan allocator
Format resolver
Variant fan-out
Copy drafter
Image and video renderer
Voice and claims guard
Evidence-span checker
Review queue keeper
Render request emitter
Export and publishing seam
Competitor creative monitor
Performance and learning reporter

How it works

01One source inputa founder interview, a sales call, a complaint you keep hearing
02Frame the narrativeeach asset gets a job: capture, authority, objection, or category
03Plan the mediabefore anything is generated, which is what protects the budget
04Produce the variantscopy, image and video, fanned out by format
05Guard the claimsevery number resolves to an approved source
06Export approved workpublication itself stays with a person

Feedback: performance by message, audience and channel decides what is reused

What it changes

Todaymarketing depends on whoever last had capacity
With itproduction at a rate the business sets
Over timewhat works is reused deliberately; what does not is retired

What you receive

Narrative and positioning architecturecategory language included
Format-specific production workflowswith approved prompt libraries
Approval and brand-safety workflowcontrol levels matched to risk
Competitor creative monitoringand emerging demand-angle tracking
An initial campaign or content setwith a performance dashboard and learning loop

Cheaper content is the trivial gain; the real one is more creative experiments per unit of spend, which is the mechanism by which acquisition cost is driven down and then held down.

Build a repeatable demand engine

AI Content and Campaign System, drawn: 6 stages from signal to decision, 3 human gates, one shared ledger and 14 named workers.

Open this map at full size ↗

05 / 10

AI Commercial Intelligence System

CRM, lifecycle follow-up and commercial reporting

One trusted commercial view that runs lifecycle follow-up and turns reporting into ranked, measurable management decisions.

19
workers: 2
human gates: 6
steps

24 customers
back a month: 9 enquiries
recovered a month: +£4.1k
a month, at £220 a sale

Modelled for a typical business. Change the inputs in the calculator ↓

What it is

Two things usually bought separately, built as one: the lifecycle system that makes sure enquiries, reviews, rebookings and referrals are handled, and the analytical layer that shows what the business is genuinely earning, and where.

Does unaided: Ingestion, identity resolution, metric definitions, lifecycle tracking and the record, deterministically.
Prepares for you: Every follow-up, reply, referral prompt and management recommendation.
Refuses: Merging two customers on a name alone, soliciting a public review on a weak signal, and any clinical prediction.

The workers

Source ingester
Identity resolver
Metric dictionary keeper
Exception and approval queue keeper
Lifecycle stage tracker
Follow-up drafter
Review request sender
Feedback router
Review reply drafter
Reputation monitor
Referral prompt sender
Reactivation and renewal watcher
Cohort and retention analyst
Channel economics and leakage analyst
Prospective model runner
Commentary writer
Decision log keeper
Cockpit reporter
Warehouse pipeline runner

How it works

01Ingest every sourcebookings, finance, ad and web data
02Resolve identitiesone customer record, never merged on a name alone
03Reconcile definitionsone agreed meaning of customer, lead, conversion and margin
04Run the lifecyclefollow-up, review requests, referrals, reactivation
05Explain the movementwhich channel, cohort or service moved, and why
06Rank the decisionsexpected value, uncertainty and downside, each recorded

Feedback: every decision is reviewed later against what actually happened

What it changes

Todayreporting is a monthly ritual everybody half-believes
With itthe thing disagreements are settled with
Over timerecommendations become recorded decisions, scored against outcomes

What you receive

A governed data modelwith a metric dictionary agreed across the business
CRM and staged follow-upreview and referral workflows with staff task queues
An executive cockpitplus role-specific operating dashboards
Cohort, channel and leakage analysisand profitability by service
Forecasts as rangesand a decision log that records outcomes

Reduced leakage and better-allocated spend fall almost directly to EBITDA, and the same reconciled numbers are what a lender, investor or acquirer will ask to see.

