Decision Support, Not Autonomy

Technology work starts at the point of collection, because no amount of later processing repairs an inconsistent form. We define entry fields, controlled vocabulary for habitat and condition descriptions, and documentation sized for use at a site in poor weather. Existing paper records and spreadsheets are treated as source material to be structured, not discarded.

• Structured inspection and collection fields
• Controlled vocabulary for condition and habitat terms
• Import paths for historical spreadsheets and paper records
• Entry conventions documented for field staff

Surveillance information becomes far more useful when it is organized geographically. Monitoring locations, habitat classes, inspection coverage, and recorded activity are placed on a common spatial reference so a program can see where it monitors, where it does not, and how areas differ.

• Monitoring location register with coordinates and habitat class
• Coverage and gap visibility across the service area
• Habitat and activity layers for review
• Site grouping for multi-location programs

AI-assisted processing summarizes long-form field notes, groups entries by location and period, and highlights where recent observations differ from a site’s own history. It is applied to organization and comparison, not prediction. We do not forecast outbreaks and do not present statistical modeling as a determination of risk.

• Field-note summarization with source traceability
• Location and period comparison
• Deviation from site history highlighted for review
• No outbreak prediction or risk determination

Every analytical output is presented for confirmation by qualified personnel before it influences an operational decision. Reviewers see the underlying records, can reject or amend findings, and their decisions are recorded alongside the information that informed them. Disagreement between staff experience and analysis is documented rather than resolved in favor of the system.

• Confirmation step required before operational use
• Underlying records available to every reviewer
• Reviewer decisions recorded with their basis
• Documented disagreement rather than silent override

Reviewed information supports planning: which locations to examine next, how routes and schedules are organized, which equipment checks are due, and how work is distributed across a field team. Plans are proposals for staff to accept, adjust, or reject. Nothing is dispatched automatically.

• Prioritized review lists for staff confirmation
• Route and schedule organization support
• Equipment check and servicing intervals
• Workload distribution across field teams

Reporting is built so a program can repeat it without a special effort each cycle. Coverage, documentation completeness, inspections performed, and recorded actions are assembled into exportable summaries for internal review, supervisory discussion, or funder requirements, with limitations stated on the output itself.

• Repeatable coverage and completeness summaries
• Exportable formats for internal and external audiences
• Period-over-period comparison
• Stated limitations carried on the report

Program information is handled under access controls appropriate to the engagement, with data handling, retention, and processing arrangements defined in the project agreement before work begins. We do not claim perfect security, because no organization can. We do commit to stating plainly how information is stored, who can reach it, and how long it is kept.

• Access limited to personnel performing the engagement
• Handling and retention defined in the project agreement
• Customer control over what is shared
• No claim of absolute security

Our tools are limited by the data available to them. Sparse records, inconsistent collection, undocumented habitat change, and gaps in coverage all constrain what any analysis can support. We state these limits on the deliverable rather than in a footnote, because a program that misjudges the confidence of its information makes worse decisions than one working from acknowledged uncertainty.

• Output quality bounded by input quality
• No prediction of disease outbreaks or transmission
• No species determination or laboratory analysis
• Limitations stated on every deliverable