Meeting intelligence
Recordings become minutes, summaries, and executive briefs — in the format the file requires, with a person in review.
Read the caseDISols · Digital solutions practice · New York
Applied AI for operating businesses.
Engagements begin with one well-defined task and a working tool. When the result proves useful, the same method carries forward: into workflow systems, governed processes, and platforms on infrastructure you control.
Every build is shaped by clear data boundaries, reviewed outputs, named ownership, and a measure of operational value.
01 / Where we begin
Bounded problems first.
Operating businesses carry repeatable work that deserves better instruments: meetings that need reliable records, receipts that need structure, meter data that should become decisions. These are precise, bounded problems. They are the right place to begin.
The meeting record
A conversation becomes a record: minutes, summaries, and executive briefs produced from recordings, formatted for the file, and reviewed by a person before the record is final.
The expense trail
Receipts, voice notes, and memos become structured entries in the owner's own spreadsheet. The service works across languages while keeping the record under the owner's account.
The unread meter
Hourly water-meter data becomes a management report: consumption patterns, anomalies, and likely leaks presented in a form a property manager can act on.
02 / The engagement model
From one task to a platform.
Three stages, one discipline. Each stage produces its own value, and each advance is earned by evidence from the work already in use.
III
A platform in your infrastructure
Customized implementations at organizational scale — records-management integration, cloud support for on-premise deployment, third-party infrastructure where requirements demand it. Scoped and staged.
II
A system in your workflow
Custom systems connected to the tools you already run, with defined data boundaries and human review where judgment matters.
I
A working instrument
One recurring task, delivered as a standalone tool in days to weeks. The first instrument proves the use case, the review path, and the value of building further.
03 / Discipline
Four commitments, every engagement.
01
Data boundaries
Data boundaries are defined before work begins: what can use an external model, what remains inside your accounts, and what belongs on infrastructure you control.
02
Reviewed outputs
AI output enters a defined review path before it becomes part of your records. Human judgment stays where the work requires judgment.
03
Named ownership
Every system is delivered with a named owner, a maintenance path, and the operating notes needed to keep it useful after launch.
04
Measured results
Each engagement defines the operational measure in advance, so the next decision is made from observed results: continue, adjust, or complete the work at the right stage.
04 / Selected work
Technical work, documented plainly.
Analytics for operations
Hourly water-meter data turned into consumption patterns, anomaly detection, and leak indications for multi-unit properties.
Read the caseTechnical video restoration
A museum exhibition video restored from SD to 2160p, with final masters prepared under public-institution documentation and data-handling constraints.
Read the case05 / The practice
Principal-led by design.
DISols is led by Igor Sevonkaev, a PhD physicist and AI/R&D transformation leader with fifteen years across industrial R&D, analytics, digital platforms, and practical AI adoption. The practice remains deliberately compact: direct judgment, close communication, and clear accountability for the systems delivered.
Specialist collaborators are brought in when a project calls for them. Assessment, architecture, and accountability remain principal-led.
Next step
Begin with one task.
Describe one recurring task in your operation. You will receive a direct assessment of the responsible path: what it would require, how it could be reviewed, and where it could lead.