F&BIntel 1.5B Model Performance

Food-service feedback intelligence, measured for buyer evaluation

F&BIntel turns food-service feedback into structured operational intelligence across themes, risk, customer intent, sentiment, confidence, and review-ready JSON output.

Validated Output

100%

JSON Validity

Evaluation outputs are valid JSON in the tested snapshot.

Schema Discipline

100%

Schema Validity

Outputs conform to the expected F&BIntel evaluation schema.

Field-Level Agreement

93.75%

Average Partial Match

Strong field-level agreement across the current evaluation sample.

92.50%

Primary Theme

Strong92.50%

85.00%

Secondary Theme

Good85.00%

90.00%

Risk Level

Strong90.00%

67.50%

Full Exact Match

Precision focus67.50%

Current F&BIntel buyer-evaluation snapshot - Structured output - Controlled pilot access path

Buyer Pilot Evaluation

F&BIntel gives buyers a clear view of food-service intelligence performance.

F&BIntel is structured around the same CXIntel access system used across product layers: public demo examples, controlled buyer pilot, and private toolkit direction.

01

Structured Output

JSON-first behavior supports review, validation, local workflows, and future API packaging.

02

Food-Service Taxonomy

Outputs are aligned with F&B operations instead of generic review sentiment only.

03

Decision-ready Confidence

Confidence labels help teams prioritize which food-service issues need faster review.

04

Buyer Evaluation Path

F&BIntel follows a controlled buyer evaluation path before full private toolkit access.

Core Intelligence

Food-service classification fields.

Core F&BIntel outputs focus on turning restaurant, QSR, cafe, delivery, and order feedback into structured operational fields.

92.50%

Primary Operational Theme

Identifies the main food-service operational issue in a review or customer comment.

85.00%

Secondary Operational Theme

Preserves secondary signals such as food quality, packaging, service, or order accuracy.

90.00%

Risk Level

Detects operational severity for escalation, review, or management attention.

Evaluation-ready

Intent Type

Separates complaints, feedback, requests, and review-style customer comments.

Evaluation-ready

Sentiment Label

Captures positive, negative, and mixed food-service sentiment.

Reviewed

Confidence

Confidence label behavior is being reviewed before final public benchmark claims.

Pilot Evaluation

Controlled pilot before full package access.

F&BIntel follows the same systematic CXIntel buyer path as TelIntel, giving users a structured way to evaluate the model before full access.

Active

Evaluation Demo

Structured F&BIntel examples are available through the unified CXIntel demo page.

2-day buyer session

Pilot Evaluation

Buyers can evaluate the layer through a controlled 2-day pilot path before full package access.

Buyer path

Private Toolkit

Evaluation can lead into private toolkit discussion using the same CXIntel structure as TelIntel.

Buyer Evaluation Notes

Food-service intelligence, structured for pilot evaluation.

01

F&BIntel converts restaurant, cafe, QSR, delivery, and food-service feedback into structured operational intelligence.

02

Buyer evaluation follows the CXIntel pilot system: a controlled 2-day session before full package access.

03

The model highlights operational themes, service issues, food-quality signals, risk level, customer intent, sentiment, and confidence.

04

F&BIntel is designed for private toolkit deployment after evaluation, keeping the same CXIntel access structure used across product layers.

Model Positioning

F&BIntel buyer evaluation track.

The F&BIntel performance section shows how food-service feedback becomes structured operational intelligence that can support review, triage, and private toolkit evaluation.

Current status

Structured evaluation is active through the CXIntel buyer path: public examples, controlled pilot access, and private toolkit discussion.

Evaluation Review

Ready to try F&BIntel through a controlled buyer pilot?

Request access to evaluate F&BIntel through the CXIntel pilot path and review how it handles real food-service feedback.