What it handles
- Failed steps and retries
- Opt-outs and consent
- Meta messaging windows
- Spam-trigger checks
- Approvals before sending
System · Automations
AI Automations
AI automations take over the repetitive part of your team's day. They capture and qualify leads, turn one recording into a week of content, send outreach, chase lapsed customers and move data between your tools, with AI doing the reading and writing in between. Each workflow is built around a process you already run.
- Inbound buyer lead matched to live listings
- < 60s
- Qualified property matches in three months
- 400+
- AI agents and automations built to date
- 150+
- Form received
- AI qualifies
- CRM updated
What is an AI automation?
An AI automation is a workflow built around a process you already run, with AI doing the reading, classifying and writing that used to need a person. It's triggered by an event such as a form, a DM or a missed call, wired into your CRM, inbox and messaging tools, and runs on its own, 24/7 or on a schedule you set.
Every build includes a mapped workflow, the AI steps, integrations with your existing tools, approval points where you want them, testing on test runs before go-live, and a documented handover. You own everything we build. No ongoing licence fees to us, no lock-in.
- 14:02 · Missed call detected
- 14:02 · SMS sent: "Sorry we missed you. How can we help?"
- 14:05 · Reply: "Need a quote for a new kitchen tap"
- 14:05 · AI tags job type and urgency
- 14:06 · Lead added to CRM · booking link sent
CapturedTaggedBooking link sent
- Mon 09:00 · Podcast episode uploaded
- Mon 09:12 · AI transcript ready
- Mon 09:30 · 5 posts, 1 email and 1 blog draft written
- Mon 09:31 · Batch sent to you for approval
- Tue 10:00 · Approved posts scheduled
DraftedAwaiting approvalScheduled
Sample runs. Illustrative only, not real customer data.
What can you automate?
You can automate any task your team repeats in the same way: capturing and qualifying leads, following up for months, writing content in your brand voice, sending outreach, reminding customers and handling back-office admin. Most businesses start with one workflow and add the next once the first is earning its keep. Use cases fall into five groups.
| Use case | What the automation does |
|---|---|
| Lead magnet generation | Swaps a personalised, AI-written report for a lead's details, then delivers it with a link to book a call |
| Social DM lead capture | Turns keyword comments and DMs on Facebook and Instagram into tagged, pre-qualified leads, in GoHighLevel or ManyChat |
| Lead capture and routing | Pulls leads from forms, portals, email, DMs and missed calls into one CRM and alerts your team to the hot ones |
| Speed-to-lead | Acknowledges every enquiry the moment it arrives, qualifies it and routes it with context attached |
| Lead qualification | Scores and segments leads from form and DM answers so sales sees the best first |
| Lead nurture sequences | Follows up for up to 10–12 months, checking before every message whether the lead has already acted |
| Missed-call capture | Texts every missed caller straight back and captures the enquiry |
| Matchmaking | Matches buyers to properties, or one side of any market to the other, and creates a deal for every match |
| Client onboarding | Welcomes each new client, collects agreements and access, sets them up in your tools and chases what's missing |
| Referral and affiliate tracking | Gives every referrer a link or code and tracks each referral through to the reward |
| Use case | What the automation does |
|---|---|
| Content repurposing | Turns a video, podcast or interview into emails, social posts, reel scripts and a blog post in your brand voice |
| Social media content | Researches ideas, drafts platform-specific posts, gets your approval and schedules them |
| Email marketing | Writes and assembles on-brand campaigns in your email platform and reports on performance |
| AI teaser generation | Writes a teaser for every listing or opportunity, holding back anything that should stay private |
| Use case | What the automation does |
|---|---|
| Cold email outreach | Builds and verifies a prospect list, researches each prospect and sends personalised emails from warmed-up inboxes |
| Personalised email drafting | Researches a lead and writes a personalised email, saved as a draft or sent straight away |
| Sales call logging | Records, transcribes and summarises every sales call into your CRM |
| Appointment setting | Sends booking links at the right moment and runs reminders so meetings happen |
| Proposal and quote generation | Prices each lead by your rules and assembles a branded PDF proposal for review |
| Use case | What the automation does |
|---|---|
| Reminders and reactivation | Sends service and renewal reminders and wins back lapsed customers by email and SMS |
| Feedback and reviews | Asks every customer for feedback and a review at the right moment |
| Customer support | Classifies requests from email and forms, answers the standard ones and routes the rest |
| Upsell and continuation offers | Moves clients to an upsell stage before their programme ends and sends the continuation offer |
| Client and partner portal | Gives clients and partners their own logins to check status, swap documents and raise requests |
| Use case | What the automation does |
|---|---|
| Payment workflows | Turns each successful payment into the full handover, and chases failed payments by SMS and email |
| Document tracking | Tracks the documents every record needs and updates statuses as files arrive |
| AI valuation | Values a listing from your data and live market research, with a price range and its reasoning |
| Call routing and IVR | Sets up ring groups, IVR menus, overflow and voicemail across your phone lines |
| Inbox and request triage | Sorts incoming emails by type and urgency, answers the standard ones and routes the rest |
| CV screening | Scores each CV against the role's requirements and shortlists the best fits |
| Operations database | Moves spreadsheets and scattered tools into one linked database with live dashboards |
| Internal comms | Posts an alert to the right channel or person when a match, hot lead or deal update happens |
How does an AI automation work?
