Compare responsibility boundaries
A platform, a consultancy and a specialist studio can sell different portions of the same workflow. Write down who prepares data, runs training, owns evaluation, deploys, monitors and handles incidents. A low platform fee may assume substantial buyer engineering. A fixed project fee may exclude ongoing operations.
Normalise one-off and recurring costs
Separate preparation, training, evaluation and exit from serving and monthly support. Mark an omitted fee as unknown rather than zero. Note volume limits, assumptions about buyer effort and the trigger for a change request. Do not add overlapping budget ranges as if they were mandatory packages.
Require comparable evidence
Ask for dated comparator details, protected slices, whole-task acceptance and failure analysis. Source references allow a reviewer to trace a claim; they do not themselves verify the claim. Treat supplier-provided scores as inputs to diligence. Block selection when data rights, operating ownership or evidence are unresolved.
Use the directory as a starting point
Frontier Development’s landscape entries identify organisations from the supplied brief. They are not an endorsed shortlist and their current scope is unverified. Build a shortlist only after checking fit for the approved task and obtaining comparable obligations. Fictional proposals in the normaliser demonstrate the method without implying a real supplier offer.
Bring to the decision review
Bring two or more proposals normalised to the same task and evaluation policy. Ask each provider to identify buyer-retained work explicitly and to answer the same clarification questions. Keep a source reference for every scored claim. Where evidence is missing, reduce coverage or block the decision rather than substituting a plausible score. This preserves a distinction between an attractive narrative and a comparable commitment.
Sources & editorial record
This is Frontier Development’s decision framework. The references below support the general technical concepts; the worked scenarios and procurement questions are our analysis.
References accessed 16 October 2026. No current price, product availability or legal clearance is asserted. Published study findings are attributed above; this guide reports no Frontier Development experiment. Read the comparison method.
Review history and research boundary
Edition 1.1 · 16 October 2026: added claim-level attribution, explicit evidence boundaries and links to the flagship synthesis. Published by Frontier Development. No named external technical reviewer is recorded. This guide provides a decision framework; it is not a Frontier Development experimental result.
Read the primary-study synthesis →Open the proposals tool
Explore a complete synthetic example before using your own inputs.