Offerings · How workflow engagements run
A workflow review, pilot to production or agent build follows the same path, scaled to your organization. Each phase is scoped and priced separately, and you can stop after any of them. Leadership and advisory work runs on a monthly retainer instead.
The first conversation is free. Priced per phase, fixed fee or time and materials, and stated in a written proposal before you commit.
The path
For a smaller firm, Frame and Prove can run as one short engagement.
Phase 1
2 to 3 weeks
We work with you to choose the workflow and agree the owner, baseline, data and target design. You get a written case before any build budget.
You decide: proceed, reshape or stop.
Phase 2
4 to 6 weeks
A working slice on your data and controls, tested by real users, with evidence.
You decide: go or no-go.
Phase 3
8 to 16 weeks per release
Certified data, release pipeline, monitoring, tested runbooks, training and redesigned roles.
You decide: release by release.
Phase 4
Through launch and handover
Your team takes day-to-day ownership while we complete the handover: documentation, training, monitoring, release procedures and an incident exercise. Continued support is available where it adds value.
You decide: what support, if any, continues after handover.
See the first 13 weeks, gate by gate See sample decision documents
Where to start
Call when: You need a data and AI leader’s judgment without a full-time hire.
A senior data and AI leader on a monthly retainer, from strategy through production.
Call when: You have AI ideas, but no agreed first case or owner.
Choose the right first workflow and agree the case before any build budget.
Call when: The demo worked, and weeks later it is still not in production.
Find what is missing, agree the release evidence and prove a working slice.
Call when: You know the workflow and want it running, with controls.
Build and release an agent for one workflow, with documented ownership, controls and a structured handover.
Call when: Two teams get different numbers from the same AI.
Certified data products and shared definitions that people and agents can trust.
Call when: Risk, audit, a customer or an examiner asks for evidence you cannot yet produce.
An identity for every agent, labels that travel with the data, and an evidence pack ready for your risk team, auditors and examiners to review.
For sponsors and procurement
Share the one-page engagement overview
Where engagements start
Illustrative examples of how a first slice could be framed.
Starting question: can relationship managers draft a credit memo section from approved sources, with every figure traceable?
First slice: one memo section, the approved source set, agreed definitions for exposure and covenant terms, and a named reviewer.
Acceptance: each figure traces to a source and version, unsupported statements are flagged and the credit officer keeps the decision. See AI Governance Evidence.
Starting question: can team A prepare an assigned group disability case in region 2 from supported facts and the applicable procedure, while examiners retain every case decision?
First slice: read-only status, events within the assigned case, resolved parties and procedure. Prior-claims history is held; internal notes, payments and customer contact are excluded.
Acceptance: current sources and access checks, independent evaluation, manageable review effort and tested recovery. The synthetic worked example connects the baseline and dependencies to the release decision. Its figures and evidence are invented. See also the Reference Implementation.
Starting question: what would business, technology and control owners need to approve an internal guidance assistant?
First slice: one user group and source collection, a permission-aware retrieval path and a documented release gate.
Acceptance: allowed and denied access tests pass, answers trace to current sources and the operating owner demonstrates escalation and rollback. See the readiness worksheet.
Starting question: which one workflow would show value and pass diligence without a large program?
First slice: one workflow on the stack you already run, a measured baseline, one owner and proportionate controls.
Acceptance: the baseline moves, the controls hold under test, and the team can run it. See AI Token Economics.
Buyer questions
You bring one workflow and the gaps that matter most, and we talk them through together, including whether a Frame phase makes sense. It is free, and nothing is billed until you sign a proposal.
Each phase gets its own written proposal: deliverables, assumptions, your team’s responsibilities, acceptance criteria and the decision at the end. You can stop after any phase. Timing depends mostly on access, the condition of the data and how quickly your risk and compliance teams can engage; the thirteen-week plan shows a typical first run.
Richard does. RJL is principal led: you work with him directly from the first conversation through launch and handover. When the scope needs more capacity or a specialist skill, he brings in partner firms and practitioners he has worked with, and names each one, with their role, in the proposal for your approval. Your organization names the business, technical and control owners who make the decisions.
Usually not. We start with the stack you already run and only propose a change when a specific requirement cannot be met without it. The platforms on the About page are ones Richard has worked with, not a shopping list.
We agree the sources, permitted uses, environment and transfer method during scoping, and start a proof with synthetic or approved material. The website forms and Ontology Studio are not channels for client data.
Tell us your required security evidence, confidentiality terms and supplier checks at the start, and we will work through them before any access is granted. Commitments are made in the contract.
The agreement sets this out before work starts: commissioned deliverables, pre-existing methods, your material and any third-party components. You receive the formats and documentation your team needs to maintain the result.
We agree a baseline and measures with the workflow owner, then test results, source support, exceptions, cost and adoption against it. Recovered capacity and cash benefit are reported separately, as in AI Token Economics, and the next scope follows the evidence.
Handover includes training, documentation, monitoring, runbooks, release evidence and tested rollback and incident procedures. Together we agree when operational ownership has transferred. Continued support is optional and separately scoped.
A free Beta tool for building an ontology on any topic, reviewing its evidence and exporting it to the AI tools you use. It is not an enterprise governance or compliance platform. See the Studio overview.
Next step
We will work through the current state, the missing evidence and whether there is a useful first slice.