We hear about cognitive automation. But what does success look like? Meet Cognitive Assist, our virtual agent. These virtual agents, powered by Watson, have ingested a vast corpus of knowledge about the applications IBM support, so they can provide the guidance an experienced coach could give – consistently and in real time. Read more about how Cognitive Assist can help you.
Scaling API-first – The story of a global engineering organization
Cognitive Automation: What does success look like?
1. C O G N I T I V E A U T O M A T I O N
WHAT DOES SUCCESS LOOK LIKE?
B A R R Y M I T C H E L L – G L O B A L L E A D , I B M A U T O M A T I O N W I T H W A T S O N
2. WHAT DOES SUCCESS LOOK LIKE?
S U C C E S S C R I T E R I A
M E E T C O G N I T I V E A S S I S T
L I K E A B I L I T Y
B U S I N E S S V A L U E
S C A L A B I L I T Y
S A M P L E S C O R E C A R D
3. M E E T C O G N I T I V E A S S I S T
‘Sounds great. But how do we know they’re working?'
These virtual agents, powered by Watson, have ingested a vast corpus of knowledge about the applications IBM support, so
they can provide the guidance an experienced coach could give – consistently and in real time.
Over 25,000 IBMers have access to Cognitive Assist solutions to guide them
in their work.
Agent
Assist
For support agents.
Provides context-
specific technical
assistance to resolve
issues without the need
to escalate to senior
support agents.
For junior developers
Delivers a consistent,
full-coverage learning
on coding practices,
while freeing up senior
developers time
Coding
Assist
Employee
Assist
For employees.
Provides context-
specific procedural
guidance to resolve
issues without the need
to call the Helpdesk.
4. Do users like Cognitive Assist?
Does Cognitive Assist deliver business value?
Is Cognitive Assist a scalable solution?
To ensure a good return on investment, we ask three questions when
we roll Cognitive Assist out to a new team.
S U C C E S S C R I T E R I A
5. Do users like Cognitive Assist?
Measuring likeability
Every time Employee, Agent or Coding Assist answers a question, it requests feedback. Users state
whether the answer given was 'Helpful' or 'Not helpful’.
Users can also provide further comments if they wish. In some pilots – for example the launch of a new corpus - we run
communications campaigns to drive up the proportion of users providing feedback.
In some pilots we have also surveyed users, asking them about their preferred method for receiving application support.
If they don’t like it, they won’t use it
And if they don’t use it, we won’t see business benefits. Plus, because Watson learns from each
interaction, we need high usage to continue to improve the tool.
WHY IT
MATTERS
L I K E A B I L I T Y
6. Does Cognitive Assist deliver business value?
We expect an investment in Cognitive Assist to pay off in months, and ongoing maintenance costs to
be offset by the significant savings Cognitive Assist delivers.
It is important to begin measuring the savings early to ensure they are on track from the start.
WHY IT
MATTERS
Three drivers of business value
1. Productivity gains. Agents, employees and coders save time by accessing expertise in real-time - especially when a subject
matter expert might otherwise take a day to respond.
2. Labor cost reductions. Junior agents and coders are upskilled faster and able to resolve more complex issues sooner. An
increase of 20% in the number of tickets resolved by L1 agents is typical - and reduces spend on L2 labor.
3. Reduced ticket volume.
⇢Roughly 40% of L1 tickets are knowledge queries which can be resolved by Watson, when employees use Employee
Assist.
⇢Because code being put into production is better quality, fewer issues occur.
B U S I N E S S V A L U E
7. Is Cognitive Assist a scalable solution?
The more user groups an application can support, the more value it will deliver.
No matter how well liked a Cognitive Assist solution is, or the savings it generates, if deploying it is
harder than launching a space mission it’s not a viable solution.
WHY IT
MATTERS
Assessing scalability
Potential usage We aim to expand our successful pilot to a larger user population. Some common opportunity areas:
Type of Activity Solution Benefits
User Queries ( ‘How to’ Questions) Employee Assist Eliminate 60 –70 % “how-to” queries over time
Service Requests ( eg Password Reset, VM Reboot) Cognitive Assist + integration layer + Dynamic Automation Eliminate 40 –60% service requests over time
Application Development Employee Assist / Agent Assist Eliminate 60-80% initial training effort
Application Management Agent Assist 12% + Productivity gains in a traditional AMS model
Time How long will it take create a corpus? For our most mature applications, such as certain SAP or Oracle modules, a vast
body of knowledge now exists. Creating a custom corpus can take 6 to 12 weeks depending on the volume and quality of existing
documentation, and subject matter expert availability.
S C A L A B I L I T Y
8. Success Criteria Proof points Criteria met?
Users like Cognitive Assist
4/5 users prefer using Cognitive to calling Helpdesk
97% of answers were rated helpful by users
Cognitive Assist delivers
business value
25-45% ticket reduction because users are resolving queries via Employee Assist
20% more tickets resolved by L1 agents, instead of being escalated to L2
Users claim 4 minutes in time saved per ticket.
Average turnaround times for user queries with Employee Assist are minutes,
compared with 2.3 days for Helpdesk.
Cognitive Assist is a scalable
solution
Cognitive successfully ingested documents from 19 sources, and 4 FTE took 6 weeks
to create over 450 Q&A pairs.
A Cognitive corpus already exists for 12 of the top 20 apps (by ticket volume)
S A M P L E S C O R E C A R D
This scorecard was compiled during a pilot for a major US bank. Following the
successful pilot, Cognitive Assist was rolled out to ten additional applications.
9. WHAT DOES
SUCCESS LOOK
LIKE FOR YOU?
V I S I T W W W . I B M . C O M / A U T O M A T I O N F O R M O R E
O N T H E B U S I N E S S I M P A C T O F A U T O M A T I O N .
T O T A L K T O A N E X P E R T , C O N T A C T
G B S A U T O @ U S . I B M . C O M