The question being asked in procurement committees this spring is whether the outsourcing contract coming up for renewal should be renewed at all. The argument for not renewing is that the work the provider does — invoice processing, reconciliations, payroll administration, master data maintenance — is precisely the work that agent-based automation is now claimed to do at a fraction of the cost, without the offshore centre, the transition project or the three-year term. It is the right question to ask and the wrong framing to ask it in. The provider is not selling you labour. It is selling you the absorption of variability, and that is a harder thing to replace.
The argument for outsourcing was never that other people are cheaper at data entry. It was that somebody else guarantees the work gets done when three people resign in the same month
An agent deployment does not make that guarantee. You do.
What the provider actually supplies
Unbundle the contract and there are four distinct things in it, and only one of them is automatable. Capacity. Bodies doing volume. This is the genuinely substitutable component and it is usually the largest line. Elasticity. The ability to absorb a month-end spike, an acquisition, or a new country without you hiring. Automation supplies this well, arguably better. Process ownership. Somebody whose job is to know how your reconciliation actually works, to maintain it when your system changes, and to tell you when it is broken. Automation does not supply this at all; it consumes it. Risk transfer. A service level with a penalty, an indemnity, a named accountable executive, and an insurance policy. This is the part nobody prices and it is the part that matters at three in the morning. Replace the first two and you have made a real saving. Replace the third and fourth without noticing and you have moved risk onto your own balance sheet while calling it efficiency.
| Contract component | Question before substitution |
|---|---|
| Capacity | Which transaction work can be replaced? |
| Elasticity | Who absorbs spikes and new scope? |
| Process ownership | Who monitors, changes and explains the process? |
| Risk commitments | Which service levels, indemnities and coverage remain? |
Qualitative summary of this article's source text, not a measured outcome or performance estimate.
The internal cost that does not appear in the business case
An agent-operated process needs an owner, a control framework, monitoring, exception handling, change management when the underlying system updates, and someone who can explain it to an auditor. In a provider relationship, all of that was inside the fee. Count it honestly and the comparison usually comes out as a substantial saving rather than an order of magnitude. That is still a good outcome. It is not the outcome in the pitch deck.
The shape that actually works this year
The organisations doing well with this are not choosing between the two. They are renegotiating on a different basis: the provider keeps process ownership, service levels and risk transfer, and deploys automation itself, with the contract priced on outcomes rather than full-time equivalents and the productivity benefit shared on a defined schedule. That conversation is available right now, because providers know exactly what is coming and would rather restructure than lose the account. The leverage is at renewal and it does not last.
Practical Guidance for Agent Swarm BPO Strategy
- Unbundle the contract into capacity, elasticity, ownership and risk.
- Price the internal overhead you would absorb, including audit support.
- Automate the volume, retain the ownership — whoever holds it.
- Renegotiate on outcomes rather than headcount.
- Share the productivity gain on a defined, dated schedule.
- Keep the indemnity and service level whatever the delivery model.
- Test the provider's own automation rather than accepting the claim.
- Use the renewal window; the leverage disappears after signature.
The Regional Angle
The first consideration is that a meaningful share of regional back office work cannot be moved to an agent because it is not a transaction, it is an errand. Renewing a trade licence, submitting documents to an immigration authority, obtaining an attestation, presenting a cheque, collecting a stamped original from a government service centre — these require a physical presence and, frequently, a person who knows which counter and which officer. Providers here supply that capability and it is invisible in a contract that describes itself as finance and accounting outsourcing. Before assuming a process is automatable, check whether any step involves someone leaving the building, because that is the step that will still be there after the automation is deployed. The second is a cost comparison that works differently in the Gulf than in the markets these business cases are usually written for. The saving from automating an offshore team in a low-cost delivery centre is real but modest per head; the saving from automating a locally employed team carrying visa costs, accommodation allowances, annual leave passage and end-of-service accrual is considerably larger. That argues for automating the onshore portion first and leaving the offshore capacity alone for the moment — which is the reverse of the sequence most business cases propose, because most business cases were written for organisations whose expensive people are the ones doing judgement work. The third concerns what happens to the provider relationship in a market where relationships carry weight. Regional outsourcing arrangements are frequently held together by a long-standing connection between principals, sometimes with shared ownership history, and the decision to withdraw scope has consequences beyond the contract. That is a genuine constraint rather than a sentimental one: the provider may also be handling your government liaison, your local sponsor arrangements, or work in another entity of the group. The practical approach is to restructure rather than exit, take the automation benefit inside the existing relationship, and keep the goodwill available for the parts of the arrangement that are genuinely hard to replace.
The objection worth taking seriously
The strongest objection is that this defence of outsourcing has been made before every technology shift and has been wrong each time. Providers said the same about robotic process automation — that the real value was governance, elasticity and risk transfer rather than labour arbitrage — and the market repriced anyway, because clients discovered that a substantial part of the fee was indeed just people. Arguing that process ownership and risk transfer justify the premium is what an incumbent says when the labour component of its pricing is under attack, and the honest reading is that fees will fall sharply and the retained functions will turn out to be thinner than claimed. That is a fair account of the last cycle, and fees almost certainly will fall. The part that does not follow is the conclusion that the retained functions were never real. They were real, and the evidence is what happened to the clients who took automation in-house during the last wave: a meaningful number of them ended up rebuilding a small internal capability that looked remarkably like the function they had removed, staffed by people who managed bots instead of processing invoices. The cost did come down. It did not come down to the number in the original business case, because the ownership work did not disappear — it relocated, and it relocated to an organisation that had not budgeted for it. The right prediction is not that outsourcing survives intact. It is that the labour line collapses, the ownership line persists, and the question is only whose payroll it sits on.
Common Questions
Should we cancel a renewal to build this internally?
Rarely as a first move. Restructure the existing contract on outcome pricing, which gets you most of the benefit without taking on the transition risk and the retained-function build simultaneously.
Can providers actually deploy this well?
The better ones can, and have more process data to train against than you do. Ask for a demonstration on your own transaction set, not a reference case.
What should we never bring in-house?
Anything where the value is a guarantee rather than a task — statutory filings with deadlines, payroll on a fixed date, anything carrying a penalty you would rather someone else bore.
What should we expect over the next twelve months?
Expect providers to move aggressively to outcome-based pricing and to lead with automation in every renewal. Expect at least one large provider to announce a restructuring driven by this. Expect the first clients who insourced to report that the retained-function cost exceeded plan. And expect the best deals to go to whoever negotiated in the next two quarters.
Agent Swarm BPO Strategy — we unbundle what your provider actually supplies, then decide which parts an agent can hold and which need an owner.