See the CRM and recovery demo Turn fragmented data into decisions

AI Commercial Intelligence System, drawn: 6 stages from signal to decision, 2 human gates, one shared ledger and 19 named workers.

Open this map at full size ↗

06 / 10

AI Market Intelligence System

Competitor, pricing and demand monitoring

Continuous monitoring of prices, offers, hiring and demand, resolved into decision-ready reads on where the market moves.

12
workers: 2
human gates: 6
steps

6 hrs → 1.2 hrs
a week watching the market: 4 decisions
a year, on evidence: +£26k
a year

Modelled for a typical business. Change the inputs in the calculator ↓

What it is

Continuous monitoring of the prices, offers, hiring and demand signals around you, resolved into where your market is moving. Most monitoring is descriptive; the useful question is what the traces imply about capacity, margin and priorities.

Does unaided: Collection, change history, materiality filtering and evidence classification.
Prepares for you: Every inference about a named competitor and every recommended response.
Refuses: A modelled range presented as fact, calm reported when a source has failed, a thesis with nothing recorded against it.

The workers

Market universe keeper
Signal collector
Change and history recorder
Materiality filter
Evidence classifier
Source health and unknown-state reporter
Baseline deconstructor
Local demand mapper
Contrary-evidence tester
Intelligence brief writer
Scoped question answerer
Decision and outcome recorder

How it works

01Define the marketdirect rivals, substitutes, do-it-yourself, and choosing to wait
02Collect on cadenceprices, offers, hiring, reviews, ads
03Record the changemovement over time, not a snapshot
04Grade the evidenceobserved, inferred, or modelled with the assumptions written down
05Test the contraryno thesis released without evidence against it recorded
06Brief the responseonly the moves that matter, with confidence stated

Feedback: decisions and outcomes are recorded against each brief

What it changes

Todaya feed of competitor news
With itonly the moves that matter, each with evidence and a response
Over timepricing set against a corridor you can see, not an impression

What you receive

A competitor and substitute mapgrouped by why each one is monitored
Baseline deconstructionsbusiness model, pricing corridor, positioning, visible vulnerabilities
Pricing, offer and review historywith change tracked over time
Alerts and a regular briefwith response playbooks
Local-demand or market mapswhere geography drives the economics

In a market consolidating around fewer operators, the two costliest failures are reacting too late to a real move and spending capital reacting to noise; this system is built to prevent both.

See the market intelligence demo See the market before it moves

AI Market Intelligence System, drawn: 6 stages from signal to decision, 1 human gate, one shared ledger and 12 named workers.

Open this map at full size ↗

07 / 10

AI Proposal and Document System

Governed proposal, quote and report generation

Governed generation of proposals, quotes and reports that stay specific to the client without being rebuilt from zero.

10
workers: 2
human gates: 6
steps

3 hrs → 54 min
per document: 63 hrs back
a month: +£3.6k
a month, from 30 documents

Modelled for a typical business. Change the inputs in the calculator ↓

What it is

Governed generation of the documents you sell with — proposals, quotes, scopes, reports — specific to the client without being rebuilt from nothing each time. Firms sit at two extremes, generic templates or documents rebuilt at senior cost, and both are expensive.

Does unaided: Assembly, section selection, rendering and version history.
Prepares for you: Every drafted narrative, routed to its named approver.
Refuses: A model-produced price or term, release with an unsupported claim open, or a proof past its review date.

