CSAT PRO builds AI agents and automation for shops, practices and small firms — starting with the single job that's eating your week. Fixed price, live in weeks, and yours to keep.
Three messy sources, one pipeline, one agent that answers customers, and a dashboard you'd actually open. Live in weeks, not quarters.
Plain language, tight scope, fixed price. Bigger organisation with security review and procurement? The enterprise track is a different conversation entirely.
No platform to buy into and nothing to rip out. We work with the tools you already pay for.
Answering customers, chasing unpaid invoices, sorting the inbox, booking appointments. It handles the routine and passes anything odd to a person with the context attached.
For exampleA shop agent that answers "where's my order" from the till system, day and night, without waking anyone up.
The steps that happen by hand every week — retyping an order into the accounts, rebuilding the same report, sending the same reminder to the same fifty people.
For exampleJob sheets and photos from the van turned into a finished invoice before the team gets back.
Spreadsheets, till, bookings and accounts pulled together every night and reconciled, so you can ask a plain question and get a straight answer.
For exampleKnowing on Monday which lines actually made money last week, rather than finding out at year end.
Software the vendor stopped supporting, a database only one person understands, ten years of records you can't afford to lose. We move it and keep the history.
For exampleA practice moving a decade of client records to a supported system over a weekend, open as normal on Monday.
When four spreadsheets and a group chat have become the system, we build the small piece of software that replaces them — and document it so anyone can pick it up.
For exampleA scheduler that knows your travel times, your van stock and who's qualified for what.
A hard monthly cost ceiling, clear rules about what the agent may and may not do on its own, and a one-page runbook. Nothing depends on us being available.
For exampleAn agent that can issue a refund up to a set amount, and must ask a human above it.
Start with one. Most clients do a second only after the first has paid for itself.
Prices are quoted after a 30-minute call, once scope is clear — add your own figures here before publishing. Discovery isn't invoiced if you decide not to proceed.
You'll have something usable in your hands well before the end, and you can stop after any stage.
A couple of sessions with the people doing the job. We map where the hours actually go and pick the one problem worth solving first.
A plain-English spec, the price, and a clickable sketch. If it isn't right, we change it now rather than after it's built.
A short demo every Friday. You see progress weekly and can redirect us at any point.
One person or one product line first, then everyone. The old way stays available until the new one has proved itself.
Documentation, a recorded walkthrough and a support line. Most clients are running it themselves within a month.
Nothing is locked to us. You own the code and the accounts. If you want to take it in-house or hand it to someone else, everything you need comes with it.
Book a 30-minute call →Three engagements in detail, then twelve places we see the same problems repeat.
Twelve places we see the same problems over and over — and the build that usually fixes them.
An agent connected to your till and stock system answers "do you have it, when's it back, where's my order" — on the website, by text and on social.
Fewer interruptions during trading hours.Reminders and rebooking handled automatically, digital intake forms landing straight in your records system, and insurance eligibility checked before the visit. Admin only — never clinical advice.
Reception freed for the people in the room.Six-month recalls, hygiene reminders and treatment-plan follow-ups sent on schedule, with replies routed back to the diary without anyone re-typing them.
A fuller book without extra phone hours.Reservations and takeaway orders reconciled nightly with your POS, plus supplier invoices read automatically so you see when a cost has crept up.
Margin visible weekly, not at year end.Handles enquiries and rescheduling around the clock, offers cancelled slots to the waitlist first, and nudges clients who haven't been in for a while.
Empty chairs offered out automatically.Customers get status updates as the job moves, approvals for extra work come back by text, and estimates draft themselves from your parts pricing.
Approvals in minutes instead of days.Vaccination and check-up reminders, repeat prescription requests triaged before they reach a vet, and client records pulled together across sites.
Routine requests handled before opening.Missing documents chased automatically, receipts and statements read and filed to the right client, and deadline reminders that escalate as they get close.
Less chasing in every busy season.New enquiries qualified and conflict-checked before they reach a fee earner, and routine documents drafted from your own precedent bank for review.
Billable time spent on billable work.Failed payments retried and followed up, class waitlists managed automatically, and members who've stopped turning up contacted before they cancel.
Churn caught early, not at renewal.Portal enquiries answered within minutes, viewings booked against agent calendars, and feedback collected and summarised for the vendor.
First response measured in minutes.Quotes drafted from your price list, jobs scheduled around travel time, and completion photos and notes turned into an invoice the same day.
Invoices out same day, not next week.Figures and outcomes shown are representative of engagement types. Replace them with your own verified client results before publishing.
CSAT PRO delivers AI agents, orchestration and complex migrations alongside your existing teams — to your architecture standards, your security review and your procurement process. Proof of value first, programme second.
Governed ingestion across systems of record, an orchestration layer with evaluation and rollback, and agents that write back into the tools your teams already use.
Architecture, security review and phased delivery. If you're a smaller team wanting one thing fixed at a fixed price, the small business track is a faster route.
We integrate with systems of record rather than replacing them. Code lands in your repositories and deploys through your existing pipelines.
Shared tooling across use cases: retrieval, action-taking against systems of record, permission scoping per role, evaluation harnesses and human-in-the-loop gates wherever the decision carries risk.
Delivered withEval suites, prompt and tool versioning, full request-level audit logging.
