Cralgo
The conversation had moved from model performance to which decisions the system is allowed to take
In operator conversations, model quality was largely assumed and the open question was routing — which decisions go to a model, which to a person, which to neither alone.
Evidence · Field Note
In operator conversations, model quality was largely assumed and the open question was routing — which decisions go to a model, which to a person, which to neither alone.
Environment
Ecosystem visit · US west coast
Evidence status
Field observation
What was being observed
Technology × Organisations
Context
A series of operator, engineering and investor conversations during an ecosystem visit. Organisations are not identified.
The context is part of the observation. Environments are described at setting level only.
Observed
Model benchmark discussion was noticeably thinner than in earlier visits. What came up repeatedly instead was context and routing: what the system is given to work with, and which classes of decision it is permitted to complete without a person. Several teams described having built the model capability first and only later discovering they had no agreed answer to the routing question.
One possible reading
One possible reading is that the harder problem has moved from the model to the boundary around it, and that boundary is an organisational decision rather than a technical one. It may also be that this cohort simply moved past benchmarks earlier than most.
A reading is not the observation. This note describes one environment and does not establish what organisations in general do.
Still open
Whether teams that answered the routing question explicitly ended up with different results, and whether this reflects the wider market or one unusually advanced set of environments.
Question this sharpens
Who in an organisation is actually entitled to decide which decisions a system may take on its own?
Basis
Direct observation in this environment, recorded by the Cralgo team present. Qualitative and unmeasured.
Related
Related reading
Industry
Knowledge · Architecture
Knowledge · Playbook
We identify emerging patterns and opportunities through research and Signals, and take them into real-world application through technology strategy and leadership, AI and data, digital commerce and consumer platforms, product and engineering, and technology-led growth and transformation.
From emerging patterns to real-world application.
Emerging Patterns → Research & Signals → Collaborate → Technology & Application → Better Outcomes
Where we apply this thinking
Technology Strategy & Leadership · AI & Data · Digital Commerce · Consumer Platforms · Product & Engineering · Technology-led Growth & Transformation
Psychology × Technology × Organisations → Outcomes — the research lens Cralgo explores through.
Explore Signals · Collaborate with Cralgo
Signals · What is a Signal? · Open Signals · Threads · Papers · Reflections · Field Notes · Questions · What we are learning · Research · Technology · Open Explorations · Collaborations · Evidence · About Cralgo · Beyond · Connect
Founding
Cralgo was founded in 2025 by Anil Kabir Kumar. Legal entity: Cralgo Innovations (OPC) Private Limited, incorporated 27 June 2025.
The three lenses
Psychology — how people perceive, decide and behave. Technology — what the system makes possible or constrains. Organisations — what conditions shape action. Outcomes emerge from their interaction.
Research — the questions underneath the work
Collaborative investigation with organisations, universities, practitioners and the people a question affects. Commissioned research is disclosed as such.
Beyond — questions Cralgo carries on its own account
Beyond describes the origin of a question: one Cralgo carries independently, or around a public-purpose context.
- Beyond
- Explore
- Beyond Ecosystem
- AI and public purpose — a provisional context inside Beyond
- Cralgo Now — a public-interest initiative about timing and attention
Working together — the belief behind all of it
Listening before concluding, room for different perspectives, useful disagreement, and evidence that still matters.