Agent
Scoring
Scores every open deal as a sum of weighted terms, each weight carrying where it came from.
Approval
Last run
1 Sep 2026 ·
Reads
Can
The workflow
Top to bottom, exactly as it runs. Dry Run walks it against the last run's trace, a step at a time, writing nothing.
From the last run · 1 Sep 2026
What this agent actually produced, the action it is asking for, and the evidence each one rests on.
Cloudspan · Compression platform
scoreWaiting on youFound
One of the healthiest deals in the book. It is at Negotiation, and 7 of the last 8 deals that got this far were won.
Left alone it should progress: Deals at Negotiation typically close in about 3 weeks from here.
It is at Negotiation, and 7 of the last 8 deals that got this far were won. Nothing in the evidence contradicts it. Something last happened 2 days ago, and deals here usually move within 26. Aisha Clarke owns it.
Evidence
Nimbus Retail · Compression platform
scoreWaiting on youFound
One of the healthiest deals in the book. It is at Negotiation, and 7 of the last 8 deals that got this far were won.
Left alone it should progress: Deals at Negotiation typically close in about 3 weeks from here.
It is at Negotiation, and 7 of the last 8 deals that got this far were won. Nothing in the evidence contradicts it. Something last happened 32 days ago, and deals here usually move within 26. Tom Nguyen owns it.
Evidence
Homicide Energy Drink · Streaming CDN
scoreWaiting on youFound
One of the healthiest deals in the book. It is at Proposal, and 7 of the last 11 deals that got this far were won.
Left alone it should progress: Deals at Proposal typically close in about 7 weeks from here.
It is at Proposal, and 7 of the last 11 deals that got this far were won. Nothing in the evidence contradicts it. Something last happened 21 days ago, and deals here usually move within 26. Tom Nguyen owns it.
Next step
Set end_customer_company to create: Stallion Media Group
Homicide is the reseller; Stallion Media Group operates the platform and must be represented as the end customer.
Evidence
Kestrel Robotics · Compression platform
scoreWaiting on youFound
One of the healthiest deals in the book. It is at Security Review, and 2 of the last 3 deals that got this far were won.
Left alone it should progress: Deals at Security Review typically close in about 7 weeks from here.
It is at Security Review, and 2 of the last 3 deals that got this far were won. Nothing in the evidence contradicts it. Something last happened 45 days ago, and deals here usually move within 26. Marcus Webb owns it.
Evidence
Pinecone Logistics · Compression platform
scoreWaiting on youFound
One of the healthiest deals in the book. It is at Proposal, and 7 of the last 11 deals that got this far were won.
Left alone it should progress: Deals at Proposal typically close in about 7 weeks from here.
It is at Proposal, and 7 of the last 11 deals that got this far were won. Nothing in the evidence contradicts it. Something last happened 55 days ago, and deals here usually move within 26. Jared Patel owns it.
Evidence
How a deal gets its score
This agent rates every open deal for how winnable it looks. The rating is built from the things below, each worth what your profile says it is worth, so a score can always be taken apart. Nothing here is a judgement Arkiv keeps to itself.
What the score is made of
Every term, the weight the profile gives it, and where that weight came from. Change the profile and every score moves with it.
- stage_depth45%
- evidence_supports_stage25%
- days_since_last_activity20%
- owner_assigned10%
| Term | Weight | Favours | Provenance |
|---|---|---|---|
stage_depth How far the deal has actually got, by evidence rather than field. | 45% | Deals whose evidence puts them at Proposal or beyond. | The only term with a measured lift: 11 closed deals reached Proposal and 64% of them won, against 37% across all 19 closed deals. It carries the largest weight because it is the one term the history can price. |
evidence_supports_stage Whether the deal's own evidence agrees with the stage it sits in. | 25% | Deals with no Disputed finding against them. | Weighted second on the CRO's rule rather than on history: too few deals carry a contradicted stage to measure the lift. |
days_since_last_activity Silence, against how long a stage normally takes. | 20% | Deals that moved within 26 days. | Priced off the same 95 stage transitions as the stale threshold: median 23 days, three quarters inside 26. |
owner_assigned Whether a person who still works here owns the deal. | 10% | Deals with a current employee as owner. | A standard starting default, and unmeasurable here: one deal in the book has no live owner, which is a story but not a sample. |