Process mining to find where work actually stalls, then orchestration that respects existing approval chains, separation of duties and regional variation.
Delivered withControl mapping, exception routing, and reporting against your existing KPIs.
Incremental and CDC ingestion, schema drift handling, data contracts, lineage, quality tests and SLAs — into Snowflake, Databricks, BigQuery or your existing warehouse.
Delivered withdbt models, orchestration in your scheduler, freshness and volume alerting.
On-prem to cloud, ERP and CRM consolidation after acquisition, database re-platforming, monolith decomposition. Zero-downtime patterns where the business requires them.
Delivered withPhased cutover, dual-running, reconciliation evidence and a tested rollback at every gate.
Internal applications and services built against your design system, deployed through your CI/CD, integrated with your identity provider from day one.
Delivered withYour repositories, your review process, your environments and release gates.
Model risk documentation, DPIA support, red-teaming, role-based access, cost governance, and enablement so your platform team runs it after we leave.
Delivered withRunbooks, on-call handover and a documented exit plan agreed at kick-off.
Three phases, each with its own gate. You can stop at the end of any of them, and the exit plan is agreed before we start.
Stakeholder interviews, system inventory, data access review and a written, signed-off problem statement with a prioritised opportunity list.
Commercial: fixed fee. Output is yours whether or not you continue.
One use case built in a non-production environment against real data, with agreed success criteria, evaluation results and a security review completed before any gate.
Commercial: fixed scope, fixed price, go/no-go at the end.
Phased rollout with your teams embedded, sprint cadence in your tooling, and enablement running alongside so ownership transfers as you go.
Commercial: milestone-based or T&M, priced per phase against your MSA.
Durations shown are typical ranges and vary with scope, data access and review cycles. Adjust to your own delivery history before publishing.
We expect security review as part of delivery rather than an obstacle to it. Happy to start under NDA.
Deployment inside your own cloud accounts and network boundaries. Data residency stays wherever your policy requires it.
Integrated with your identity provider. Role-based scoping so an agent can only reach what the person it acts for could reach.
Request-level logging of inputs, tool calls, outputs and human overrides, retained to your schedule and queryable by your teams.
Documentation aligned to your existing model risk framework, plus DPIA support and red-teaming before anything reaches production.
PII classification, masking and minimisation applied in the pipeline rather than trusted to the model. Contracts and lineage make the flow provable.
A documented exit plan agreed at kick-off. Code in your repositories throughout, so there is nothing to extract if you part ways.
Each stage produces an artefact your governance functions can review before the next one starts.
Stakeholder interviews, system inventory and data access review, signed off before design begins.
Integration and security design, evaluation criteria and cutover plan, reviewed with your architecture board.
Sprint cadence in your tracker, code in your repositories, environments promoted through your existing pipelines.
Dual-running and reconciliation evidence, go/no-go at each phase, and a rollback path tested rather than assumed.
Runbooks, on-call arrangements and enablement for your platform team, against the exit plan agreed at kick-off.
Procurement-ready. We work to your MSA, complete your security questionnaires, and can operate entirely within your cloud tenancy under your change control.
Arrange a scoping call →Three engagements in detail, then the eight functions where a scoped proof of value usually pays for the wider programme.
The eight areas where a scoped proof of value tends to justify the wider programme.
Retrieval agents that pull the full picture from every system a handler would otherwise open by hand, scoped by role and fully logged.
Handling time down, decisions still human.Automated matching across ledgers and bank feeds, with agents preparing exception packs for review rather than clearing them unsupervised.
Close cycles shortened, audit trail intact.Ingestion across ERP, WMS, carrier feeds and supplier portals into one governed model, with exception alerting on the things that move stock.
Disruption seen days earlier.Control evidence collected and structured automatically, with model risk documentation and red-teaming built into delivery rather than bolted on.
Reviews evidenced, not reconstructed.Ticket classification, enrichment and routing against your CMDB, with automated resolution only where the runbook already permits it.
Fewer reassignments per ticket.An assistant grounded in your own policy set that answers common questions and always shows the source, escalating anything sensitive to a person.
Consistent answers across regions.Activity capture, account enrichment and forecast hygiene, with reverse ETL putting the enriched view back into the CRM your reps already use.
Forecast built on data reps trust.CDC ingestion, data contracts, lineage and quality tests into a governed lakehouse, phased with dual-running and a tested rollback at each gate.
Migration evidenced phase by phase.Figures and outcomes shown are representative of engagement types. Replace them with your own verified client results before publishing.
We exist to close the distance between what AI can do and what it actually does for a business on a Tuesday morning.
Most organisations don't need another platform. They need someone to sit with the work, understand where it snags, and build the specific thing that unsnags it — then leave behind something the team can run and trust.
That's why we don't sell a product. We take the requirement seriously enough to build to it, whether it comes from a founder with fourteen staff or a transformation office with a three-year roadmap. The two need different conversations, which is why this site runs them as separate tracks.
Thirty minutes, no pitch deck. We'll tell you honestly whether this is worth building — and if it isn't, we'll say so.
A scoping conversation with the people who'd actually deliver. Happy to start under NDA and work to your procurement process.
Thanks — we'll come back to you within one business day with a time and a short agenda. If it's urgent, email sales@csatpro.com directly.