Risk and control mapping
Map AI model risks to the controls that mitigate each one
Try in PlaygroundExample brief
We are taking an AI credit-scoring model to production and need a slide that maps each deployment risk to its control. On the model side, training data can drift from the current applicant population, mitigated by monthly population-stability checks with an auto-retrain trigger, and the scoring logic is opaque to credit officers, mitigated by per-decision reason codes in the underwriting screen. On the operations side, there is no fallback when the scoring service is down, mitigated by a documented manual scorecard with a degraded-mode SLA, and overrides are logged inconsistently, mitigated by mandatory override-reason capture with a quarterly review. The data-drift risk is the highest priority. The message is that no model reaches production until every high-severity risk has an owned, testable control.
Render it
Structured intent — create_slide payload
Copy-paste-ready. Renders deterministically (verbatim, no LLM). Add a theme_id for any of the built-in themes — the layout stays identical.
{
"form": "editorial_split",
"variant": "risk_control_split",
"headline": "Each AI deployment risk is matched to a named control before go-live",
"context": "Model risk register and the control response",
"takeaway": "No model reaches production until every high-severity risk has an owned, testable control.",
"data": {
"risks": [
{
"kind": "header",
"label": "Model"
},
{
"risk": "Training data drifts from the current applicant population",
"control": "Monthly population-stability checks with an auto-retrain trigger",
"severity": "High",
"emphasis": "primary"
},
{
"risk": "Scoring logic is opaque to credit officers",
"control": "Per-decision reason codes surfaced in the underwriting screen",
"severity": "High"
},
{
"kind": "header",
"label": "Operations"
},
{
"risk": "No fallback when the scoring service is unavailable",
"control": "Documented manual scorecard with an SLA for degraded mode",
"severity": "Medium"
},
{
"risk": "Overrides are logged inconsistently across branches",
"control": "Mandatory override-reason capture with a quarterly review",
"severity": "Medium"
}
],
"before_label": "Risk",
"after_label": "Control"
}
}cURL (REST)
curl -X POST https://api.slideforge.dev/v1/render/intent \
-H "Authorization: Bearer sf_live_..." \
-H "Content-Type: application/json" \
-d '{"form":"editorial_split","variant":"risk_control_split","headline":"Each AI deployment risk is matched to a named control before go-live","context":"Model risk register and the control response","takeaway":"No model reaches production until every high-severity risk has an owned, testable control.","data":{"risks":[{"kind":"header","label":"Model"},{"risk":"Training data drifts from the current applicant population","control":"Monthly population-stability checks with an auto-retrain trigger","severity":"High","emphasis":"primary"},{"risk":"Scoring logic is opaque to credit officers","control":"Per-decision reason codes surfaced in the underwriting screen","severity":"High"},{"kind":"header","label":"Operations"},{"risk":"No fallback when the scoring service is unavailable","control":"Documented manual scorecard with an SLA for degraded mode","severity":"Medium"},{"risk":"Overrides are logged inconsistently across branches","control":"Mandatory override-reason capture with a quarterly review","severity":"Medium"}],"before_label":"Risk","after_label":"Control"}}'Related templates
Agenda
Show two-day workshop schedule details
Bars vs target marker
Margin recovery is visible but three units are below target
Diverging variance bars
Cloud hosting and contractor spend offset hiring delay savings
Ranked magnitude bars
Three offer families drive 68% of FY25 growth
Frequently asked questions
How do I render the Risk and control mapping template?
Send a POST to /v1/render/auto with a brief, or call create_slide via MCP with the structured intent shown below. The response includes an editable .pptx, a PDF, and a PNG. $0.05 per slide.
Does an AI write the slide?
No. SlideForge renders deterministically from a structured intent — no LLM draws the slide, so text is never mangled or fabricated. The optional brief path uses a router to pick the pattern, then renders deterministically.
Can I customize the colors and fonts?
Yes. Pass a theme_id to apply a built-in or custom brand theme; the layout stays the same.
Try this template
Render Risk and control mapping deterministically — editable PowerPoint in under a second.