Incident sheet · /enterprise-ai

Filed
Read
5 min · 926 words
Severity
Routine

AI Is Turning Every Enterprise Process Into Legacy — Camunda

Camunda says 79% of firms bolt AI onto existing processes rather than redesign them — and cites Danica and Angle Finance cases where end-to-end re-engineering delivered dramatic results.

The Great Process Re-Engineering: Is Every Process Now Legacy?
FeedThe Great Process Re-Engineering: Is Every Process Now Legacy? — geralt / pixabay

Event log

  1. 79% of organisations integrating AI into existing processes do so to avoid redesigning the process itself, per Camunda's 2026 AI Process Gap research across 1,000 senior decision-makers.
  2. Danica cut customer onboarding from two to three months to a matter of days using Camunda ProcessOS after a projected 20% AI improvement proved insufficient.
  3. Angle Finance reduced credit decisioning approval from several days to two hours and cut its cost-to-income ratio by 24%.
  4. Average enterprise processes span more than 50 endpoints, per Camunda research.

Danica, the Danish pension and insurance provider owned by Danske Bank Group, cut customer onboarding from two to three months down to a matter of days — not by adding AI to individual steps, but by redesigning the process end to end. That case sits at the centre of what Camunda calls The Great Process Re-Engineering: the argument that AI now renders most existing enterprise processes legacy, regardless of their age.

Camunda's regional vice-president APAC, Jared McMahon, frames the problem in plain terms. "Legacy is not necessarily about age," he told iTnews. "It's about who the process was actually designed for."

For decades, enterprises built processes around a simple pattern: a person applies judgement, completes a task, and passes the work to the next person. AI, McMahon argues, breaks that assumption — and inserting models into workflows designed for handoffs between humans leaves most of the value on the table.

The takeaway: a process becomes legacy when its design assumes human judgement at every step, and AI has just made that assumption obsolete.

Why do organisations bolt AI onto broken processes?

Camunda's 2026 AI Process Gap research, conducted by Sapio Research among 1,000 senior IT, operations and transformation decision-makers at large organisations in the US and Europe, found 79 per cent of organisations starting to integrate AI into existing processes do so because it is an easier internal conversation than redesigning the process itself.

That preference is understandable. Process re-engineering historically meant lengthy projects: stakeholder consultation, workshops, design phases, approvals. McMahon argues AI changes what is possible, compressing those timeframes and letting organisations reconsider how work gets done rather than where technology fits into the current design.

The takeaway: most enterprises choose AI insertion over redesign because redesign is politically and operationally harder — not because it delivers better results.

When does a 20 per cent improvement fail the business?

Danica's numbers show the gap between the two approaches. The insurer first explored introducing AI into individual onboarding steps that required significant judgement, and calculated it could improve the existing process by around 20 per cent.

"Twenty per cent on two to three months was really and realistically, for their customers, not moving the needle far enough," McMahon said.

Danica instead became an early beta customer of Camunda's ProcessOS. Working with the vendor, it mapped its desired business outcomes, existing process knowledge and bottlenecks, then rebuilt the onboarding process end to end using Camunda's intelligence layer. The result: onboarding dropped from two to three months to days.

The takeaway: trimming a fraction off a months-long process rarely changes the customer experience; redesign can collapse it entirely.

Does every step need an AI hammer?

No. "Not every nail requires an AI-based hammer," McMahon said. His rule of thumb:

  • Deterministic workflows still fit steps that are known in advance, repeatable and consistent.
  • AI suits work requiring greater judgement.
  • AI agents or humans step in when a process moves outside defined criteria.

The discipline matters as enterprises try to move AI projects past experimentation and escape what practitioners call "PoC prison." "For me, it's always about the business outcome, because if there isn't one, then why do it?" McMahon said. The starting point is identifying what will make the biggest difference to the organisation, and whether that calls for deterministic processing or agentic orchestration.

The takeaway: target AI where judgement is genuinely needed, and leave repeatable work to deterministic flows.

How do you build for models that change?

The AI model or agent an organisation chooses today may not be the one it wants tomorrow. Camunda's research puts the average enterprise process at more than 50 endpoints spanning numerous systems — a surface area that makes per-system rework expensive.

McMahon's answer is an orchestration layer between core systems and customer engagement. It provides visibility across the entire process while individual technologies change underneath. When an AI agent or underlying system is swapped, the organisation modifies the orchestration logic and points to a different target rather than redesigning the business process. "It's just about changing that logic and potentially pointing to a different target," he explained.

The takeaway: orchestration decouples process design from technology choice, so model churn does not force constant re-engineering.

Can you re-engineer without ripping and replacing?

Yes, and McMahon insists you should. "Re-engineering is not necessarily about throwing the baby out with the bathwater," he said.

Australian lender Angle Finance is his proof case. Formed through the acquisition of two companies, it inherited duplicated systems and siloed processes. Rather than rebuild, it deployed Camunda as an end-to-end orchestration layer, coordinating established process steps across existing systems while layering human judgement and AI onto more dynamic activities.

The results were concrete:

  • A priority credit decisioning process fell from several days to two hours, including during peak periods such as end of financial year.
  • The private equity-backed lender cut its cost-to-income ratio by 24 per cent.
  • Duplicated systems are being progressively retired without disrupting critical business operations.

"No one wants to have duplicated systems, but you also don't want to interrupt those processes that effectively run the business," McMahon said.

The takeaway: orchestration lets an enterprise modernise while it keeps extracting value from sunk technology investment.

For CIOs, the broader challenge is to look past incremental gains and ask what processes could achieve if redesigned for an AI-enabled world from the start. McMahon's closing instruction is short: "Think big to begin with."

via itnews.com.au (Source)

Filed under

  • enterprise-ai
  • process-re-engineering
  • process-orchestration
  • workflow-automation
  • ai-agents
Share this article:

More from Amara Osei

Amara Osei

Show full bio

Correspondent covering regulation, policy and their impact on enterprise information technology and the leaders who run at CIO Monitor.

188 articles

The newsroom digest

One email a day with the stories that matter. No noise.

By subscribing you agree to the privacy policy.

Cross-references

‹ Previous articleNext article ›