Every AI automation runs in four stages. A trigger fires, an AI step reads, classifies, scores or writes, the results are saved to your systems of record, and your team is notified or asked to approve before anything goes out. The workflows run on Make.com, n8n or Claude routines; Facebook and Instagram DM flows run in GoHighLevel or ManyChat.
Trigger
A form is submitted, someone comments a keyword or sends a DM, a call is missed, a file lands in a folder, a CRM record changes, or a schedule fires.AI processing
An AI step does the part that used to need a person: transcribing, reading, classifying, scoring, researching or writing, from your own knowledge base, brand voice guide and rules.Systems of record
The results are written to where your team already works: your CRM, an Airtable base, a Google Sheet or Doc, or your email or social platform.Notify or approve
Your team is notified, or asked to approve, before anything goes out. Once approved, the automation sends, posts, books or files it.
- Make.com
- n8n
- Claude routines
- GoHighLevel
- ManyChat
- Airtable
- HubSpot
- Google Sheets
- Slack
We pick the platform that fits your tools, volumes and budget.
Where does a person approve the output?
You decide where a person signs off. In an AI automation, approval steps usually sit before anything public or personal goes out: lead magnets, personalised emails, social posts and blog articles. Approvals can be switched off for low-risk steps once you trust the output, and kept on for anything public.
| Workflow | What gets reviewed | Where |
|---|---|---|
| Lead magnets | Each personalised report before it's emailed (or sent instantly, if you prefer) | Airtable, email or Slack |
| Personalised emails | Each drafted email before it's sent (or sent straight away, if you prefer) | Your email drafts or CRM |
| Social media | Content ideas first, then the finished posts before scheduling | Your CRM, project tool or Airtable |
| Content repurposing | The weekly batch, with edits or tweak requests | Airtable content database |
| SEO articles | The AI draft, so a writer can add first-hand insight | Google Docs |
| Email marketing | Suggested changes to flows and timings, applied only once approved | Your email platform, with a summary to Slack or email |
| Cold email | Replies, which go to your team to handle | Your sending tool's inbox or CRM |
What's included in every build?
Every AI automation we build includes a mapped workflow agreed before we start, your brand voice and knowledge base captured at onboarding, error handling so failed steps are retried and flagged, a log of every run, notifications when something needs a person, and testing on test runs before anything goes live.
- Mapped workflow: every trigger, step, branch and fallback agreed before we build
- Brand voice and knowledge base: your tone, niche language, offers and FAQs, so the AI writes like you
- Error handling: failed steps retried and flagged, never dropped silently
- Logging: every run recorded, so you can see what happened and when
- Notifications: alerts when something is ready, needs approval or needs a person
- Testing and refinement on test runs: every branch run with test data before go-live
How do automations work in your industry?
AI automations follow the same four stages in every industry, but each industry automates different paperwork. Real estate matches buyers to listings, coaching businesses run DM flows and upsells, recruiters screen CVs. Each industry page shows the workflows that pay off there, in that industry's own words.
What are the limits?
