Work has outgrown the setup
The business still runs, but too much depends on people moving context between disconnected tools.
Practical AI systems studio · UK · Pre-launch
Useful digital systems for the work your business has outgrown.
DG Workflow builds websites, automations, data flows, dashboards, and AI-assisted tools that help small businesses run with less manual effort and more useful information — with a person reviewing everything that matters.
Live simulation
Website enquiry captured — classified into 8 structured fields
Simulated activity · synthetic dataAI-assisted tools, Automation, Data flows, Dashboards, Web systems, Integrations, Internal tools, Human review built in
The starting point
The work that quietly eats the week.
Most ambitious small businesses do not need enterprise software. They need the scattered work between their website, inbox, spreadsheets, and reports to behave like one system.
Information lives apart
Websites, inboxes, spreadsheets, documents, and reports hold useful data that rarely joins up cleanly.
Decisions need clearer signals
Owners and managers need dashboards, summaries, and checks that make the current picture easier to trust.
AI needs a practical job
AI is useful when it drafts, classifies, extracts, or checks work inside a human-reviewed business system.
Before
A business area depends on manual handoffs between the website, inbox, spreadsheets, and documents.
After
A practical system captures the important fields, keeps uncertainty visible, and prepares reviewed next steps.
Before
Useful information exists, but it is hard to see quickly or trust without manual checking.
After
Data flows and dashboards turn the same inputs into a clearer operating picture and decision-ready views.
Before
AI is discussed as a separate tool rather than part of how the business actually works.
After
AI-assisted drafting, classification, and extraction sit inside a wider system with human review and handover.
The capability stack
Everything orbits one system.
The tools rotate — AI models, automation platforms, data stores, web frameworks. The job in the middle never does: a business system a person can understand, trust, and hand over. Hover any node in the orbit to see what it does.
AI-assisted tools Draft, summarise, classify, extract, and check work with human review kept visible.
Automation Turn repeated handoffs into scoped, tested system steps across the tools the business already uses.
Data flows Shape inputs into cleaner records that can drive follow-up, reporting, and handover.
Dashboards Create decision-ready reporting views from data people can understand and trust.
Web systems Modern websites that support enquiries, content, follow-up, and operational visibility.
Integrations Connect existing apps, inboxes, sheets, APIs, and lightweight databases where it adds value.
Internal tools Small interfaces, trackers, review screens, and handover packs for repeatable work.
ToolOpenAIAI extraction, summarisation, and assistant flows.System layerAI-assisted toolDrafting, extraction, and checked assistant flows.System layerHuman-reviewed outputReview gates before customer or business action.System layerAutomationRepeated handoffs turned into repeatable system steps.
ToolAstroFast static websites and service pages.ToolReactInteractive tools, demos, and dashboards.ToolTypeScriptTyped front-end and system code.ToolTailwind CSSResponsive interfaces with practical polish.ToolPythonData cleanup, scripts, and automation glue.ToolSupabaseDatabases, auth, storage, and prototypes.ToolPostgresStructured business data and reporting tables.
ToolCloudflareHosting, Workers, and edge deployment.ToolGitHubVersioned delivery, review, and deployment flow.Tooln8nAutomation across business tools.ToolMakeNo-code and low-code automation scenarios.ToolZapierLightweight app-to-app system links.BIReportingPower BIOperational dashboards and reporting views.System layerWeb systemsSites that support enquiry and follow-up work.System layerEnquiry systemInbox messages turned into clearer next steps.System layerCalendar draftsMeeting preparation that stays reviewable before invites go out.System layerTask queuesChecked records turned into visible owner and next-action lists.System layerData flowsRows and records shaped into usable business data.System layerDashboardsNumbers translated into useful review views.
Proof, not promises
Watch one enquiry become a system.
This is our flagship demonstration: a fake local-services business, one scattered customer request, and the exact journey from noise to a reviewed, connected operating picture. It runs itself — or take the controls.
Signal to System Pipeline
Synthetic data only
Capture scattered customer signals.
Website, email, phone, sheet, and job-note details are kept visible rather than flattened into one vague note.
Website enquiry
websiteUrgent maintenance request
Private landlord asks for a same-week leak inspection and wants to know if the website lead-time promise is still accurate.
