Try it small before you commit
A prototype is a working answer to a narrow question. It is not a product. It is built quickly, tested honestly and thrown away or promoted depending on what it shows.
ServiceNow prototypes
Most of our prototypes are on ServiceNow. Examples of questions we build prototypes to answer:
- Will this catalog design be understood? We build a small catalog item and portal page and watch real users try to complete a request.
- Can this workflow be simplified? We model an approval process in a test instance, remove steps and see what breaks.
- Is this CMDB data good enough to rely on? We run data checks on a sample of records to find gaps, duplicates and stale entries before anyone builds reports on top.
- Can these two systems exchange data reliably? We build a small integration between ServiceNow and another tool and test it with realistic volumes and messy input.
- What will this upgrade change? We review customizations against a new release in a test instance and list what needs attention.
A ServiceNow prototype is most useful when it uses the customer’s real processes and real data, in a safe test environment. We insist on test instances and copies of data, never live systems.
What we check on a ServiceNow prototype
- ServiceNow roles and access. Whether the people who should see a ServiceNow form can, and the people who should not cannot.
- ServiceNow approvals. Whether every approval path in the workflow ends somewhere sensible, including when an approver is away.
- ServiceNow notifications. Whether the right people are told at the right moment, and nobody is told twice.
- ServiceNow reporting. Whether the numbers on a ServiceNow dashboard match the underlying records.
- ServiceNow performance. Whether a portal page or a list stays quick with realistic amounts of data.
AI prototypes
Our AI prototypes answer a different set of questions. Is the data behind this use case good enough? Can an AI tool draft a useful first summary of a ticket history, and how often does a reviewer need to correct it? What does the tool do with unusual input? We test with real examples, record what happened and note where the tool was wrong, so the decision to proceed rests on evidence.
Salesforce prototypes
We build occasional Salesforce prototypes, usually to test a CRM workflow idea or an integration with another system. We take on a limited number and say early whether a request fits.
What you get
A short brief stating the question and how we will know the answer.
A working prototype in a safe environment.
A write-up describing what we built, what we found and what we would do next.
A recommendation to stop, change direction or turn the prototype into a project.
What a prototype is not
A prototype is not something to put into production. If it proves the idea, it should be rebuilt properly, with the testing, documentation and security a production system needs.