Skip to main content

What FedGrade does — the capability map

This is the single reference for every capability FedGrade ships, the tier it needs, and how it behaves on provenance. The platform's hard rule runs through all of it: nothing leaves the system without a citation, and a result with thin evidence is graded down honestly rather than dressed up as confident.

The provenance grades

Most capabilities below carry one of four grades on each result. The ladder is enforced in code, not marketing:

  • verified — passed FedGrade's two-pass citation validation (verbatim-substring match + claim-support entailment).
  • source-cited — a direct retrieval from a cited primary source; there is no synthesized claim to validate.
  • modeled — pipeline-derived (entity-graph edges, rankings), with the underlying sources attached.
  • insufficient-evidence — FedGrade cannot support an answer and says so instead of fabricating one.

Intelligence reads

CapabilityTierProvenance behavior
Daily Brief (synthesized read)Free (top-3) / Pro+ (full)Every claim validated; red claims dropped pre-delivery
Forum Forecast (pre-RFP prediction)Free (view) / Pro+ (watch) / Team+ (alerts)Rationale cited to signal ids; materialization log shows misses
Forecast calibration ledgerPublic + Team+ workspaceBrier + reliability over resolved predictions; insufficient-evidence pre-data
Hybrid searchFree+Ranked passages, each source-cited
Forum Network entity graphTeam+ (depth scales by tier)Every edge carries a source pointer; competes_with is heuristic
Cited network paths ("how do these connect")Team+Each hop cited; uncited hops are skipped
Agency profilesPro+Each section a cited bundle; honest empty-state when research library is thin

Capture workflow

CapabilityTierProvenance behavior
pWin capture assessmentTeam+Decomposed factors, each graded; grade rolls up to the weakest weighted factor
pWin calibration ledgerTeam+ (tenant-private)Per-model-version Brier; never pooled, never published
RFP shred (solicitation → requirements)Pro+Every requirement cited to its section + character span
Compliance matrix (L ↔ M ↔ SOW)Pro+Deterministic cross-walk; gap rows flagged; every row editable
Opportunity ranking + pipelineTeam+Decomposed rank (pWin × value × proximity × fit), each component cited; unrankable items surfaced honestly
Intelligent alertsTeam+Every fired alert carries the triggering event + a citation

Deliverables (exports)

CapabilityTierProvenance behavior
Brief PDFFree (5/day) / Pro+ (unlimited)Inline citations + source list; retractions struck through
Brief DOCXPro+Footnoted citations, editor-friendly
Brief deck (PPTX)Team+Per-claim slides + source appendix
Capture package (DOCX/PPTX)Team+Brief + pWin + compliance matrix + agency snapshot, each section cited or honestly absent
Pipeline board report (DOCX/PPTX)Team+Ranked pipeline with cited factor drivers + not-yet-rankable section

Trust + integrations

CapabilityTierProvenance behavior
Signed claim artifactsResearch library-level briefsEd25519-signed; verifiable without trusting FedGrade
Public claim verification (/verify, POST /api/verify)PublicRecompute canonical form + signature + excerpt hash
MCP server (9 tools)Sample (free) / Enterprise+ (live)Every tool payload graded + cited; no ungraded answers
REST API v1Enterprise+Scoped keys; provenance block on every object
WebhooksEnterprise+Event delivery with retry

What is roadmap, not live

These are committed directions, set up during Government-tier onboarding, not shipped capabilities today: PIV-CAC sign-in, FIPS mode, AWS GovCloud residency, and FedRAMP authorization. The Government tier article and /pricing carry the current status. Treat any claim that these are live as wrong.

For the buyer-facing version of this map, see Solutions. For the developer surface, see the API docs and MCP server docs.

Last updated 2026-06-13.


Was this article helpful?
Still need help? File a ticket via the floating "?" button on any app view, or email support@fedgrade.com.
What FedGrade does — the capability map · Help