Service-as-software is the business model in which AI does the work itself — resolving the support ticket, collecting the debt, producing the research — and charges for the completed outcome rather than for access to a tool. It is replacing the per-seat SaaS model that defined enterprise software for twenty years, and IDC now predicts pure seat-based pricing will be effectively obsolete by 2028 (1). For African founders, this is not a disruption to fear but a convergence to exploit: the global software industry is adopting the pay-for-results logic that mobile money economies were forced to invent years ago.
Key Takeaways
- The per-seat SaaS model is structurally breaking. IDC predicts that by 2028, 70% of software vendors will refactor pricing around consumption, outcomes, or capability rather than seats (1); Bain finds roughly 65% of major SaaS vendors have already layered AI consumption meters onto seat pricing (2).
- AI-native challengers price the work, not the login. Decagon charges per conversation handled; Intercom’s Fin charges $0.99 per resolved ticket — pricing models incumbents cannot copy without cannibalizing their own seat revenue (3)(4).
- In February 2026, markets repriced the fear: roughly $285 billion in value was wiped from global SaaS and IT-services stocks in a single sell-off as investors digested agentic AI’s threat to seat economics (5).
- Africa never fully adopted per-seat SaaS. Low card penetration, prepaid culture, and tight cash flow forced usage-based and outcome-based models — from PAYG solar to per-transaction mobile money fees — making outcome pricing the regional default, not a pivot (6)(7).
- The opportunity is asymmetric: East African founders fluent in selling outcomes through mobile money rails can build service-as-software businesses for markets the global vendors will not localize for — if they move before the window closes.
What Actually Changed: From Tools That Assist to Agents That Do
For two decades, the SaaS contract was simple. The vendor sold a tool; the customer supplied the labor. Salesforce did not close deals — salespeople did, inside Salesforce. Zendesk did not resolve tickets — support agents did, inside Zendesk. The pricing unit, the “seat,” was a proxy for the human doing the work. More humans, more seats, more revenue. It was, as a16z’s enterprise team put it, the golden rule of software economics (3).
Agentic AI breaks the proxy. When an AI agent resolves the ticket end-to-end, there is no human in a seat to count. The natural pricing unit becomes the unit of work: the conversation handled, the invoice collected, the report produced. a16z’s analysis of this shift names the pattern precisely — Decagon prices customer support per conversation handled and is moving toward pricing per resolution achieved, “fundamentally a better alignment of incentives between vendor and buyer” (3). Intercom’s Fin agent charges $0.99 per resolved conversation, and competitive pressure pushed HubSpot’s equivalent to $0.50 in April 2026 (4). The product is no longer software-as-a-service. It is service-as-software: human work, delivered by code, priced like labor.
The strategic insight underneath is counterpositioning. An incumbent like Zendesk cannot match Decagon’s per-resolution pricing without destroying its own seat-based revenue — the same trap that kept Blockbuster from matching Netflix and PeopleSoft from matching Workday (3). Bain’s technology research confirms the squeeze: generative and agentic AI are automating the very workflows SaaS tools were built to host, and leaders are being told to shift pricing from seats to outcomes and to train sales teams to sell business results, not features (2). IDC frames the endpoint bluntly: by 2028, pure seat-based pricing is obsolete, with 70% of vendors refactoring around new value metrics (1).
Markets have already voted. The February 2026 sell-off — traders called it the “SaaSpocalypse” — erased roughly $285 billion in value from SaaS and IT-services companies in a single day as the repricing fear went mainstream (5). Whether that was overreaction or early recognition, the direction of travel is no longer seriously contested. The debate has moved to a second-order question: if AI eats the application layer, where do moats go? The emerging consensus — visible in a16z’s own counter-argument that AI eating application software is good news for builders — is that defensibility migrates from the software itself to proprietary data, workflow ownership, and trusted distribution (8).
Why Does the Service-as-Software Shift Favor African Founders?
Here is the part the Silicon Valley discourse misses: Africa never lived under the per-seat regime it is now watching collapse.
