Score · AI Recommendations
Most AI pilots start from the tool, not the process.
Every process contains potential AI opportunities, but not every opportunity creates meaningful value. Technology-led approaches start with what AI can do, while use case-led approaches start with a specific task to automate. Crismo takes a process-led approach: analysing how work actually happens, including volume, effort and cost, to reveal which opportunities are truly worth pursuing.
- AILLMsAgentsCopilotsChatbots1
Technology-driven
- Summarise documentsClassify ticketsGenerate content2
Use-case-driven
- 3+€34k / yr
Opportunity-driven
Use cases tell you what AI can do. Opportunities tell you where it’s worth doing.
Score where AI pays off.
AI Recommendations read your process model and rank every step by value, effort and confidence. Not “what can AI do”, but where it moves the number.
AI opportunities✦ scored
3 foundFollow-up agent after proposal
Quotes wait 9.4 d in Proposal sent with no owner activity in 71% of cases. An agent drafts the follow-up from the proposal and the last thread and files it for the owner.
Readiness scoring before solution design
Discount approval agent
How a score is built.
Every opportunity is tied to a step in your model and backed by the attributes and live data on it. Nothing on the card is a guess.
Follow-up agent after proposal
Quotes wait 9.4 d in Proposal sent with no owner activity in 71% of cases. An agent drafts the follow-up from the proposal and the last thread and files it for the owner.
- 1
Where
The exact step in your process model. The opportunity has a location and an owner from the start, not a department.
- 2
Value
Volume × waiting time × cost, taken from the attributes on that step. Euros per year, not a maturity score.
- 3
Effort
How large the change is: the weeks it takes to build and roll out.
- 4
Confidence
How solid the data behind the estimate is. Low confidence is shown, not hidden.
Example · Quote-to-cash · 620 cases a month
Three opportunities found. Ranked, costed, ready to build.
Follow-up agent after proposal
Quotes wait 9.4 d in Proposal sent with no owner activity in 71% of cases. An agent drafts the follow-up and files it for the owner.
Readiness scoring before solution design
Enterprise quotes run the solution design session 2.1× because scope moves. A scoring agent checks scope, stakeholders and budget at intake.
Discount approval agent
6% of quotes carry an unmodelled discount approval that adds 1.9 d. An agent pre-checks against policy and routes only exceptions to the sales lead.
- a year, combined
- €74k
- total effort
- 6 weeks
- out of 14 in the process
- 3 steps
Found it. Now build it.
“Link to path” creates a new innovation path: the future-state version with the agent task placed at that step. Review it with the team, publish, and connect the agent through MCP so it runs on the approved process.
How Ship works- ScoreFollow-up agent after proposal+€44k / yr
- ShipInnovation path: Quote-to-cash v8agent task at “Quote and present offer”
- ShipReview and publishSales, Legal approved
- ShipConnect via MCPagent runs on v8
With Agentic Process Mining
The same scores, from live data.
Connect your systems and the finder scores against execution data instead of attributes. Waiting times, rework and volumes come from what actually happened, per case.
About the add-onQuotes wait 9.4 days in Proposal sent with no owner activity in 71% of cases.
Measured across 1,860 cases · last 90 days · HubSpot
See your first three opportunities this week.
Model one process, run the finder, get a ranked list.