Connect what your business means to what its systems do.
Start where meaning, information or responsibility gets lost. Bring clarification, data modeling, implementation, human review and verification together around a useful result.
This framework connects practices demonstrated across Blake’s independent projects. The projects have their own scope and maturity; they are not one integrated commercial platform.
AI-assisted delivery, shaped around the work.
AI supports how I develop a solution. Depending on the problem, the result may be a better-configured tool, an integration, a dashboard or a custom workflow. It does not necessarily need AI to operate.
The separate fictional demonstration illustrates missing information and changed requirements. It does not execute these projects or establish a business outcome.
A workbench connecting language, context, operations, scenarios, evidence, and human control.
Evidence: Reviewed implementation and controlled studies support these bounded mechanisms. The public exhibit uses synthetic examples and explains the scope of the work.
Limits: Research mechanisms and an authored finite grammar do not establish general language understanding or autonomous business execution. Recorded compute studies do not establish universal speedups. Commercial reuse is subject to licensing.
A Python controller connecting structured file-audit requirements to plans, execution, receipts, and revision.
Evidence: The public preview includes tests, deterministic fixtures, validation notes, and known issues. The reviewed demonstration changes a resource budget while retaining required digest and byte-count checks.
Limits: Implemented operations are bounded file-audit tasks. Independent validation is pending. Requirement prose is retained for human alignment, not automatically judged. Source is available under PolyForm Noncommercial; commercial use outside its grant needs separate permission.
The separate fictional demonstration illustrates missing information and changed requirements. It does not execute these projects or establish a business outcome.
An explorer for North Dakota permit, spud, and well-status records with explicit decision boundaries.
Evidence: Representative source includes role-specific decision gates and bounded ranking logic. The packet architecture distinguishes design intent, control envelopes, measured information, and authorized action.
Limits: The inspected public index does not supply oil, gas, or water production volumes. Source review does not establish commercial performance, operator affiliation, or a production authorization system.
A well-log workbench for importing local data, comparing signals, and inspecting estimates with supporting evidence.
Evidence: The reviewed pipeline connects imported data, geometry and priors, residual calculations, acceptance checks, and export.
Limits: Imported-data analysis and advisory behavior do not establish field validation or authorize rig control. No industrial performance or certification claim is implied.
The separate fictional demonstration illustrates missing information and changed requirements. It does not execute these projects or establish a business outcome.
A Python controller connecting structured file-audit requirements to plans, execution, receipts, and revision.
Evidence: The public preview includes tests, deterministic fixtures, validation notes, and known issues. The reviewed demonstration changes a resource budget while retaining required digest and byte-count checks.
Limits: Implemented operations are bounded file-audit tasks. Independent validation is pending. Requirement prose is retained for human alignment, not automatically judged. Source is available under PolyForm Noncommercial; commercial use outside its grant needs separate permission.
An explorer for North Dakota permit, spud, and well-status records with explicit decision boundaries.
Evidence: Representative source includes role-specific decision gates and bounded ranking logic. The packet architecture distinguishes design intent, control envelopes, measured information, and authorized action.
Limits: The inspected public index does not supply oil, gas, or water production volumes. Source review does not establish commercial performance, operator affiliation, or a production authorization system.
Trace one recent change across an existing process or tool boundary. Define the smallest integration or modernization pilot that makes its effects inspectable.
PROPOSED BUSINESS APPLICATION
A concrete first output
A change-impact map and a pilot that connects requirement versions, review decisions and repeatable checks.
How we would evaluate it
Introduce a changed requirement and inspect which decisions need review. Recheck the affected output and distinguish passing, failing and unresolved results.
These are scope options to discuss, not delivered client results or guarantees.
The separate fictional demonstration illustrates missing information and changed requirements. It does not execute these projects or establish a business outcome.
A workbench connecting language, context, operations, scenarios, evidence, and human control.
Evidence: Reviewed implementation and controlled studies support these bounded mechanisms. The public exhibit uses synthetic examples and explains the scope of the work.
Limits: Research mechanisms and an authored finite grammar do not establish general language understanding or autonomous business execution. Recorded compute studies do not establish universal speedups. Commercial reuse is subject to licensing.
A Python controller connecting structured file-audit requirements to plans, execution, receipts, and revision.
Evidence: The public preview includes tests, deterministic fixtures, validation notes, and known issues. The reviewed demonstration changes a resource budget while retaining required digest and byte-count checks.
Limits: Implemented operations are bounded file-audit tasks. Independent validation is pending. Requirement prose is retained for human alignment, not automatically judged. Source is available under PolyForm Noncommercial; commercial use outside its grant needs separate permission.
An independent checker for finite operator proposals, with explicit verdicts and inspectable certificates.
Evidence: The reviewed source separates checker imports from the generator and retains specification and digest checks. Incomplete coverage or exhausted checking budgets can produce UNKNOWN.
Limits: The checker applies to finite declared domains. It is not a universal solver or an arbitrary natural-language-to-software compiler. Separation of generator and checker does not establish independent professional validation.
Keep the intended meaning connected to the working result.
These practices can repeat as understanding changes. A failed check may return the work to definition; a new constraint may require a different structure. The next step follows the evidence and the people responsible for the decision.
Agents and harnesses can assist with planning, implementation, checks and revisions. People establish requirements, permission to act and acceptance. Source review describes implementation; recorded tests describe the cases that were exercised; functional review considers observed behavior. Independent specialist validation is a separate activity and is claimed only when documented. An automated checker does not itself establish independent professional review.
An updated plan, a record of the change, and evidence that the revised path still serves its intended purpose.
Start with one boundary worth improving.
A workflow assessment brings together someone familiar with the process, non-sensitive examples and known constraints. The output is a reviewable brief and scope options for a bounded pilot.
Scope and pace adapt to available resources. Reduced-fee or pro bono work may be considered for aligned causes, subject to capacity. Agree on responsibilities, acceptance checks and a stopping point before expanding.