Read a deal's fit report

The Fit tab on a deal shows how the company compares with your thesis: the latest screening against your rubric, with every criterion, the value found, where it came from and why. It is an AI-assisted reading for your team to review. A screening never changes a deal's stage and never contacts the founder. A person at your fund decides.

When a screening runs#

• On its own, each time a deck version finishes its analysis. It uses your fund's default rubric, at its live version. See Build your thesis rubric.
• By hand, with Run screening (or Run again) on the Fit tab, by anyone who can edit the deal.

While it runs, the Fit tab shows Screening against your thesis with the steps Uploaded, Scanning, Extracting, Analysing and Screening. Updates on its own means you don't need to refresh. If a newer deck is still being read, the screening waits for it, then checks the rules against the facts on file. When it ends you see Screening finished. or The screening failed.

A deal that hasn't been screened shows Not screened yet, with Run screening if you can edit the deal.

The top line#

Above the readings you see:

• Fit report, labelled as an AI suggestion.
• The rubric and version used, such as Core seed rubric · v2. Hover over it for when it was screened, which deck version, the AI model and the prompt version. Click it to open the rubric.
• Screened and the date and time, in your fund's time zone.
• Includes overrides when your team has changed a criterion. See Override an assessment.
• Offline scoring when no AI model was available, so preferences were scored by simple keyword rules. Hard rules are unaffected.

The three readings#

The three readings answer different questions, so they are shown side by side and never blended into one score.

Under Thesis fit you also see the hard rules count, for example "Hard rules: 3 met · 1 mismatch · 1 unknown". A mismatch adds To review. Not part of the score and not a rejection.

Under Evidence quality, a line such as "2 disputed or unresolved claims." counts claims on the deal's Claims tab that are Disputed or Unresolved. When a claim was edited, only its latest statement counts, so an earlier wording that was disputed doesn't count against the deal twice. A claim your team dismissed doesn't count at all. The count is taken when the screening runs: after you settle claims, click Run again to update it. See The claim tracker.

Deck quality on the Brief tab is different again: it says how well the deck makes its case, not whether the company fits your thesis. See How deck analysis works.

Hard rules#

Each hard rule shows its name, the rule (such as "Rule: Seed" or "Rule: USD 1M to USD 4M") and a result:

Unknown always has a dashed outline and is never red, because a missing value is not a failure. Rules that aren't Met open automatically. Open any rule to see:

• Found: the value, its source (Deck, Founder correction, Your team or Application form) and a Slide 4 chip when it came from the deck.
• The reason, marked AI suggestion in lilac. Hard rules are checked in code, but every screening output is shown as a suggestion for your review.
• Exception applies or Exception may apply, with what was found or what to check.
• Becomes a question: for an Unknown rule.
• Any overrides, with who made them, when and why.

Preferences#

Each preference shows its share of the fit (such as "28% of the fit") and a score such as 7.5 / 10, or Not scored. Open it for the description, the model's rationale and the quotes it relied on, each with its slide. If a quote couldn't be found on its slide, it was dropped and the count is shown. If the deal had no deck text to read yet, you see "No deck text to read yet, so preferences weren't scored."

Slide citations#

Hover over a Slide 4 chip to preview the slide with the quote. Click it to open the slide on the Evidence tab. A chip marked older deck comes from an earlier deck version.

Questions for the founder#

Questions for the founder lists what to ask, each marked Unknown rule, Exception check or Rubric question.

To ask them, click Send as a request. Delsivo saves a draft request on the deal with one long-answer question per item, titled Questions from our first review, due in 14 days with reminders every 3 days, and opens the request composer on it. If the company has one founder on your portal, they are already picked. Nothing goes to the founder yet:

1. Edit, reorder or remove questions, add your own, and change the title, message, due date or reminders.
2. Check the Email preview tab.
3. Click Send to founder and confirm with Send request. Partners and Admins send; an Analyst clicks Save draft and a Partner sends it.

Clicking Send as a request again for the same screening opens the same draft, so you never get two. Close the composer without sending and the draft waits on the deal's Requests tab. The button is greyed out when there are no questions or when you can't edit the deal. See Send a request to a founder.

On the Brief tab#

The Brief tab repeats the readings as three cards next to Deck quality: Thesis fit (with the met, mismatch and unknown counts and the rubric version), Evidence quality and Diligence readiness, worded as on the Fit tab: 2 missing, or Nothing missing with "Every core item is on file." Click a card to open the Fit tab. Before the first screening the cards say Not yet, or Running while one runs.

Who sees and does what#

Other Analysts don't see the deal at all. A Delsivo support view can read the report but can't run or change anything. Founders never see screenings, scores or overrides.

Frequently asked questions#

Does a low fit or a mismatch move the deal to Passed?#

No. Nothing in a screening changes a stage or writes to a founder. Someone at your fund decides what happens next.

Why is thesis fit Not scored?#

The rubric has no preferences with a weight above 0, the deck had no readable text yet, or the model found nothing to score any preference on. The hard rules still run.

Which screening does the Brief show?#

The current one: the newest finished screening. Earlier runs stay in Run history on the Fit tab. See Rubric versions and re-runs.