We map every task in your business, build the agents to handle them, and validate they work. Then — and only then — the headcount conversation happens. 33 roles automated across 8 departments. £1.13M in annual savings.
Most headcount reductions fail because they remove people before replacing the work. The business loses capability, remaining staff absorb unsustainable workloads, and within six months the roles are quietly re-hired.
We do it the other way round. We embed with your team, document every task a role performs, build the AI agent or offshore process to absorb it, and validate that the replacement works at equal or better quality. Only then does the restructuring conversation happen. The business never loses a capability. It gains efficiency.
When we did this for a major UK e-commerce retailer, we automated 33 discrete roles across 8 departments over 8 months — each one tracked against the P&L. The total annual savings: £1.13M.
Every role goes through a three-layer assessment before any change is made.
We deploy agents for product descriptions, first-line customer queries, image generation, report compilation, and data analysis. These are purpose-built agents using Claude and other models — not generic chatbots. Each one is trained on your data and validated against the output of the person it replaces.
Tasks that require human judgment but not on-site presence — order processing exceptions, supplier communications, marketplace listing management — move to trained offshore operators at a fraction of the cost. We recruit, train, and manage them.
Only after layers one and two are validated does the headcount change happen. Some roles consolidate — two people become one with automation support. Others are removed entirely because the work no longer exists in its previous form.
We worked through the business one department at a time. Each month delivered a discrete, measurable change. No big-bang restructuring. No shock announcements. The business absorbed each change, validated it was working, and we moved to the next.
In Customer Experience, we deployed an AI chatbot agent that handled first-line queries — password resets, order tracking, returns status — reducing the volume that reached human agents enough to consolidate the team in two phases. In Finance, we built end-to-end process automation that replaced manual reconciliation and reporting. In Tech, we deployed AI coding assistants and automated testing pipelines that reduced the engineering workload enough to restructure the team.
The eight departments: Customer Experience, Finance, Merchandising, Buying, Data, Tech, Operations, and Studio/Samples.
Some of the biggest savings came not from individual role automation but from closing or restructuring entire functions.
We closed the photography studio entirely. Product imagery moved to a combination of supplier-provided images and AI-generated lifestyle photography. The quality was comparable; the cost was a fraction. The studio lease, equipment, and associated headcount were all eliminated.
The samples team was consolidated after we redesigned the buying process to use AI-assisted trend analysis and digital visualisation, reducing the volume of physical samples needed for range planning.
We also identified and executed the closure of a margin-negative international business unit. The data showed it had been losing money for three consecutive years once fully-loaded costs were included. Closing it freed headcount, warehouse capacity, and management attention for the profitable core business.
None of this was possible without deploying purpose-built AI agents first. These agents absorbed the routine work before any role changed. We built them, we deployed them, and we still run them.
Handles first-line customer queries — order tracking, returns, account issues — with escalation to human agents only for complex cases.
Automates product listing, pricing updates, and inventory sync across multiple marketplace channels.
Manages customer segmentation, triggered campaigns, and lifecycle automation that previously required dedicated CRM analysts.
Generates daily and weekly reports, flags anomalies, and produces the analysis that previously occupied a full-time data team.
Every initiative is tracked against the P&L. Each role change has a documented annual saving, an implementation cost, a payback period, and a risk rating. We report monthly showing cumulative savings against plan.
If an agent underperforms, the role stays. If an offshore transition doesn't meet quality standards, it's brought back. The data drives every decision. This is not cost-cutting by spreadsheet — it is measured, validated automation with a clear rollback path.
33 roles automated across 8 departments over an 8-month deployment. Every change backed by a working AI agent, offshore operator, or structural redesign. No capability was lost.
Workforce automation was the single largest initiative set within our ongoing engagement with this retailer. The savings freed budget for reinvestment into revenue-generating work and funded the AI agent deployments that made the restructuring sustainable long-term.
Read more about how AI agents should be deployed before headcount cuts.
No. We automate the work before any role changes. Every task is documented, an AI agent or offshore operator is deployed to handle it, and we validate the replacement works at equal or better quality.
The business can do everything it did before — often more — with fewer people and lower cost. The difference is we replace the work, not just the person.
Typically 6-10 months depending on scope. We work department by department — each one gets a discrete window where we deploy agents, train offshore operators, and validate everything works before moving to the next.
No big-bang restructuring. The phased approach gives the business time to absorb each change and spreads the HR workload.
Every agent runs in parallel with the existing human process before any role changes. If the agent doesn't match the quality or throughput of the human role, the role stays.
We track performance metrics for every agent — response accuracy, processing time, error rates — and only proceed when the data confirms the automation is ready.
We build the operational blueprint and the business case with full P&L impact. The actual HR process — consultations, redundancy packages, legal compliance — is handled by your HR team or employment lawyers.
We make the business operationally ready for the change. You handle the people process. Clean separation.
High-volume, repetitive roles with clear input-output patterns: first-line customer service, manual data entry, product copywriting, basic financial reconciliation, and routine reporting. These have the highest automation success rate and fastest payback.
Complex roles involving judgment, negotiation, or creative strategy are addressed last, if at all. The goal is to remove toil, not talent.
Not because the people aren't working hard — because the work itself hasn't been re-engineered. We'll map where AI agents can absorb the toil and show you the savings before anything changes.