An AI automation can only do what each tool's API allows, and Facebook and Instagram limit when a business can message someone first. AI-written content is only as good as the brand voice and examples behind it. Running costs for platforms, AI models and sending tools are billed by the providers, and we estimate them before the build.
Guardrails
Sample settings. Agreed with you at kickoff.
Built-in guardrails
- Nothing sensitive goes out without approval unless you choose automatic sending
- AI output is checked against your brand rules, and emails are checked for spam triggers and compliance footers
- Marketing messages only go to contacts with the right consent, and opt-outs are honoured automatically
- Review requests go to every customer. We never filter out unhappy customers before asking.
- Facebook and Instagram flows follow Meta's messaging rules
- Built in your own accounts wherever possible, so the workflows and data stay yours
Known limitations
- Platform limits. Some features, such as editing live email flows in some platforms, need a workaround or a person in the loop.
- Meta's messaging rules. DM flows and broadcasts work within Facebook's and Instagram's limits on messaging someone first.
- AI output needs a voice. We refine it on test runs and recommend keeping approvals on for anything public at first.
- Running costs. Platforms, AI models and sending tools are billed by the providers and rise with volume.
- Deliverability. Email and SMS reach depends on sending reputation, domain setup and list quality, which nobody can fully guarantee.
- Keeping it current. When your offers, tools or processes change, the workflow and prompts need updating.
Automations we've built
Real builds, anonymised. Each case study shows the brief, the build and how it runs.
400+ qualified matches in three months
< 60s
Lead matched to live listings
From spreadsheets to one portal with AI matching and valuations
1 build
From a proven template
What's included in each package?
Every AI automation package includes a mapped workflow, error handling, logging and handover documentation. Core covers one trigger and one AI step and ships in days; Advanced adds multi-step AI writing and approvals over a few weeks; the Full Automation System is a custom pipeline delivered in phases. Every build is scoped and priced on the discovery call.
| Core Automation | Advanced Automation | Full Automation System | |
|---|---|---|---|
| In one line | One workflow. One trigger. Running on its own. | AI that writes, converses and decides across more steps. | A custom end-to-end pipeline, built and handed over. |
| Ideal for | One repetitive task with a clear trigger | Automating a whole process, such as content or lead follow-up | Teams and agencies running automation at volume |
| Example: missed calls | Missed-call detection, instant SMS, lead into your CRM, booking link | Adds AI two-way SMS that answers questions and books | Adds reactivation and reminder sequences on the same contacts |
| Example: content | Upload by folder or form, AI transcription, content stored, ready notification | Adds AI-written emails, social posts, reel scripts and blog posts | Adds brand-voice tuning, a rewriting pass and scheduling and publishing |
| AI | A single AI step: transcribe, summarise, classify or draft | Multi-step generation and conversation in your brand voice | A fully customised generation pipeline |
| Integrations | Your existing CRM or database | Several tools, such as CRM, calendar and email platform | Your full stack, including publishing tools |
| Approvals and notifications | Notifications when work is ready | Approval steps before anything is sent | Approval and feedback loops at every stage |
| Mapped workflow, error handling, logging, documentation | ✓ | ✓ | ✓ |
| Testing and refinement rounds | 1 | 2 | 3 |
| Build timeline | Days | A few weeks | A few weeks, in phases |
Add-ons, scoped before the build so there are no surprise costs: CRM or database setup (GoHighLevel, HubSpot or Airtable) · extra channels or platforms · scheduling and publishing · a review interface · conversational follow-up with an AI chat agent or AI voice agent · a reporting dashboard.
Prefer to pay monthly? See the Build Partner RetainerMake.com, n8n or Claude routines?
We pick Make.com, n8n or Claude routines by your existing tools, your volumes and your budget. If you already use Make.com or n8n, we'll usually build on it. Facebook and Instagram comment and DM flows are built in GoHighLevel or ManyChat. We recommend a platform on the discovery call, before anything is built.
We also work with Zapier where a client already runs on it.
