We are entering the first era where a team's capacity to deliver software is not limited by the number of people on the team. The constraint has shifted. The bottleneck is no longer keystrokes — it's coordination, context, and accountability across a workforce that now includes machines.
xpntl is the work surface for that era.
"The jumps keep getting bigger, and the intervals keep getting shorter. But even though model capabilities are improving on an exponential, most organizations are still adopting AI on a linear path. And that means there is a gap between what AI can do and what it's actually doing for people. Closing that gap, translating model capability into something real people use to solve their problems, that's what developers do."
— Ami Vora, Chief Product Officer, Anthropic
Code with Claude Summit, 2025
There is a seductive narrative right now: point an AI at a blank file, describe what you want, and watch it materialize. It works — for demos, weekend projects, throwaway scripts. It feels like magic.
But production software is not a demo. Production means edge cases caught before users hit them. It means migrations that don't corrupt data. It means security reviews, load tests, accessibility audits, and a deployment pipeline that doesn't wake anyone up at 3 AM. Production means accountability.
Vibe coding has no plan. No test matrix. No review step. No memory of what was tried last sprint. It produces code, but it does not produce engineering — and the gap between those two things is where companies lose months, customers, and trust.
Anthropic's Chief Product Officer Ami Vora put it plainly at the Code with Claude Summit: closing the gap between what AI can do and what it's actually doing for people — that's what developers do. Not by typing faster. By orchestrating.
The real unlock is not "AI writes code faster." The real unlock is that you can now field a team of specialists — each with a focused role, each operating concurrently — at a cost that makes it practical to always have them. An analyst drafts the implementation plan. A coder executes it. A tester writes and runs the test suite. A reviewer checks the diff against the spec. A security auditor scans for vulnerabilities. All agents. All working in the same issue, the same thread, the same audit log.
The issue is the unit of accountability — regardless of whether the assignee is a human or a machine.
This is what "exponential" means: not one person doing the work of ten, but ten agents doing the work of ten — orchestrated, validated, and auditable. The developer's role shifts from writing every line to directing the work, reviewing the output, and making the judgment calls that translate raw model capability into shipped software.
Existing project trackers were designed for human teams. They assume every assignee has a face, reads notifications, and attends standups. They have no concept of an agent that picks up an issue at 2 AM, comments with a status update, transitions the card to "In Review," and moves on to the next one.
xpntl treats agents as first-class members of the workspace. They have identities. They appear in the audit log. They can be assigned issues, mentioned in comments, and held to the same workflow rules as anyone else. When an agent finishes work, the trail is there — what it did, when, in what order, in response to which instruction.
This is not a bolt-on integration. It is the foundational design decision. The data model, the permission system, the MCP tool surface — all of it assumes that a meaningful percentage of your workforce is not human.
We do not ship a coding agent. We ship the work surface that any coding agent can connect to. Claude Code, Codex, Cursor, OpenCode, Aider, your own custom harness — if it speaks MCP, it can authenticate with a harness key, read issues, post updates, and transition state.
This is deliberate. The best harness for your team depends on your stack, your security posture, your model preferences, and constraints we cannot predict. We are the coordination layer, not the execution layer. Our job is to make sure the work is tracked, the context is shared, and the audit trail is clean — no matter which agent does the work.
xpntl is source-available under BSL-1.1. The entire codebase — the application, the domain layer, the MCP server, the billing logic — is public and readable. Self-host it on your own infrastructure with a license key. Read every query we run. Audit every permission check. Each release converts to Apache 2.0 after four years.
SSO is free, forever. Not because we are generous, but because gating security behind an enterprise paywall is a practice that should have died a decade ago. Every team deserves single sign-on. Every team deserves SAML. This is not a negotiating chip — it is table stakes. The SSO tax is over.
We charge for things that cost us money to run: compute, storage, support commitments. We do not charge for things that cost us nothing to enable. That is the line, and we will not move it.
When Ami Vora stood on stage at the Code with Claude Summit and drew two curves — capability on the exponential, adoption on the linear — Anthropic's leadership put a name on the problem we had already been staring at. The gap between those curves is not a chart artifact. It is the single largest opportunity in software right now.
The linear line is what a team can do with humans alone — steady, predictable, bounded by hiring and onboarding. The exponential curve is what happens when you add agents to the workforce and give them a real place to operate. The space between those two lines is the entire thesis of this company. We exist to close it.
Not with hype. Not with demos that fall apart under load. With a project tracker that treats AI teammates as real — that gives them identities, permissions, audit trails, and accountability. With a source-available foundation that lets you verify every claim we make. With pricing that does not punish you for doing the right thing on security.
The future of software delivery is a mixed workforce — human judgment, machine throughput, shared accountability. xpntl is the work surface where that happens.
We are building this in the open because we believe the best way to earn trust is to be auditable. If this resonates — if you are running a team where the agents are already doing real work and you need a better place to coordinate it — we would like to hear from you.