The signal layer
The part nobody else brings
Every services firm tells you the work is getting done. We instrument it. The layer underneath SignalImpact is OLi — Dataken's platform, in production for over five years, capturing a million work signals a day.
Signal, context, impact
The name is the mechanism. Each stage is a different technical problem, and skipping any of them is how most 'AI-powered' services claims fall apart.
- 01Signal
The work emits signals
Which applications are open, what moves between them, how long a task actually takes, where it stalls. OLi's computer vision reads the active window with machine learning at the endpoint — it does not store or transmit raw screenshots.
- 02Context
Signals become context
One event means nothing. The same event across weeks, roles, and systems is a process. Process mining turns the stream into the actual shape of how your operation runs — which is almost never the shape of the documented procedure.
- 03Impact
Context becomes impact
The findings turn into changes: a step removed, a license cancelled, a task moved to a teammate who is faster at it, a person coached in the moment rather than at the annual review. Then it is measured again.
Capabilities
What runs underneath every engagement
These are modules of the OLi platform, not a roadmap. They are live today because Dataken has been operating them for years in commercial deployments.
Operational analytics
Where the hours go, by person, process, and system. The baseline that makes every later claim of improvement checkable.
Process mining
The real path a task takes through your systems, including the loops and the rework nobody documented.
Computer vision
Machine learning on the active window at the endpoint. Structured metadata is transmitted; raw images are not stored or sent.
Real-time coaching
Help at the moment of the task, not in a quarterly review — which matters most for a new remote teammate learning your systems.
Micro learning
Short, targeted training delivered against the specific gap the signals identified.
Security insights
Shadow IT, unsanctioned tools, and risky data paths surfaced from the same signal stream rather than a separate audit.
Limits
What this is not
Instrumentation in a mission organization is a governance question before it is a technical one. These are commitments, not caveats.
- It is not surveillance of your congregation, members, or clients. The signal layer instruments the staff and teammates delivering the services we are contracted for — nobody else.
- It is not keystroke logging or screenshot archiving. Computer vision runs at the endpoint and sends structured metadata; raw screen images are neither stored nor transmitted.
- It is not deployed without disclosure. Every person under instrumentation is told, in writing, before it starts. For a mission organization that is not a legal nicety, it is the whole ballgame.
- It is not a performance-policing tool we will help you point at your own employees. If that is what you want, we are the wrong firm.
Built by Dataken, not assembled for this launch
Every number on this site is sourced. These are the platform's, and they belong to Dataken rather than to SignalImpact — we say so rather than implying a track record we have not earned yet.
- work signals in the graph
- 10B+work signals in the graphSource: Dataken production infrastructure
- years in continuous production
- 5+years in continuous productionSource: Dataken production infrastructure
- signals captured per day
- 1Msignals captured per daySource: Dataken production infrastructure
SignalImpact launched in 2026 and has one client engagement in production, described above by type and scale. The platform figures are Dataken's, not a claimed SignalImpact track record. We publish no outcome percentages because we do not have any yet; when we do, they will appear here with the method attached.