Enterprise AI Code Testing.
tailtest's open-source foundation works at any scale: install the plugin per engineer, commit a team config, and the test cycle fires deterministically across every AI session. Enterprise-grade additions (centralized config, audit logging, SSO-aware install) are on the roadmap. We're talking with early adopters who need these now.
What's shipping today vs. on the roadmap
Today: the open-source plugin (1,234 tests across 4 AI coding hosts, MIT) covers per-edit test generation for any team size. No SaaS account, no per-seat cost.
On the roadmap (Q3-Q4 2026): centralized policy enforcement, audit log streaming, SSO-aware install, on-prem MCP server, compliance reporting evidence generation. If you need these now, get in touch, we are building this in collaboration with early adopters.
What works at enterprise scale today
The open-source plugin is genuinely production-grade. We dogfood it. Real OSS projects use it. The R1-R15 rule layer has caught 16 real bugs across 47 OSS Python repos in production runs.
- Per-engineer install across Claude Code / Cursor / Codex / Cline
- Commit-tracked
.tailtest/config.jsonfor team-uniform depth and runner settings - Monorepo support: per-package configuration, runner-per-language resolution
- Accepted-failure baselines: silence pre-existing failures without fixing them all on day one
- R12 failure classification: real_bug / test_bug / environment routing
- Adversarial mode for high-stakes modules
- No data leaves your machine; no SaaS account required
Enterprise gaps (working on these in 2026)
Honest framing: what the open-source plugin does NOT yet have:
- Centralized policy enforcement: today, depth and runner settings are committed to the repo. Enforcement is on engineer goodwill (or git pre-commit hook).
- Audit log streaming: tailtest sessions log to
.tailtest/session.jsonlocally. No central aggregation. - SSO-aware install: per-engineer install today. No org-level provisioning.
- Compliance reporting: the underlying data (scenarios run, failures classified, fixes applied) exists. Evidence packaging for SOC2/ISO/etc. doesn't yet ship.
- On-prem MCP server: the Cline variant's MCP server runs locally per engineer. An org-level on-prem mode is on the roadmap.
Why we're being public about the gaps
Most enterprise software pages over-promise. We'd rather you adopt the OSS plugin with clear-eyed expectations than try to sell you something we haven't built.
If the gaps above are blockers, talk to us. We're picking 2-3 early enterprise adopters in 2026 H2 to co-design the features with. The OSS foundation will stay free; enterprise features will likely be a separate offering. Open-source-then-enterprise is the model (think Sentry, GitLab).
Get in touch
Email Nikhil at AvanSaber for enterprise conversations.