Keelforge

Screen 09 · Feedback Agent · fn_learn

Learn & outcomes

Win or lose, the outcome is written back. Evidence that scored well is promoted in the retrieval index; evidence that drew a finding is demoted.

Demo snapshot as of 15 Aug 2026. Every tile, row and panel below reads from this snapshot.

58%

Win rate · rolling 12m

+9pts

Governance score uplift

Explain the +9pts

14

Evidence items re-scored

2

Sources demoted

Period

Showing rolling 12 months. Period recalculation ships in a later wave - this selector currently pins the label, not the data.

Sector

Outcome capture

4 of 4 bids
Bid IDBuyerSectorStatusScoreGovernance scoreLessons capturedNote
BID-2026-009Caledon TrustFinancial servicesWon88 / 10092 / 1003Governance section scored highest of five bidders.Case study →Requirement matrix →Audit trail →
BID-2026-007MoD - Strategic CommandDefenceLost71 / 10078 / 1005Price 12% above winner; technical scored joint-first.Case study - none yetRequirement matrix →Audit trail →
BID-2026-005Perrin Asset ManagementFinancial servicesWon84 / 10087 / 1002Signed case studies cited as a differentiator in feedback.Case study →Requirement matrix →Audit trail →
BID-2026-002Northgate Building SocietyFinancial servicesLost63 / 10066 / 1004Stale regulatory citation flagged by evaluator.Case study - none yetRequirement matrix →Audit trail →

Win rate by sector

The rolling 12 month win rate broken down across the three sectors the platform competes in.

SectorBids submittedBids wonWin rate
Defence4250%
Financial services5360%
Government and public sector3267%
Overall12758%

Sample sizes reflect a demo dataset - production win rate calculations will apply the same method to the live bid outcome ledger.

Governance uplift breakdown

The +9pts headline is the difference between the platform's average governance score today and the same average twelve months ago. This panel shows how the gap opened.

Component12 months agoTodayUplift
Signed evidence coverage68%84%+16pts
Named signatory coverage71%89%+18pts
Audit trail completeness82%94%+12pts
Regulatory citation recency61%66%+5pts
Composite governance score71/10080/100+9pts

Signed evidence coverage

Twelve months ago 68 · today 84

Named signatory coverage

Twelve months ago 71 · today 89

Audit trail completeness

Twelve months ago 82 · today 94

Regulatory citation recency

Twelve months ago 61 · today 66

Composite governance score

Twelve months ago 71 · today 80

Regulatory citation recency at +5pts is the smallest mover and is the candidate for the next improvement wave.

See the retrieval-index moves that produced this uplift →

Retrieval-index movements

Every outcome writes a move into the retrieval index. Evidence that scored well moves up; evidence that drew a finding moves down. The moves below reflect the four outcomes in the capture panel above.

Evidence IDTitleSectorMoveReasonWritten by
EV-2026-0142Caledon Trust - governance section anchor citationFinancial servicesPromoted +2 tiersCited in Caledon Trust winning response, evaluator called out the governance section as strongestfn_learn on 2026-08-15
EV-2026-0138SS1/23 model risk framework summaryFinancial servicesPromoted +1 tierReused in Caledon Trust and Perrin, scored highest on model risk in bothfn_learn on 2026-08-15
EV-2026-0121Named signatory register - Kevin Telford SM&CR mappingFinancial servicesPromoted +1 tierCited in every winning FS response, no findingsfn_learn on 2026-08-15
EV-2026-0109Procurement Act 2023 MAT criteria referenceGovernment and public sectorPromoted +1 tierCited in winning government responses across the rolling 12 monthsfn_learn on 2026-08-15
EV-2025-09872019 outsourcing framework referenceFinancial servicesDemoted -2 tiersCited in Northgate Building Society lost response, evaluator flagged as stalefn_learn on 2026-08-15
EV-2025-1042Defence Standard 05-138 legacy referenceDefenceDemoted -1 tierSuperseded by 2024 revision, cited in MoD Strategic Command lost responsefn_learn on 2026-08-15
EV-2024-0876Generic AI ethics framework citationFinancial servicesDemoted -2 tiersCalled out by evaluator as insufficiently specific in Northgate responsefn_learn on 2026-08-15
EV-2026-0155Social Value Model Theme 5 referenceGovernment and public sectorRe-scored - no moveCited but was neither differentiator nor detractorfn_learn on 2026-08-15
EV-2026-0161Third-party assurance PS21/3 excerptFinancial servicesRe-scored - no moveCited without commentaryfn_learn on 2026-08-15
EV-2026-0148Cabinet Office SME participation guidanceGovernment and public sectorRe-scored - no moveCited without commentaryfn_learn on 2026-08-15

The 14 evidence items counted in the third stat tile above include the ten shown here plus four intake-side re-scores held on the source Signal Fabric.

Every move is written to the audit log as an fn_learn call - see the audit log →

Lessons library

14 of 14 lessons

Every outcome produces a learning that changes how the next similar bid is drafted. The lessons below carry a named owner and a status - planned, in-flight or applied.

Lesson IDBid IDSectorLessonOwnerStatus
L-2026-042BID-2026-009 Caledon TrustFinancial servicesGovernance section anchor citation belongs at the top of every FS response, not buried mid-wayKevin TelfordApplied
L-2026-041BID-2026-009 Caledon TrustFinancial servicesNamed signatory register with SM&CR mapping should be linked from every citation, not attached separatelyKevin TelfordApplied
L-2026-040BID-2026-009 Caledon TrustFinancial servicesTwelve-month recency threshold for regulatory citations reads as governance discipline to FS evaluatorsKevin TelfordApplied
L-2026-039BID-2026-007 MoD Strategic CommandDefencePrice must be benchmarked to the winning band published in Defence Digital OGM data, not to internal cost-plusKevin TelfordIn flight
L-2026-038BID-2026-007 MoD Strategic CommandDefenceTechnical joint-first with lower price still loses - technical differentiation must be quantified, not assertedKevin TelfordIn flight
L-2026-037BID-2026-007 MoD Strategic CommandDefenceDefence Standard citations must reference the latest revision - legacy references demoteKevin TelfordApplied
L-2026-036BID-2026-007 MoD Strategic CommandDefenceCleared personnel must be named at the intake stage, not deferred to mobilisationKevin TelfordPlanned
L-2026-035BID-2026-007 MoD Strategic CommandDefenceSocial value alignment to defence-relevant themes (skills, veterans employment) should lead, not followKevin TelfordPlanned
L-2026-034BID-2026-005 Perrin Asset ManagementFinancial servicesSigned case studies from named clients outperform anonymised case studies as differentiatorsKevin TelfordApplied
L-2026-033BID-2026-005 Perrin Asset ManagementFinancial servicesModel risk framework references should quote SS1/23 tier language directly, not paraphraseKevin TelfordApplied
L-2026-032BID-2026-002 Northgate Building SocietyFinancial servicesRegulatory citations older than 24 months should trigger a stale-source gate before draft submissionKevin TelfordApplied
L-2026-031BID-2026-002 Northgate Building SocietyFinancial servicesGeneric AI ethics frameworks read as filler - every AI reference must cite the buyer's own model risk regimeKevin TelfordApplied
L-2026-030BID-2026-002 Northgate Building SocietyFinancial servicesThird-party assurance PS21/3 detail should be surfaced as a mandatory schedule item on FS intakesKevin TelfordIn flight
L-2026-029BID-2026-002 Northgate Building SocietyFinancial servicesNamed signatory of appropriate seniority for outsourcing sections requires SMF-mapped attributionKevin TelfordApplied
Every applied lesson is written to the audit log - see the applications →