- Automation
- Make.com, n8n, Claude routines
- Social DM flows
- GoHighLevel, ManyChat
- AI models
- Chosen per build, based on the task, availability and the latest versions
- Research
- Perplexity, web search and website scraping
- Transcription
- Whisper, Fireflies
- CRM and data
- GoHighLevel, HubSpot, Airtable, Google Sheets, Supabase
- Email and SMS
- Klaviyo, Instantly, GoHighLevel
- Content and publishing
- Google Docs, Dropbox, Meta Developer Platform, Buffer, Hootsuite
- E-commerce
- Shopify
- Calendars
- Google Calendar, Cal.com
- Notifications
- Slack, email, Telegram
What do we need from you?
To build an AI automation we need access to the tools involved, one contact who can grant access and approve output quickly, your brand voice, your knowledge base and examples of output you're happy with. For DM flows we also need your Facebook page and Instagram account connected to a Meta business account.
- Access to your tools: CRM, email platform, social accounts, calendar, file storage
- Your Facebook page and Instagram account, connected to a Meta business account, for comment and DM flows
- One point of contact who can grant access and approve output quickly
- Your brand voice: a style guide, or examples of content you're proud of
- Your knowledge base: services, offers, FAQs and anything the AI needs to write accurately about you
- Examples of good output, so we can refine the AI to your standard
- Sending setup where needed: verified email domains and a business number for SMS
How long does it take?
A simple AI automation ships in days. Multi-system builds usually take a few weeks, and the Full Automation System is delivered in phases. After launch we watch the first live runs with you, and bugs found in the first 2 weeks after go-live are fixed at no extra cost under the warranty.
| Step | What happens |
|---|---|
| 1. Discovery | We map the process as it runs today: who does what, in which tools, how often and where it breaks. |
| 2. Workflow design | We agree the trigger, every step, the AI's job, the approval points and what "done" looks like. |
| 3. Onboarding | We gather access, your brand voice, examples of good output and your knowledge base. |
| 4. Build | Integrations first, then the AI steps and prompts, then notifications, approvals and error handling. |
| 5. Testing and refinement | We run test data through every branch, you review the output, and we refine from both. |
| 6. Go-live and handover | A recorded walkthrough plus documentation: workflow map, scenario exports, field mapping and prompt library. |
| 7. Launch check | We watch the first live runs with you to confirm every step works as tested. |
2-week warranty. Bugs found in the first 2 weeks after go-live are fixed at no extra cost. After that, support is optional: Maintenance, Active Management or Growth Partner.
Frequently asked questions
Questions
Answer
Will AI-written content sound generic?
Not if the AI automation is set up properly. We capture your brand voice, niche language and examples at onboarding, and the AI writes from those. During testing we refine the prompts on your feedback, and approval steps let you edit or request changes before anything goes out.
Do I have to approve everything an automation produces?
No. You choose where approval steps sit in each AI automation. Most clients keep them on for public content at first, then switch them off for low-risk steps, such as internal notifications, once they trust the output.
Should we use Make.com, n8n or Claude routines?
Whichever fits your tools, volumes and budget. We recommend one on the discovery call. If you already use Make.com or n8n, we'll usually build on it. Facebook and Instagram comment and DM flows are built in GoHighLevel or ManyChat instead.
What does an AI automation cost to run?
An AI automation runs on third-party tools billed by usage: the automation platform, the AI models and any sending or data tools. Costs rise with volume. We estimate them before the build, so you know what the system costs to run as well as to build. The build itself is scoped and priced on the discovery call.
Can you work with the tools we already have?
Usually, yes. If a tool has an API or webhook, an AI automation can connect to it. If it doesn't, we'll suggest an alternative or a workaround. If you don't have a CRM yet, we can set one up first.
What happens when an automation breaks?
Failed runs are retried and flagged, not dropped. Bugs found in the first 2 weeks after go-live are fixed under the warranty. After that, fixes and updates, including when the tools underneath an AI automation change, are covered by the optional monthly support plans.
Is a comment-to-DM flow a chat agent?
No. Keyword comment-to-DM flows, and DM flows that ask set qualifying questions, are AI automations built in GoHighLevel or ManyChat. If people need open-ended answers in their DMs, an AI chat agent can answer there too. Many businesses use both.
Can we change an automation later?
Yes. Prompt tweaks and small changes are covered by the optional monthly support plans. Bigger changes, such as a new workflow or a new channel, are scoped as a new piece of work.