AI Features Overview
AI runs through all of Celeredge, but it follows one rule everywhere: outputs are grounded in real evidence and cited back to the source. The AI doesn't replace the consultant's judgment — it does the retrieval, drafting, and pattern-finding, and shows its work so you can verify it.
The AI surfaces
| Feature | What it does | Where |
|---|---|---|
| Autopilot baseline | AI-generated baseline analysis of the engagement, with a pre-flight readiness check | Analysis |
| Ask the Data | Ask questions in natural language over the engagement's data; get grounded answers with sources | Analysis |
| AI Interviewer | Conversational AI that interviews stakeholders to capture tacit knowledge as citable evidence | AI Interviewer |
| Signals | AI-detected observations from the engagement's data, each with evidence and an assignable action | Signals & Actions |
| Causal & impact | Causal-intelligence analysis runs | Analysis |
| Notebook | Analytical cells that produce insights for Execute | Analysis |
| Assessments | Framework assessments grounded in evidence | Running Assessments |
| Framework authoring (Compile) | AI generates the evaluation prompts + retrieval queries that let an authored framework run (Enterprise) | Framework Authoring |
| Reports & decks | Generated, evidence-backed deliverables | Reporting · Deck Studio |
The common thread: grounding
Every one of these is built on the same evidence-and-citation spine described in Citations & Trust:
The Analysis workspace
Most AI tools for an engagement live under Analysis, which offers four tools over the engagement's data:
- AI baseline — autopilot analysis, gated by a pre-flight readiness check.
- Causal & impact — causal-intelligence runs.
- Notebook — analytical cells that feed Execute insights.
- Ask the data — knowledge-graph chat with grounded answers, a financial-KPI snapshot, and report generation.
What to expect with AI jobs
Some AI work runs in the background (assessments, deck generation, causal runs). You'll see live progress and statuses rather than a frozen screen. Larger jobs take longer — start them and come back; the engagement keeps their results.
Related
- Citations & Trust — how to trust any AI output
- Autopilot baseline
- Ask the Data
- AI Interviewer
- Signals
- Framework Authoring — AI-compiled evaluation prompts