Are VPNs Banned in the UK? AI Agents Reshape SaaS Pricing

Per-seat SaaS pricing is cracking as AI agents do the work. See real 2026 contract numbers, hybrid pricing tips, and whether VPNs are banned in the UK.

## The Invoice That Broke the Seat Model Last month a procurement lead at a mid-sized UK logistics firm showed us a renewal quote from an AI support vendor. Same 40 seats, same plan name, and a 34% price increase justified by "expanded agent capabilities." Her team had cut ticket volume by half using those agents. She was paying more for less work. That is the subscription era's tell, and it is why so many of you keep asking whether **[are vpns banned in the uk](/blog/nvidia-s-jensen-huang-on-ai-doomsday-why-you-shouldn-t-panic)** style questions about AI tools and pricing in the same breath: the rules of the software economy are shifting under your feet, and nobody sent a memo. Is the subscription era over? Not yet, but it is cracking. The per-seat model made sense when humans did the work and software was the tool. When [AI agents](/blog/ai-safety-showdown-what-the-doomsday-debate-means-for-everyday-users) do the work, seats become a terrible proxy for value. Vendors know it, buyers are starting to notice, and the contracts landing on desks in 2026 reflect the confusion. ## Why per-seat pricing is coming apart The logic is simple. A seat license assumes output scales with headcount. AI agents break that link. One analyst with an agent can process what used to take five people. The vendor's revenue falls exactly as the customer gets more value. Nobody designed a pricing model to punish efficiency, but here we are. We have watched three things happen across US, UK and European buyers this year: - **Seat counts shrink** as teams consolidate around agents, so vendors raise per-seat prices to protect revenue. - **Usage caps appear** in fine print, turning "unlimited" plans into metered plans with a friendly name. - **Outcome clauses creep in**, where vendors tie part of the fee to tickets resolved, leads qualified, or hours saved. That third one is the interesting shift. Outcome-based pricing SaaS is no longer a conference talking point. It is showing up in real contracts, especially in support, sales ops, and finance tooling. ## The named options on the table If you are renegotiating a SaaS contract in the next two quarters, you are choosing between four models. Here is what each actually costs and where it bites. | Model | How it bills | Typical 2026 range (USD) | Where it hurts | |---|---|---|---| | Per seat (legacy) | Fixed fee per named user | $20 to $120 per user/month | You pay more as you get more efficient | | Usage based | Per API call, token, or task | $0.002 to $0.05 per unit | Bills spike when agents run wild | | Hybrid (seat + usage floor) | Base seats plus metered overage | $15 to $60 per seat plus overage | Overage clauses are where budgets die | | Outcome based | Fee per resolved ticket, qualified lead, or verified saving | $1.50 to $12 per outcome, or 10% to 25% of documented savings | Disputes over what counts as an outcome | The headline numbers are less important than the definitions. A $4 per resolved ticket sounds cheap until you read the clause that says a "resolution" includes any ticket the agent touched, even the ones a human had to fix afterwards. ## A worked example with real numbers Take a 60-person customer support org in the Netherlands running an AI agent across email and chat. In 2025 they paid $45 per seat per month for 60 seats: $32,400 a year. The agent handled 40% of volume, and the team dropped to 45 seats. The vendor's new quote: $58 per seat for 45 seats, or $31,320 a year. Barely a saving, despite doing far less human work. The outcome-based alternative quoted $3.20 per resolved ticket. Last year the team closed 96,000 tickets. At that rate the bill would be $307,200. Absurd. But the vendor's definition of "resolved" excluded escalations and reopenings, and the real eligible count was 41,000. That is $131,200. Still worse than seats. The model that actually worked was a hybrid: $18 per seat for 45 seats ($9,720 a year) plus $1.10 per eligible resolution above a 30,000 baseline. At 41,000 resolutions, that added $12,100. Total: $21,820. A 33% saving versus the legacy renewal, with a shared incentive to keep quality high. The lesson: outcome pricing only works when the baseline is fixed and the definition is tight. ## What buyers should demand before signing We have read enough 2026 contracts to know the traps. Before you sign anything with an outcome or usage component, get these in writing: 1. **A written definition of the outcome.** Not "resolved." Specify what counts, what is excluded, and who adjudicates disputes. 2. **A baseline and a cap.** Fix last year's volume as the baseline and cap annual overage at a percentage of base fee (we suggest 15%). 3. **Audit rights.** You need the right to see the vendor's counting logic, not just their invoice. 4. **A quality floor.** Tie a share of the fee to accuracy or CSAT, otherwise the vendor profits from volume, not results. 5. **Exit terms.** If the model stops working, you need a clean way out without losing your data. If a vendor refuses all five, that tells you something. Walk. ## Our take We would not sign a pure outcome-based contract in 2026 unless the vendor is small and hungry enough to accept tight definitions. The accounting is too fuzzy and the disputes too expensive. Instead: - **Push for hybrid pricing** with a low seat floor and a capped usage component. It is the only model where both sides win as efficiency improves. - **For tools under $500 a month**, stay on seats. The admin cost of outcome tracking exceeds any saving. - **For anything above $50,000 a year**, insist on outcome clauses with a baseline, a cap, and audit rights. That is where the money is. - **Benchmark against alternatives** every renewal. Vendors are pricing in the assumption that switching is painful. Prove them wrong once and watch the quote change. On the VPN question that keeps landing in our inbox: no, VPNs are not banned in the UK as of October 2026. The Online Safety Act and the Investigatory Powers Act have tightened rules around age verification and [data retention](/blog/ai-in-public-spaces-the-flock-cameras-controversy-and-what-it-means-for-your-pri), and some consumer VPN marketing has been pulled for misleading claims, but using a VPN remains legal. If a SaaS vendor tells you a VPN ban justifies a pricing change, they are bluffing. Ask for the clause. ## What to do this week Pull your three largest SaaS renewals and find the seat count versus actual usage. If seats are shrinking while the bill grows, you have leverage. Ask the vendor for a hybrid quote with a capped usage component and a written outcome definition. Do it before the renewal window opens, not after. The vendors who move first on pricing models are the ones who keep customers through the transition. The ones who defend seats to the last invoice will lose them. ## FAQ **Is the subscription model dead for SaaS?** No. Seats still work for tools where humans drive the output. The model is breaking specifically where AI agents do measurable work, because seat counts no longer track value. **What is outcome-based pricing in SaaS?** You pay per verified result (a resolved ticket, a qualified lead, a documented saving) instead of per user. It works only when the outcome is tightly defined, baselined, and capped. **Are VPNs banned in the UK in 2026?** No. VPN use remains legal in the UK. Recent legislation tightened data rules and age verification, but a ban is not in force. Treat any vendor claim otherwise as a red flag.

Are VPNs Banned in the UK? AI Agents Reshape SaaS Pricing

Frequently asked questions

Related articles