Email note
inboxFollow-up details
Customer adds photos, access instructions, and asks whether the same visit can include a recurring service quote.
Phone note
call noteAppointment preference
Caller can do Thursday morning, but only if a property manager and engineer are both included on the invite.
Operations sheet
sheetCapacity row
Harbour Services | leak inspection | owner unassigned | value unknown | lead time page may need checking
Job note
documentRisk and context
Previous visit mentions restricted rear access and a recurring issue near the external drain.
Synthetic demo. Nothing is sent, scheduled, published, stored, or connected to a live system.
Open the full walkthrough0
Useful system first
One working improvement before any bigger programme.
0
Real client records in demos
Every public example runs on synthetic data.
0
Delivery steps
From fit screening to a documented handover.
0%
Important outputs reviewable
AI drafts and structures; people approve.
Recommended first step
Start with a Website and Workflow Review.
Before a rebuild, dashboard, or automation sprint, the review works as a practical Systems Review and recommends the first useful system worth discussing — like the one you just watched.
Map the current path
Where websites, enquiries, data, dashboards, automations, and internal tools should connect.
Find the safe first system
The smallest useful system that can create clearer action without a large software project.
Keep limits visible
Assumptions, data notes, handover needs, and human-review points stay part of the recommendation.
The gallery
Systems you can watch working.
Every piece in the gallery is a live, code-native demonstration built on synthetic data — the kinds of systems we build, moving on the page instead of described in a paragraph.
Wheel / drag to browse · synthetic demos
The service ladder
Start with one business area.
The normal first step is a Website and Workflow Review: a low-risk look at where one useful system would make the biggest difference. Everything else builds from there.
Website and Workflow Review Start here The Systems Review entry point: find where your website, enquiries, data, dashboards, automations, and internal processes should work together better. Systems review map Data and AI suitability First useful system Workflow Automation Sprint Available to discuss A focused business-system build around one repeated task, using AI, APIs, automation, and human review where they help. Structured inputs Review gates Handover notes Website Modernisation Available to discuss A clearer, faster customer-facing system layer that supports enquiries, follow-up, and the work behind the website. Clear service pages Better enquiry path System-aware content Dashboard and Reporting Improvement Available to discuss Cleaner inputs, repeatable reporting, and dashboard views that turn scattered data into a practical decision system. Data cleanup Decision view Repeatable reports Ongoing Improvement Retainer Future support route After a delivered project, keep websites, automations, dashboards, and internal tools improving through scoped support. Small fixes System tweaks Clear response limits
How a project runs
Careful is a feature.
Six steps, in writing, every time. The discipline around the build matters as much as the build itself — that is what makes a system safe to hand over.
01
Screen fit
Check the business, system opportunity, data sensitivity, and boundary fit before treating it as a project.
02
Discovery
Understand the business area, current tools, inputs, outputs, and what useful progress means.
03
Review or proposal
Recommend the safest first useful system, with assumptions, effort, data notes, and expected outcomes visible.
04
Written scope
Agree deliverables, exclusions, client responsibilities, review points, support, and change control in writing.
05
Build carefully
Use test, synthetic, or minimised data first, then test the system with realistic examples.
06
Handover and support
Document how it works, where it can fail, what needs human review, and what support is included.
Data and AI posture
Boring promises, kept.
Minimum useful data
Use only the fields the system needs. Public demos use synthetic business data.
Human-reviewed AI
AI can extract, summarise, classify, or draft, but important outputs stay reviewable by a person.
Synthetic demos only
Public examples do not use real client records, testimonials, confidential data, or employment-related data.
No autonomous high-impact decisions
Early systems should support people, not make legal, financial, HR, medical, safety, or regulated decisions.
Who builds this
Founder-led, by design.
DG Workflow is run by one builder who scopes, builds, documents, and hands over every system — with the discipline of someone who has to stand behind what they ship. No account managers, no hand-offs, no inflated team page.
More about the approachPre-launch
Bring us one workflow.
DG Workflow is currently in pre-launch. You can get in touch to discuss a business system, workflow, or practical improvement — online purchase and final payment are not active yet.
No pressure chat Future-fit conversation
Practical advice Honest and helpful
Clear next steps If it is a good fit