Software as a Service is broken. Here’s what comes next.

The Promise vs. the Reality

SaaS was a revolution 20 years ago. No argument from me. Move from on-prem to subscription, get out of the server-patching business, log in, get value. For its first decade it delivered exactly that.

Then it degraded. The promise — subscribe, log in, get value — quietly became: subscribe, scope a six-month implementation, hire a consultant, appoint an internal champion, run a steering committee, hope for adoption. By the time the platform “goes live,” the budget and the patience are both gone.

The numbers are not a side issue. The average company runs 106 SaaS applications today — down from a 2022 peak of 130, but large enterprises still average 131 apps each, per BetterCloud’s 2025 State of SaaS. Gartner has long pegged unused SaaS spend at over 30% of the bill, and Gartner has separately warned that organisations failing to centrally manage SaaS lifecycles will overspend by at least 25% on unused entitlements and overlapping tools alone. Some applications have nobody who remembers why they were bought.

For asset managers — lean firms with no dedicated IT team, where one person on the operations desk is also the data engineer, the project manager, and the compliance lead — the gap between “access to software” and “outcome delivered” is where millions of dollars and entire quarters disappear.

Why SaaS Fails Asset Managers, Specifically

Three failure modes show up at every asset manager I’ve worked with, regardless of size.

Integration burden. Every tool has to talk to your custodian, your fund administrator, your registry, your CRM, and your market data provider. That integration work doesn’t end at go-live. APIs change. A custodian gets acquired. A new platform shows up on your APL. Someone at your firm — usually someone who already has three other jobs — has to manage the plumbing in perpetuity.

Configuration gap. Generic defaults don’t fit a fund manager. Compliance rules are mandate-specific. Reporting is investor-specific. Distribution workflows are platform-specific. Closing the gap between the demo and your reality requires a project manager, a consultant, and six months. Vendors quote “go live in 4 weeks” but mean access in 4 weeks, value in 6 months. If ever.

Maintenance tax. Updates break workflows. Staff turnover means the configuration knowledge walks out the door with the person who built it. Data quality degrades because nobody owns the input layer. You’re paying full price for a tool running at 40% of its capability, and once that’s the steady state nobody can justify the project to fix it — and nobody can justify the project to replace it either.

What Service as Software Means

Invert the model. Don’t sell the tool. Deliver the outcome.

• You don’t buy reporting software. You get reports.
• You don’t buy a compliance platform. You are compliant.
• You don’t buy analytics. You get insights.
• You don’t buy a distribution CRM. You get a distribution strategy executed.

How this works at Datafabric:

We own the integrations. Custodian, fund admin, CRM, registry, market data — connected, harmonised and maintained by us inside The Foundry. APIs change, we handle it. You never touch a data pipeline.

We own the configuration. Compliance rules, reporting templates, distribution workflows — set up for your firm, not a generic default. Mandates change, we update. No Jira board on your end. No project manager.

We own the outcomes. Our success metric is your success metric. Did reports ship on time? Did breaches get caught? Did the distribution team hit targets? If not, that’s our problem to fix. Not yours.

One Person, Not a Village — an Embedded Engineer

SaaS requires an ecosystem to deliver value: a vendor account manager, an implementation consultant, an internal champion, a data engineer, and a project manager. Five people to make one tool work. For a 30-person asset manager, that ecosystem is the cost.

Service as Software inverts that. We embed an engineer in your team.

That engineer sits in your stand-ups when it makes sense. They learn your mandates, your reporting cadence, your custodian quirks. They’re on first-name terms with your operations lead and your CFO. Functionally, they feel like an internal hire — the kind of senior platform engineer a $5B asset manager would never have on payroll on its own.

But they don’t report to you. They report to Datafabric. And that distinction is the whole point.

Because they’re a Datafabric engineer, they have direct, hands-on access to the platform’s core engine and services. When you need a new custodian feed, a new compliance rule, a new report template, or a new Sherpa workflow, that engineer ships it themselves — into the same platform that runs every other client. There’s no ticket queue, no cross-vendor handoff, no “we’ll raise it as a feature request and see where it lands on the roadmap.”

This is the model leading service-as-software firms are converging on (Sierra reportedly embeds engineers inside Fortune 500 support teams for weeks at a time, per Foundation Capital’s analysis). It’s the only way the economics work. You get a single point of contact who is both deeply on your side and structurally able to change the product itself.