Per-seat SaaS assumed three things that East African markets never supplied: corporate credit cards for frictionless recurring billing, IT budgets sized to pay for capacity whether or not it was used, and a labor model in which white-collar workers sat in licensed seats. None of these held in Kampala, Nairobi, or Dar es Salaam. Card penetration stayed low while mobile money became the rail — Africa’s mobile payments market reached an estimated $75 billion in 2025 and is growing at nearly 40% annually (6). Budgets were tight and lumpy, which made paying for idle capacity irrational. So the models that actually scaled on the continent were usage-based and outcome-based from birth: prepaid airtime, per-transaction mobile money fees, and pay-as-you-go solar that meters energy by the day and repossesses the asset remotely if payments stop (7).
In other words, East African commerce has run on “pay for what you get” for fifteen years. The global software industry is now converging — under AI pressure — on the pricing logic African constraint already invented. That is a rare and valuable asymmetry. The hardest part of outcome-based selling is not the technology; it is the commercial muscle: defining the outcome, measuring it credibly, pricing it so both sides win, and collecting in small increments from cash-flow-constrained buyers. East African founders have been doing all four since before “agentic” was a word.
This connects directly to how agentic payments will work in the region. The infrastructure question — can an AI agent hold a mobile money float, initiate a payment, settle per outcome? — is the subject of active building, and the founders wiring AI agents into M-PESA and mobile money rails are constructing the settlement layer service-as-software needs in African markets. Outcome pricing only works if the outcome can trigger the payment.
What Happens to African Service Businesses When AI Does the Work?
Most East African “tech” businesses are, candidly, service businesses with software wrappers: agencies, brokers, recruiters, accountants, customs clearers, lead-generation shops. The service-as-software shift cuts through this segment in two directions at once.
The threat: any service priced on effort rather than outcome is now exposed. If a global AI agent can produce the market research report, reconcile the books, or qualify the leads at near-zero marginal cost, the African firm selling hours of junior labor competes against software priced like labor but costed like electricity. The Mastercard Foundation’s research on the adjacent BPO sector — 40% of tasks automatable by 2030, entry-level roles most exposed — is the early warning for the whole effort-priced economy (9).
The opportunity: the same agents, in local hands, turn small service firms into scalable outcome factories. A three-person Kampala accounting practice that deploys agents for data entry, reconciliation, and draft reporting stops selling staff time and starts selling “books closed by the 5th, guaranteed” — an outcome it can price as a product and deliver at a marginal cost incumbents with payroll cannot match. The playbook for treating AI agents as your first five employees is the operational core of this move: the firm keeps the client relationship, the local trust, and the judgment; the agents absorb the toil.
The decisive variable is who owns the customer and the data. a16z’s moat analysis points here, and it lands with extra force in Africa: the defensible assets in an agentic world are proprietary data and workflow ownership (3)(8). African SMEs generate data that no foreign model has — payment behavior on mobile money, informal-sector credit performance, local-language customer conversations, supply-chain reality on murram roads. The founder who captures that data while delivering outcomes builds a moat a global agent vendor cannot scrape.
The Three-Meter Ladder: A Framework for Pricing in the Agentic Era
To make the shift operational, I use a framework I call the Three-Meter Ladder. Every software or tech-enabled service business meters its value one of three ways, and the agentic era is forcing everyone up the ladder.
Meter 1 — Access (the seat). You charge for the right to use the tool. Value capture is decoupled from value delivery: the customer pays whether or not work gets done. This is the meter IDC says dies by 2028 (1). If your business charges East African SMEs monthly fees for tool access they barely use, you are running a model that is now structurally obsolete on two continents.
Meter 2 — Activity (the usage unit). You charge per API call, per message, per transaction. Better alignment — this is where mobile money and PAYG solar live (6)(7) — but the customer still bears outcome risk: they pay for activity even when it fails to produce results. Bain’s finding that 65% of SaaS vendors have bolted consumption meters onto seats shows the industry crowding onto this rung as a halfway house (2).
Meter 3 — Achievement (the outcome). You charge when the work is done: the ticket resolved, the debt collected, the qualified meeting booked. Decagon and Intercom’s Fin operate here (3)(4). This meter demands three capabilities: a definition both sides accept, a measurement both sides trust, and a margin model that survives paying for compute on failed attempts. It is the hardest meter to run and the most defensible, because switching away from a vendor who reliably delivers outcomes means re-accepting risk the customer already offloaded.