The workers

Document-estate auditor
Content library keeper
Client-fact assembler
Section selector
Narrative drafter
Price and terms binder
Unsupported-claim flagger
Evidence freshness keeper
Approval router
Renderer and version keeper

How it works

01Stable versus variablemethod, terms and proof on one side; this client on the other
02Assemble client factsfrom CRM, pricing and project sources
03Draft the narrativelanguage and synthesis are generated; nothing else is
04Bind price, termsfrom approved sources and deterministic rules
05Flag the unsupportedand check every proof is still within its review date
06Route, render, versionto the named approver, then out of the building

Feedback: the document carries on as plan, decision log and performance record

What it changes

Todaygeneric templates, or senior time on every deal
With itspecific, consistent with the last one, free of unapproved claims
Over timethe next conversation starts from work that already exists

What you receive

A document architecturewith a governed, reusable content library
Client-specific fields and rulesconnected to CRM, pricing and project sources
Drafting and approval workflowswith unsupported-claim flagging
Branded templates and version historywith a full audit trail
Live document familieswith training and governance documentation

It converts scarce senior time into a reusable asset, shortens quote-to-signature, and keeps the firm’s best commercial reasoning inside the business rather than inside one employee’s head.

Turn every proposal into a working asset

AI Proposal and Document System, drawn: 6 stages from signal to decision, 2 human gates, one shared ledger and 10 named workers.

Open this map at full size ↗

08 / 10

Bespoke Operational Systems

Custom operational engineering

Custom engineering for proprietary internal processes: agent pipelines, deterministic core, clean interface, human control at the decisions.

9
workers: 3
human gates: 6
steps

6.5 hrs → 2.8 hrs
a week, per person: 193 hrs back
a month, across the team: +£4.7k
a month, for 12 people

Modelled for a typical business. Change the inputs in the calculator ↓

What it is

The engineering service for the processes no off-the-shelf product fits: the particular sequence of information, judgement and action by which your business delivers what it sells.

Does unaided: Set per workflow, because the workflow is the product; the deterministic core does the things that must never drift.
Prepares for you: Every exception and approval, for the person accountable.
Refuses: Any worker without a written authority before it is built; the staff answering worker refuses to answer without naming its source.

The workers

Workflow mapper
Integration and export feasibility checker
Deterministic core runner
Document and record extractor
Exception and approval runner
SOP writer
Staff question answerer
Procedure drift detector
Monitoring and reporting runner

How it works

01Map the flowtrigger to handoff, as it really runs
02Choose the layerconfigured software, integration, or custom code where justified
03Code versus AIidentifiers, permissions and state in code; extraction and drafting in AI
04Set worker authorityobserve, recommend, approve to execute, or act within limits
05Run the exceptionsan inbox for the person accountable
06Watch for driftprocedures written down, changes gated

Feedback: cases, exceptions and throughput are reported back into the map

What it changes

Todaythe process lives in one person’s head
With itdocumented software: queues, case records, an exception inbox, reporting
Over timethe process survives the person who used to hold it

What you receive

A workflow and opportunity mapwith technical and commercial scope
Working softwarewith an employee or customer-facing interface
Data, integration and permission architecturebuilt around one source of truth
Approval and exception workflowswith tested failure paths
Monitoring, reporting and a maintenance planplus documentation and training

It raises throughput without proportional headcount, expanding contribution margin as volume grows and leaving behind an operating capability competitors cannot buy off the shelf.

Build the system your business is missing

Bespoke Operational Systems, drawn: 6 stages from signal to decision, 3 human gates, one shared ledger and 9 named workers.

Open this map at full size ↗

And the surfaces they install behind

The systems above are the operating layer. They earn most where they sit behind a front door built for the same standard — the other half of the two-surface build.

09 / 10

Experiential Website

Where the systems are met

A cinematic, sector-defining flagship surface optimised for trust, capture, booking and analytics.

6
workers: 2
human gates: 6
steps

1.5% → 3%
visitors who enquire: 60 more enquiries
a month, same visitors: +£6.8k
a month, from 4,000 visitors

Modelled for a typical business. Change the inputs in the calculator ↓

What it is

A flagship built as two surfaces rather than one compromise: one creates attention, ambition and memorability; the other explains, reassures, captures demand and connects to the machinery behind the business. Forced onto a single page, neither job is done properly.

What governs it: A surface the agent systems are met through. Authority is the wrong frame for a website; what governs it is consent, accessibility and the honesty of what it shows.

