Innovation & Tech

Raiffeisen Cuts Token Use 80% as Appian Reworks Its AI Agent Play

Appian expanded its AI-agent control plane and launched Ask Appian in London on November 17, 2026, after Raiffeisen reported an 80% cut in tokens and processing time.

By Grace Kim

4 min read

Appian Expands AI Agent Controls and Unveils Ask Appian
Appian Expands AI Agent Controls and Unveils Ask Appianniekverlaan / pixabay

What's News

  • Raiffeisen Bank International reported an 80% reduction in token consumption and processing time testing a hybrid agent-and-rules approach.
  • Appian unveiled Ask Appian and an expanded agent control plane at Appian Europe in London on November 17, 2026.
  • A study by Appian and Harvard Business Review Analytic Services found 92% of respondents say agents need guardrails; only 48% have them.
  • Cross-platform tracing, an A2A protocol and third-party agent registry are planned but not yet available.

By Grace Kim

Raiffeisen Bank International cut token consumption and processing time by 80% in testing by moving predictable work away from AI agents and into deterministic rules. The bank reported the figure on November 17, 2026, at Appian Europe in London, where Appian expanded its AI-agent control plane and unveiled Ask Appian, a plain-language interface for requesting work.

The 80% number is the story's hardest evidence that hybrid agent-and-rules designs can beat agent-only automation on cost. Claus Mitterlehner, Raiffeisen's head of smart automation, said his team initially gave most tasks to AI agents. "After testing it, the team changed course and moved the predictable steps to deterministic automation, keeping agents for the steps that need judgment," he said.

Raiffeisen intends to use the setup to automate payment inquiries.

What does the control plane actually do?

The control plane registers Appian and third-party agents in a single management view. It monitors activity and cost, restricts which models agents can use, tracks policy compliance and generates alerts. An optimization layer runs evaluations to track accuracy and connects agents to the specific cases and process steps they handled. Business KPIs — cost per case, cycle time — supply the operating context.

Appian's core design choice: agents live inside deterministic business processes. The process controls the sequence, and each step goes to an agent, a person, a system or a rule. Rules handle predictable work. Agents get guardrails and shared context for steps requiring judgment. If an agent fails or gets stopped, the task reroutes to a person, rule or another agent, so the case continues.

Sanat Joshi, Appian's executive vice president of product management, argues that permissions, guardrails and model access cannot establish whether an agent is doing useful work. "Appian puts business process at the center," he said, describing how each agent affects business outcomes.

A November 16, 2026 Appian blog post lays out the trade-off. Process-led steps are consistent, repeatable and auditable. Agent-led steps handle open-ended work but add cost and latency, with outcomes that vary by case. In one worked example, an agent assessing a loan hands debt calculations to deterministic formulas while a loan officer makes the final call.

The takeaway in plain words: Appian treats agents as replaceable steps in an auditable process, not autonomous actors.

What is Ask Appian?

Ask Appian is a separate interface where employees request work in plain language. It pulls relevant context from systems connected through Appian's data fabric and dispatches agents to finish the task. The stated goal: spare employees from hopping between underlying applications to assemble information and complete a request.

What's shipping now versus what's promised?

Appian's documentation draws a sharp line between current functions and later additions.

Available today:

  • Third-party agents as steps in Appian processes, via MCP or direct integration, without moving into Appian
  • Measurement of Appian agents against process KPIs
  • Controls, orchestration and process intelligence, plus an early-adopter program ahead of general availability

Planned, per the roadmap:

  • Support for the A2A protocol
  • A registry covering third-party agents
  • Tracing for agents running outside Appian
  • Estimated cost per case and per step for Appian and third-party agents
  • Before-and-after comparisons with manual work and agent-improvement recommendations

Appian said additional capabilities would start arriving during 2026. The company itself cautions that planned features and timing can change, and purchasing decisions should rest on generally available functionality. That caveat matters most for buyers eyeing the cross-platform pieces.

Does the evidence back the pitch?

A study by Appian and Harvard Business Review Analytic Services, which surveyed AI decision-makers on modernization and workflow integration, found 92% of respondents said agents need rules-based guardrails to be safe and effective. Only 48% had such rules in place.

That gap — 92% who want guardrails, 48% who have them — is the market Appian is chasing. The control plane's optimization loop aims to close it: evaluations and process KPIs identify agent performance, and repeated agent steps can become deterministic workflows or business rules, concentrating agent spending on judgment-based work.

There's a catch, though. The control plane still lacks cross-platform tracing and measurement for agents built outside Appian's platform. Until the planned registry, tracing and cost-per-step metrics for third-party agents ship, the Raiffeisen result stays provable mainly inside Appian's own walls — and that boundary will determine how far the 80% story travels.

Source: unite.ai

Share this article:

More from Grace Kim

Grace Kim

Show full bio

Market editor covering people, launches and day-to-day practice in entrepreneurship, startups and venture funding at Business Bearings.

2,206 articles

Related articles

« Previous articleNext article »