The ladder yields three rules for East African builders. First, never enter at Meter 1 — you would be adopting the dying model with none of the incumbency that makes it temporarily survivable. Second, use Meter 2 as your on-ramp, not your destination: per-use pricing matches mobile money rails and SME cash flow, but plan the climb to outcomes from day one. Third, price Meter 3 against the labor it replaces, not the software it resembles — an agent that does the work of a UGX 1.5 million-per-month employee is mispriced at UGX 50,000 per month, and underpricing outcomes is the most common way founders donate their margin. Pricing remains the most neglected lever in African SME strategy, and the agentic shift raises the cost of neglecting it.
How Should East African Software Builders Respond — Build, Wrap, or Wait?
For the region’s developers and technical founders, the shift rewrites the build decision.
Building thin SaaS is over. A dashboard that organizes work humans still do, sold per seat to African SMEs that never had seat budgets, was always a hard sell. Now it is a hard sell into a dying category.
Wrapping global agents in local outcomes is the near-term play. The models are commoditizing; the scarce assets are local: the outcome definition that fits a Ugandan distributor, the mobile money settlement integration, the Luganda-language customer conversation, the trust to be allowed into a firm’s books at all. A founder who wraps frontier models in those assets is not reselling AI — she is selling completed work into a market global vendors cannot reach. The caution is real, though: when anyone can generate the software, the software is not the moat, and the judgment to build the right thing — and maintain it — carries the premium.
The deeper build is the data flywheel. Every outcome delivered generates proprietary performance data — what worked, for whom, at what price. That data improves the agent, which wins more outcomes, which generates more data. Service-as-software businesses are data businesses wearing work clothes. The evidence on whether AI actually pays rent for SMEs suggests returns concentrate where deployment is disciplined and measured — which is exactly the discipline outcome pricing forces, because you only get paid when it works.
The honest constraint: outcome pricing transfers risk to the vendor. You absorb compute costs on failures; you carry the measurement disputes; you need working capital to survive the gap between doing the work and proving it. For capital-constrained founders, hybrid structures — a small platform fee plus per-outcome charges, mirroring how mobile money itself blends float and fees — are the pragmatic bridge.
What Does This Mean for the Next Five Years?
Three predictions follow from the logic above.
First, the African service sector will bifurcate. Firms that wield agents will sell outcomes at software margins; firms that sell effort will compete with software on price and lose. The dividing line will not be capital — agent tooling is cheap — but operating discipline and willingness to re-price.
Second, outcome-priced exports become viable. A Kampala firm selling “qualified B2B meetings in East Africa, pay per meeting held” or “GTM research on African markets, priced per validated insight” can sell globally on Meter 3 terms, because outcome pricing strips away the buyer’s risk of hiring an unknown foreign vendor. Service-as-software is, quietly, a trade-enablement technology for African knowledge work.
Third, the pricing knowledge will flow south-to-north. The mechanics of collecting small, frequent, outcome-linked payments from millions of low-balance accounts — the operational core of agentic commerce — were industrialized in Nairobi and Kampala, not San Francisco. Expect the case studies of the agentic pricing era to cite African precedents, and expect the founders who lived those precedents to have an edge building what comes next.
The seat is dying. The work remains. The founders who learned to sell work — because their markets never let them sell anything else — are better prepared for this transition than they have been told.
Frequently Asked Questions
What is service-as-software?
Service-as-software is a business model in which AI agents perform complete units of work — resolving support tickets, collecting payments, producing research — and the vendor charges per outcome delivered rather than per user seat. It inverts SaaS: instead of selling tools that assist human labor, it sells the finished labor itself (1)(3).
Is per-seat SaaS pricing really dead?
Not instantly, but it is structurally declining. IDC predicts pure seat-based pricing will be obsolete by 2028, with 70% of vendors refactoring toward consumption or outcome metrics (1). Bain finds about 65% of major SaaS vendors have already added AI consumption meters, signaling the industry-wide transition is underway (2).
