Case Study

An AI Knowledge Assistant for Customer Service Consultants

Softwise.AI implemented an AI-powered knowledge assistant at Santander Bank Polska - the third largest bank in Poland by assets and by number of branches - to support nearly 400 customer service consultants through a complex post-merger period. The platform unified knowledge from the bank's intranet, network drives, and external sources into a single natural-language search experience, helping consultants find accurate answers in seconds and reducing reliance on second-line support.

CLIENT

Santander Bank Polska

YEAR

2019

INDUSTIRES

Banking Customer Service

DELIVERABLES

AI-Powered Knowledge Search, Document Integration, Inquiry Analytics

The problem

In 2018, Santander Bank Polska completed a complex merger with Deutsche Bank, with Deutsche Bank's customers becoming Santander's customers on November 10. The merger created an immediate operational challenge for customer service: nearly 400 consultants suddenly had to support an expanded product range and a new customer base, with limited time to absorb the change.

The consequences were felt on both sides of the line. Customers experienced longer waiting times across chat, audio, video, email, and contact forms. Inside the bank, the need to search a broad and complex knowledge base had grown more pressing than ever - and the existing approach was not keeping up.

Santander turned to Softwise.AI for a way to give consultants fast, reliable answers from across their fragmented document landscape, without rebuilding the underlying systems or burdening teams with manual tagging and content preparation.

{Key Challenges}
Why a post-merger contact center couldn't scale knowledge access?

The contact center was contending with several pressures at once, all amplified by the merger:

Extended response times to customers
Longer service times across every channel were eroding customer experience at a critical moment.
Slow, expensive consultant onboarding
New hires needed roughly a month of preparatory training before taking calls, then worked under supervision for up to six months - a cost the bank could not absorb at scale.
High employee turnover
The customer service department experienced high churn, compounding the onboarding burden.
Large thematic specialization across teams
Knowledge was divided across specialized teams, making it hard for any single consultant to handle an unfamiliar topic.
High dependency on second-line support
A significant share of inquiries required escalation to more experienced consultants.
Risk of incorrect answers
Without easy access to consultation or verification, the risk of giving wrong information - and generating complaints - was elevated.

{Process}

How we deployed the knowledge assistant with zero manual content prep?

Softwise.AI implemented the assistant with a deliberately low-friction approach, prioritising rapid time-to-value over a long content-preparation phase.

1. Source integration

The platform was connected to the bank's external websites, intranet resources, and network drives, bringing fragmented knowledge under one searchable surface.

2. Automated document preprocessing

All document ingestion and preprocessing happened automatically. There was no need to manually rework documents, tag them, or build conversation scenarios - a major factor in how quickly consultants could begin using the system.

3. Natural-language search and source retrieval

For each consultant query, the system returned the relevant fragment of the document together with a link to the original - so answers were both fast and verifiable.

4. Inquiry analytics and continuous improvement

The platform analyzed incoming queries and surfaced patterns Santander could act on: the most popular topics, which consultant groups were asking most, periods of heavy traffic, and questions customers were asking that the existing knowledge base did not yet cover.

Solution

Softwise.AI was deployed as a unified knowledge platform sitting on top of Santander's existing document estate, with a single goal: let consultants ask in natural language and get the right piece of the right document, instantly.

One search across many sources

Consultants could query external websites, intranet resources, and network drives from one interface, removing the need to remember where each piece of information lived.

Zero-friction onboarding for content

Document preprocessing was automatic - no manual tagging, scripting, or scenario design - which meant the platform could go live across the team quickly and continue ingesting new content with minimal effort.

Answers backed by original documents

Every result combined a focused excerpt with a link to the original document, making it easy for consultants to verify and cite their source.

Built-in analytics layer

The system did not just answer questions - it analyzed them. Santander gained ongoing insight into popular topics, demand peaks, and content gaps, feeding a continuous-improvement loop for the knowledge base itself.

{Results}

97.6% correct answers and 7% shorter customer conversations

Results from the deployment, covering approximately 300 consultants:

97.6% correct answers after a month of working with the platform.
7% reduction in average conversation time with customers.
89.7% user satisfaction among consultants using the application.
Faster onboarding for new customer service employees.
Fewer complaints as the risk of incorrect answers fell.
Lower second-line dependency - a smaller share of calls required escalation to senior consultants.
8,827 queries sent to the assistant in the analyzed period, with intent recognition on 97.6% of them.

{Business impact}

Faster onboarding and fewer escalations during a critical merger

Better customer experience at a critical moment. Faster, more accurate responses landed exactly when customer trust mattered most.
Cheaper, faster onboarding. New consultants ramped up more quickly with a knowledge assistant at their side, reducing the cost of high turnover.
Reduced load on senior consultants. Fewer escalations meant experienced staff could focus on genuinely complex cases.
Stronger compliance posture. Source-linked answers and reduced error rates lowered the risk of complaints and incorrect information.
A data-driven knowledge base. Inquiry analytics gave the bank a structured way to identify and close content gaps continuously.

{Key recommendations}

5 rules for AI knowledge assistants in large service organizations

A few takeaways apply to any large service organization deploying AI knowledge assistants:

Automate document preprocessing. Manual tagging and scenario design kill momentum; AI ingestion that just works is what makes adoption realistic at scale.
Always return the source, not just the answer. In regulated and customer-facing settings, verifiability is what makes consultants trust and use the tool.
Measure intent recognition, not just usage. Tracking how well the system understands queries (97.6% in this case) is the cleanest signal of real quality.
Treat inquiry data as a knowledge-base feedback loop. Unanswered questions are the highest-value content backlog you have.
Deploy first to a critical moment of pressure. A merger, regulatory change, or product launch is precisely when an AI assistant proves its worth - and when adoption is easiest to win.

SUMMARY

Softwise.AI helped Santander Bank Polska absorb the operational impact of its Deutsche Bank merger by deploying an AI knowledge assistant to nearly 400 customer service consultants. With 97.6% correct answers, a 7% reduction in conversation time, and 89.7% user satisfaction, the platform shortened onboarding, reduced complaints, and lowered the share of calls requiring second-line involvement - all while running on top of Santander's existing document estate, with no manual content preparation required.