Bellwether Labs

Your customers already told you what to fix. Bellwether ships it, then checks it worked.

Bellwether reads your tickets, errors and analytics, finds the problem the most customers share, opens the pull request that fixes it, and measures the result.

1,284signals this week
9problems found
$184kat risk in the top one

Live demo, illustrative data. Click a sheep to inspect it, click the grass to lead the flock, or scrub through the season.

Week 41

Where signals graze. The tools you already pay for.

Connect in minutes. Personal data is redacted before anything is stored.

One loop, from complaint to measured fix. Then around again.

Analytics tools tell you something is wrong. Coding agents write code when asked. Bellwether connects the two, then checks the result.

  1. Ingest

    Tickets, errors, calls and reviews flow in. PII is redacted first.

  2. Cluster

    Everything about one problem becomes one sheep, tied to revenue.

  3. Spec

    The change, its acceptance criteria and the metric that will judge it.

  4. Code

    Your coding agent builds it in a sandbox, behind a flag, and opens a PR.

  5. Ship

    You review and merge. Guardrails roll back anything that regresses.

  6. Measure

    The result is written back and picks the next bellwether.

Bellwether runs itYour team signs off

The pens. Every problem has a place.

Problems move right as Bellwether and your team work on them. Pick a card to find its sheep in the field.

Every pull request arrives with its reasons. No evidence, no PR.

Each problem carries the errors, sessions and conversations that created it, with links back to the source.

The numbers are computed from stored evidence, not written by a model, so “12 accounts affected” is a query you can rerun.

  • Who it hurts and how much revenue is at risk
  • The metric that will judge the change
  • Guardrails that roll it back automatically
Stream CSV exports instead of loading every row#218 · opened by Bellwether · OPP-118
Affected
12 accounts, $184k ARR
Success metric
export_success_rate 62% → 95%
Guardrails
p95 API latency, error rate
Rollout
flag export_streaming, 1% → 100%
  • sentryTypeError in /export, 212 events in 7 days
  • zendesk“Export is greyed out for our big tables”
  • gong“We’d pay for a bigger plan if exports worked”
  • posthog31.4% of export_clicked never reach export_done
- const rows = await db.select(query)
- return toCSV(rows)
+ for await (const page of db.pages(query, 1000)) {
+   stream.write(toCSV(page))
+ }
Win+34
export_success_rate, 62% → 96% after merge. Dotted line is the 10% holdout.

You set the fence. Bellwether roams as far as you let it.

Autonomy is set per type of change, so copy fixes can ship on their own while new features still wait for a person.

Bellwether

Your team

Authentication, billing, data migrations and anything touching personal data need a person at every level, by default.

Three bets. Why Bellwether is different.

PostHog, Amplitude and LogRocket each shipped an agent that turns signals into PRs in 2026. Each one runs mostly on its own data.

Neutral

Works on top of the tools you already have, combining tickets, calls and reviews with events, errors and replays.

Bring your own agent

Claude Code, Codex, Cursor or Copilot writes the change. Bellwether decides what to build and checks that it worked.

Measured

No evidence, no problem. No success metric, no PR. Every shipped change ends with a verdict.

Shepherd’s log. Every loop ends with a verdict.

Wins, flat results and rollbacks are all written down, and the ranking learns from each one. Illustrative data.

  1. Week 40
    Billing page timeoutPR #204 · cache plan lookups
    p95 6.1s → 0.9sWin
  2. Week 39
    Onboarding checklist copyPR #199 · shorter step titles
    activation 41% → 42%Flat
  3. Week 38
    Bulk invite modalPR #192 · new layout
    errors +3.1%Rolled back
  4. Week 37
    Password reset email delayPR #188 · move to a queue
    tickets −64%Win

Join the first flock. Make the next thing you ship the thing they asked for.

We are working with a small group of design partners who have real users, real logs, and more feedback than time.

Become a design partnerhello@bellwetherlabs.dev