The downstream consequences:

• No internal champion dependency. When they leave, nothing breaks — the knowledge sits with us.
• No consultant fees to “drive adoption”. Adoption is our job, not yours.
• No cross-functional steering committee to get a report template changed. The engineer who knows the request is the engineer who ships the change.

This is not a nice-to-have for a lean firm. It is the difference between technology that gets used and technology that sits on a shelf with a renewal date.

Built to Last, Tailored from Day One

SaaS is generic by design. Same product for every client, configured at the edges. When your needs outgrow the configuration options — and they always do — you hit a wall. Feature request submitted. Roadmap reviewed. Maybe next quarter. Maybe never.

Service as Software is tailored from day one:

And when needs change — a new fund launch, a new mandate, a custodian switch — the platform adapts in days, not quarters. No feature request queue. No roadmap politics.

Fast Shipping, No Excuses

AI changes the economics of delivery. What used to take a team of consultants six months, we deliver in weeks — not because we’re cutting corners, but because the underlying cost of integration, configuration and bespoke build has collapsed.

• New integrations built and deployed in days, not months.
• Configuration changes shipped same week.
• Reporting templates created in hours, not sprint cycles.
• The platform improves continuously, not in quarterly releases.

This is not a future promise. Clients go live in four weeks. Changes ship within days of being asked for. There is no backlog sitting between your need and the solution.

Future-Ready by Design

SaaS tools are snapshots of what the vendor thought you needed when they built them. The world moves — AI capability evolves, regulations tighten, new data sources emerge, custodians get acquired — and the tool stays frozen until the next major release.

Service as Software evolves with you:

• AI capabilities improve → your platform improves automatically.
• New regulations land → compliance rules updated within days.
• New LLMs outperform old ones → Sherpa routes to the best model for the task. No migration. No retraining your team.
• Market shifts → analytics and distribution intelligence adapt in real time.

You’re not locked into a 2022-era tool paying 2026 prices. You’re on a platform that compounds.

The End Goal: SaaS Dies. Tailored AI Becomes the Default.

Step back from the day-to-day mechanics and look at where the market is actually going. This is not a Datafabric thesis — it’s the consensus view from the analysts and venture investors who price the software industry.

Foundation Capital frames it as a $4.6 trillion services-as-software opportunity: AI doesn’t just compete with SaaS, it competes with the labour markets SaaS was built to support. Software was never the goal — the work was. SaaS was a useful waypoint. AI lets you sell the work directly. McKinsey, Forrester, AlixPartners and Gartner are all describing variations of the same shift.

Gartner’s numbers tell the same story from the inside of the software industry:

Translate that for a COO of an asset manager.

The SaaS tools you’re integrating against today are themselves trying to bolt agentic AI onto architectures designed for the click-and-configure era. Most of them won’t make the transition cleanly. Gartner’s forecast that more than 40% of agentic AI projects will be cancelled by 2027 is not a story about AI failing — it’s a story about integration, data and configuration foundations failing underneath AI. The same three failure modes that broke SaaS, on a faster clock.

So the end state isn’t SaaS plus a chatbot. It isn’t even “AI-enhanced SaaS.” The end state is that deeply integrated, mandate-specific, AI-native operations — the thing that costs $5M and 18 months to bespoke-build today — becomes a commodity. Delivered as a service. Priced against outcomes. Maintained by the vendor.

That’s what we’re building toward. It’s also why Service as Software isn’t a marketing position — it’s an operating model designed to survive the transition. The firms that win the next decade won’t be the ones who licensed the most tools. They’ll be the ones whose data, configuration and outcomes are owned by a partner who ships every week, not every quarter.

An honest counterpoint, from McKinsey: SaaS will not disappear cleanly. Multiple software models will coexist for years — some vendors will graft agents on top of existing products, some will rebuild, some will simply be replaced. The transition is messy. But the directional arrow on the chart is the same. Outcomes > access.

What to Ask Your Next Vendor

Four questions. They’re short. They’re unambiguous. The answers will tell you which model you’re actually buying.

If even one of those answers gives you pause, the model isn’t fit for a lean asset manager. Not in 2026. Not with the speed at which AI, regulation and adviser-flow dynamics are moving.