Why is Africa well positioned for outcome-based pricing?
African markets skipped the per-seat era. Low card penetration and tight SME budgets forced usage- and outcome-based models years ago — prepaid airtime, per-transaction mobile money, pay-as-you-go solar. The commercial skills outcome pricing requires — defining, measuring, and collecting on results — are already regional defaults (6)(7).
How should an African SME service firm respond to AI agents?
Re-price before you are re-priced. Deploy agents on repetitive delivery work, keep humans on judgment and relationships, and shift contracts from effort (hours, retainers) to outcomes (books closed, leads qualified). Firms that keep selling effort will compete directly with software on price — a losing position (2)(9).
What is the biggest risk in outcome-based pricing?
Risk transfer. The vendor absorbs costs when the AI fails, carries measurement disputes, and finances the gap between work done and outcome proven. Founders should start with hybrid pricing — a small base fee plus per-outcome charges — and invest early in measurement both sides trust (3)(4).
Related Reading
- AI Agents as Your First Five Employees
- Agentic Commerce on M-PESA and Mobile Money Rails
- Vibe Coding and the Judgment Premium
- Does AI Pay Rent? The SME Evidence
Sources and Evidence
- IDC, “Is SaaS Dead? Rethinking the Future of Software in the Age of AI” — global technology research firm; source of the 2028 seat-pricing obsolescence prediction and the 70% vendor-refactoring forecast. https://www.idc.com/resource-center/blog/is-saas-dead-rethinking-the-future-of-software-in-the-age-of-ai/
- Bain & Company, “Will Agentic AI Disrupt SaaS?” (Technology Report 2025) — top-tier strategy consultancy; vendor-survey finding that ~65% of major SaaS companies have layered consumption meters onto seat pricing. https://www.bain.com/insights/will-agentic-ai-disrupt-saas/
- Andreessen Horowitz, “AI Is Driving a Shift Towards Outcome-Based Pricing” (Enterprise Newsletter) — leading venture firm with direct portfolio visibility; source for Decagon’s per-conversation pricing and the incumbent counterpositioning argument. https://a16z.com/newsletter/december-2024-enterprise-newsletter-ai-is-driving-a-shift-towards-outcome-based-pricing/
- Getmonetizely, “The 2026 Guide to SaaS, AI, and Agentic Pricing Models” — pricing-strategy publication; documents Intercom’s $0.99 per-resolution pricing and HubSpot’s April 2026 move to $0.50, plus hybrid-pricing adoption at 41%. https://www.getmonetizely.com/blogs/the-2026-guide-to-saas-ai-and-agentic-pricing-models
- Outlook Business (xHub), “The SaaSpocalypse of 2026: How Agentic AI Killed Per-Seat SaaS” — business media; reports the February 2026 sell-off of roughly $285 billion in SaaS and IT-services market value. Market-event reporting; figures are estimates. https://www.outlookindia.com/xhub/blockchain-insights/the-saaspocalypse-of-2026-how-agentic-ai-killed-per-seat-saas
- Market Data Forecast, “Africa Mobile Payments Market” — market research; Africa mobile payments at ~$75.2 billion (2025) growing ~39% CAGR. Industry-research estimates. https://www.marketdataforecast.com/market-reports/africa-mobile-payments-market
- McKinsey & Company, “The Future of Payments in Africa” — global consultancy; context on mobile-money rails, low card penetration, and usage-based commerce models across the continent. https://www.mckinsey.com/industries/financial-services/our-insights/the-future-of-payments-in-africa
- Andreessen Horowitz, “Good News: AI Will Eat Application Software” — the optimist’s case that application-layer disruption expands the market and shifts moats to data and workflow ownership. https://a16z.com/good-news-ai-will-eat-application-software/
- Mastercard Foundation (with Caribou and Genesis Analytics), “40% of Tasks in Africa’s Growing Tech Outsourcing Sector May Be Affected by AI by 2030” — foundation-commissioned research; automation exposure of effort-priced African service work. https://mastercardfdn.org/en/news/40percent-of-tasks-in-africas-growing-tech-outsourcing-sector-may-be-affected-by-ai-by-2030/