| --- |
| license: cc-by-4.0 |
| task_categories: |
| - text-generation |
| - question-answering |
| language: |
| - en |
| tags: |
| - sales |
| - salesforce |
| - hubspot |
| - gong |
| - agents |
| - mcp |
| - deterministic-evaluation |
| pretty_name: SalesBench-100 |
| size_categories: |
| - n<1K |
| configs: |
| - config_name: default |
| data_files: |
| - split: test |
| path: data/tasks.jsonl |
| --- |
| |
| # SalesBench-100 |
|
|
| SalesBench-100 is a synthetic long-horizon sales-agent benchmark with 100 original workflows across Salesforce, HubSpot, Gong, and a seeded evidence room. Each task begins with a high-level employee request and has its own authored causal rule and provider transition. Identity, operating facts, authority, governed policy, live-system indexes, and exceptions are separated so no mounted business asset publishes a selected option or precomputed change. Every task has 12 multi-row source assets. The evidence—not a fixed quota—determines 5–12 authorized CRM mutations and 56–91 calls, including an exact post-write readback for every mutation. |
|
|
| ## Public release |
|
|
| - Runnable Harbor world: <https://hub.harborframework.com/datasets/blobfishai/salesbench-100> |
| - Dataset and test assets: <https://huggingface.co/datasets/SamuelChien821/salesbench-100> |
| - Benchmark page: <https://blobfish.ai/benchmarks/salesbench-100> |
| - Builder and verifier source: <https://github.com/blobfishai/sales-agent-simulation> |
|
|
| ## Contents |
|
|
| - `data/tasks.jsonl`: portable task records with prompts, context paths, deterministic rubric metadata, and public gold output. |
| - `tasks/`: one readable JSON record per workflow. |
| - `task_files/`: 1,200 unique multi-record assets across Markdown, text, JSON, CSV, email, XML, and HTML. |
| - `world/`: the four-surface offline MCP world and verifier. |
| - `contracts/`: pinned contract metadata and exact published tool schemas. |
| - `trajectories/`: accepted tool traces produced by the qualification run. |
| - `reports/`: build, qualification, MCP-conformance, and container-probe evidence. |
|
|
| ## Objective release gates |
|
|
| | Gate | Required | |
| |---|---:| |
| | Tasks | 100 | |
| | Workflow families | 10 | |
| | Multi-record assets per task | 12 | |
| | Authorized mutations per task | 5–12; at least six distinct workload sizes | |
| | Reference MCP calls per task | 56–91 | |
| | Unique tool-name sequences | 100/100 | |
| | Unique semantic action graphs | 100/100 | |
| | Authored causal decision rules | 100/100 unique | |
| | Task-specific deterministic criteria | 301–420 | |
| | Precomputed answer objects in business evidence | 0 | |
| | Evidence and provider reads before mutation | required | |
| | Exact post-write readback per mutation | required | |
| | Verifier network/model/clock/random calls | 0 | |
| | Oracle and replay passes | 100/100 each | |
| | False accepts across ten negative controls | 0 | |
| | Reward for the pristine no-op control | exactly 0.0 | |
|
|
| Measured results live in `reports/qualification.json`. Reference traces are implementation proofs, not model scores. |
|
|
| ## Contract fidelity |
|
|
| The world keeps Salesforce, HubSpot, Gong, and filesystem as separate MCP endpoints. Names and input JSON Schemas are pinned to official hosted documentation or real open-source implementations at immutable commits recorded in `contracts/upstream-pins.json`. The Gong surface is read-only. Output envelopes mirror the corresponding MCP/API JSON shapes, with documented offline simplifications. |
|
|
| ## Data and licensing |
|
|
| All people, entities, records, messages, amounts, and commercial facts are synthetic. CRMArena, SCUBA, AutomationBench, Harvey Labs, and Apex Accounting informed quality and packaging analysis only; their task text and data are not included. Synthetic task data is CC-BY-4.0; generator/runtime code is Apache-2.0. Third-party contract facts retain their upstream terms. |
|
|
