UK SaaS sprawl: stop buying point tools, buy an operating layer in 2026

By Dr Aalok Shukla, CEO · Published 9 May 2026 · Updated 9 May 2026 · 8 min read

Forrester says enterprises will defer 25 per cent of AI spend in 2026

The number frames the year. Per Forrester's 28 October 2025 press release on its 2026 technology and security predictions, enterprises will defer 25 per cent of planned AI spend into 2027. The reason is operational, not financial. Fewer than one-third of decision-makers can tie the value of AI to their organisation's financial growth, so CEOs will lean more on CFOs to approve AI investments based on provable ROI in 2026. Sharyn Leaver, Forrester's chief research officer, framed the year directly: "In 2026, the AI hype period ends as the pressure to deliver real, measurable results from secure AI initiatives intensifies."

The signal under the deferral is not that AI does not work. It is that AI bought as another point tool inside a single product surface does not move the P&L, because the value of automation lives in the seams between systems, not inside any one of them. UK mid-market boards are catching up to this. The deferral is rational. It is also a window: firms that switch from buying tools to building an operating layer over the tools they already own get the lift while peers wait.

This is why AIOS Command (Implement AI's operational platform for connecting commercial systems and deploying AI operators) is built around the operating-layer thesis rather than as another SaaS product to add to the stack.

UK mid-market is buying more software and hiring more people in parallel

The spend trajectory is rising in two directions at once. Per Gartner's Q1 2026 forecast, global software spend reaches 1.44 trillion US dollars in 2026, with year-on-year growth revised back up to 15.1 per cent. The slowdown predicted in late 2025 did not arrive. Most of that incremental spend is going to price increases on existing licences and to AI add-ons inside SaaS apps the buyer already runs.

Headcount is rising alongside the spend. Per BDO's 2026 UK Mid-Market Survey of 500 mid-sized business leaders, 88 per cent of UK mid-market firms are increasing headcount in 2026, and 36 per cent identify investment in technology and AI as the primary route to growth. The two numbers should embarrass each other but currently coexist. UK leaders are buying more software and hiring more people, hoping the combination will eventually produce productivity. The Forrester deferral says the combination is not converting fast enough.

The corroborating UK data point is direct. Per Accenture's UK 2026 research surveying 1,891 employees and 510 business leaders, 46 per cent of executives report that AI has so far delivered little impact on profit and loss. The lift is missing because the AI sits inside individual SaaS products, helping the user of that product, not the operating model that runs across products.

The marginal next SaaS purchase has near-zero P&L lift

Three structural reasons sit underneath the missing lift in the UK mid-market.

First, every SaaS product solves a problem that lives inside its own surface. CRM solves CRM-shaped problems. Finance ERP solves finance-shaped problems. Support tools solve support-shaped problems. The buyer's actual problems sit between those surfaces. A missed renewal is a CRM signal that needs a finance action and a support follow-up. None of the three products is the right place to fix it. Adding a fourth product does not help.

Second, AI inside a SaaS app is a copilot for that app, not an operator across the stack. A copilot inside CRM drafts a better email. It does not reconcile contracts against billing, then ping the right human, then update the next renewal forecast. That motion crosses three systems. It is invisible to any single product's copilot.

Third, the cost shape is wrong for the work being done. SaaS is priced per seat. The work an operating layer moves is between seats: inventory across two warehouses, tickets that escalate from support to operations, deals that need procurement and legal in series, payroll exceptions that touch HR and finance. Per-seat pricing on a between-seat motion produces invoices that look high and outcomes that look thin.

None of this is an argument against SaaS. UK mid-market firms have, over the last decade, bought a rich and largely correct set of SaaS tools. The point is that the marginal seventh tool is the wrong purchase. The marginal investment that moves the P&L is the layer that operates over the six tools the firm already owns.

Connect and operate all your systems in one place: the operating-layer thesis

Connect and operate all your systems in one place. CRM, finance, support, contracts, calls, emails, integrations, and approval workflows feed a single signal layer. The insight team reads continuously. The action team responds. This is the operating model that turns the SaaS investment a UK firm has already made into a P&L lift the board can defend.

The two-layer model maps directly to the work that lives between products. The insight team is the read-only analysis layer that connects 900 plus tools into one signal graph and surfaces the gaps a single product cannot see. The action team is the named AI operators that act on what the insight team finds, under human approval. AVA (the revenue analyst) reads commercial signals across CRM, billing, and contracts. DEX (the deal-flow analyst) watches inbound RFPs and quote requests across email, portal, and CRM. LEXI (the support analyst) reads tier-one ticket flow against contract entitlements. KIA (the knowledge agent) keeps the policy library current as procurement and finance rules shift. KORA (the customer engagement agent) sequences the right next step back to the customer or counterparty.

Read alongside the integration library of 900 plus connectors, AIOS Workforce for the named-agent footprint, and the case-study library for examples in production. The integration depth matters because the operating layer is only as honest as the systems it can read.

A 90-day operating sequence for UK boards

The mistake to avoid is treating this as a tooling project. It is an operating-model decision the board owns. The sequence below maps to the four-meeting structure UK mid-market boards typically use for an investment review.

Day 0 to 30: connect and observe. Wire the read-only signal layer across the systems that matter to one function (commercial, finance, support, or operations). Do not deploy any agent in this phase. The output is a weekly read-out that names the gaps a single product would not see: stale renewals, ticket escalations against contract entitlements, quote requests sitting in email, payroll exceptions touching three systems. The point of the phase is to give the board a view of what the existing SaaS stack is currently missing, not to demonstrate automation.

Day 30 to 60: deploy at the seams. Pick one workflow that crosses at least three systems and is repeatable. Deploy a named agent with a named human owner. Every action is logged and reviewed. Track three metrics from day one: cycle time on the workflow, exception rate, and policy adherence. The cycle-time improvement here is what closes the Forrester deferral argument; the board can see the lift before approving a wider rollout.

Day 60 to 90: tighten and extend. Move the lowest-risk policy band to auto-approval. Higher bands stay under human review. Add a second workflow that depends on the first. Report cycle time, exception rate, and policy adherence to the board monthly alongside the existing SaaS spend. By the end of the window, the board narrative shifts from "what should we buy next" to "which workflows do we move next, with the tools we already own".

Two patterns separate UK mid-market firms that succeed from those that stall:

• The action team only acts on what the insight team has already read. Skipping the read-only phase produces agents that fire on incomplete context, which is the failure mode behind Gartner's earlier prediction that more than 40 per cent of agentic AI projects will be cancelled by the end of 2027.
• Every agent action has a named human owner with a written policy band. Without that, the board cannot defend the AI investment under the financial scrutiny Forrester forecasts for 2026, and the deferral argument wins by default.

What to put on the board paper

The metrics that matter shift the moment the buying decision moves from a single product to an operating layer. Seat counts and licence utilisation fade in importance because the value moves from inside seats to between systems. Four metrics stay diagnostic.

Track cycle time per workflow before and after the agent is live; the operating layer earns its keep by compressing the time work spends crossing systems. Track exception rate, defined as the share of agent actions that escalate to a named human, and the proportion of those escalations that turn out to be policy-correct. Track policy adherence, the share of agent decisions that fall inside the band the board approved. Track integration coverage, the share of existing systems that feed the insight layer; this is the leading indicator that the layer is honest about what it sees.

Read these alongside the data-silos guide and the UK SME productivity paradox piece. The data-silos frame and the productivity-paradox frame are the same problem viewed from infrastructure and from outcomes. The buying-side response stitches both into a single decision: stop buying point tools, buy an operating layer.

Frequently asked questions