Blog
How the review engine works, what it costs to run, and what we learn building it — including the parts that did not go well.
- Pricing1 September 2026· 5 min read
Why we bill code review by pull request size, not per seat
Per-seat pricing charges a ten-person team the same whether they open two pull requests a month or two hundred. Here is why Pullora meters by the size of the diff instead, and what that costs us.
- Integrations1 September 2026· 4 min read
AI code review for Bitbucket, and why so few tools support it
Most AI code reviewers are GitHub-only. If your team is on Bitbucket Cloud, here is what is actually available, and how Pullora's Bitbucket support works.
- Integrations1 September 2026· 4 min read
AI code review for Azure DevOps
Azure DevOps is where enterprise teams live and where AI code review tools mostly aren't. How Pullora connects, what access it needs, and why it stores a token when GitHub doesn't.
- Engineering1 September 2026· 6 min read
How we stop the AI inventing code review findings
Every AI review tool claims it doesn't hallucinate. Here is the actual mechanism Pullora uses — evidence grounding, diff anchoring and confidence gating — and what it discards on a real pull request.
- Building Pullora1 September 2026· 7 min read
35 signups, zero activations: what our first Google Ads campaign taught us
We ran paid search for a week and got 35 signups, none of whom did anything. The database and Google Analytics disagreed about why. Here is the full data.
- Engineering1 September 2026· 5 min read
Your code review policy belongs in your repository, not a dashboard
Dashboard settings drift, are invisible in code review, and nobody knows who changed them. Here is why Pullora's review policy is a committed file — and the guardrails that stop it weakening security.
- Practice1 September 2026· 4 min read
Read before you post: the case for dry-run AI code review
The fastest way to make a team hate an AI reviewer is to let it comment on their pull requests before anyone has checked what it says. Dry-run mode exists for the first two weeks.
- Pricing1 September 2026· 5 min read
The economics of AI code review: what a review actually costs
Real numbers from running Pullora: what a small review costs, what a large deep review costs, and why the difference makes flat pricing awkward for everyone selling it.
- Security1 September 2026· 5 min read
Prompt injection when your input is other people's code
An AI code reviewer reads untrusted text by definition — pull request descriptions, comments, and code written by anyone who can open a PR. How Pullora treats repository content as data, never instructions.
- Engineering1 September 2026· 5 min read
Why we don't index your whole codebase
Some AI reviewers build a persistent graph of your entire repository. Pullora builds context per review instead. That is a real trade-off with real losses, and here is the honest accounting.
- Engineering1 September 2026· 5 min read
What we measure when we say a review is good
Bug catch rate is the number everyone quotes and the easiest to game. Here are the four numbers Pullora actually tracks, including the two that make us look worse.
- Engineering1 September 2026· 4 min read
Re-reviewing a pull request should not mean reviewing it again
Push three commits to an open PR and a naive reviewer comments on the same code three times. How incremental review works, when it doesn't apply, and why re-running an unchanged PR costs full price.