The workers

Enquiry capture surface
Journey analytics capture
Concierge surface mount
Accessibility and performance check
Consent state keeper
Demonstration access gate

How it works

01Decide the splitone site, flagship plus commercial surface, microsite, or site plus portal
02Commercial thesisjourney, information architecture, conversion copy
03Storyboard chaptersthe visual world planned before any media is made
04Produce the mediafilm, motion, 3D and typography, holding continuity
05Commercial layerproof, pricing logic, booking, search, CRM routing
06Sign off, measureattribution end to end, from arrival to booking

Feedback: journeys and drop-off feed the next chapter

What it changes

Todayyour category’s usual standard
With itpeople arrive expecting that standard and do not get it
Over timeattention converted into booked, tracked, attributable demand

What you receive

A commercial and creative thesiswith information and journey architecture
Positioning, messaging and conversion copyfor both surfaces
A produced visual systemimagery, video, motion and interaction
The working experiential siteand, where required, its commercial sister surface
Booking, CRM, analytics and AI integrationswith performance and accessibility QA

Distinctiveness a competitor cannot cheaply reproduce supports premium positioning, while the commercial surface is what converts that attention into booked, tracked and attributable demand.

See the website directions Build the flagship in your category

Experiential Website, drawn: 6 stages from signal to decision, 1 human gate, one shared ledger and 6 named workers.

Open this map at full size ↗

10 / 10

Mobile Phone App

Where the relationship is held

A direct line of communication with your customers facilitating initial contact, purchasing and ongoing follow up.

6
workers: 2
human gates: 6
steps

+12% repeat visits
from the app: 33 more visits
a month: +£15k
a year

Modelled for a typical business. Change the inputs in the calculator ↓

What it is

A bespoke iOS and Android application, built only where a recurring job justifies it. An app is not automatically worth more than a website; it earns its cost when the phone does a repeated job better than a browser can — holding identity, prompting at the right moment, capturing structured data, carrying a programme.

What governs it: A surface, like the website. What governs it is identity, permission, consent and retention.

The workers

Identity and session holder
Role and permission gate
Notification dispatcher
In-app capture and history
Offline capture queue and reconciler
Consent, retention and deletion keeper

How it works

01Name the jobthe recurring reason a customer would return to it
02Map the rolesuser, payer, administrator, provider: each its own interface and permissions
03Ship the minimumaccount, one core journey, notifications, history, support
04Connect, not copythe website, CRM, booking system and customer record you already run
05Prompt on timenotifications a person has approved
06Read the evidenceadoption and repeat use decide the next release

Feedback: later features come from observed behaviour, not ambition

What it changes

Todaya platform is paid to put you in front of customers again
With ita surface that remembers them and reaches them directly
Over timethe first release establishes whether the recurring job was ever real

What you receive

A product and commercial thesiswith user and workflow architecture
An MVP specificationwith UX and visual design
The working iOS and Android applicationwith secure backend and integrations
An administration interfacewith analytics and monitoring
Release support and a roadmapwith documentation and training

The economics of an app are frequency and switching cost: more repeat purchase, stronger retention and better first-party data, provided the recurring job is real enough to justify the build.

Try the working app Put the service in your customer’s hands

Mobile Phone App, drawn: 6 stages from signal to decision, 2 human gates, one shared ledger and 6 named workers.

Open this map at full size ↗

Built for your stack

Every system integrates with the CRM, phone, email and booking tools you already run. No migration required.

Maintained over time

AI systems degrade without maintenance. Retainer clients get monthly reviews, optimisation, and priority support.

Diagnostic-led

We only recommend a system once the diagnosis names it as a priority leak. No pre-priced bundles, no upsell you did not need. See the method →

Find out which system your business needs first.

The diagnostic names your highest-cost leak. The recommendation follows from the numbers, not a sales agenda.

Run the diagnostic Send us your site