Glasshouse enters beta.

Closed alpha was built to find out what a decision-maker actually does with a rehearsal, which is not always what the product assumed. Four months later, three things about Glasshouse are different, and one is not.

The pipeline got renamed by the people using it

Alpha shipped with the engineering names for the five stages: seeds, graph, personas, simulation, report. Partners never used them. They said "the cast" and "the reading" from the first week, so the product says it too. The stages do the same work they always did. They are now called what the room calls them.

1. 01

   Tell it

   You bring the decision and the documents behind it.

2. 02

   Scenario

   Glasshouse drafts the question the room will answer. Nothing runs until you approve it.

3. 03

   Cast

   The Library is polled, and the most interested profiles become personas with a job, a stake and a temper.

4. 04

   Rehearsal

   The cast reacts over timed rounds across up to six simulated platforms.

5. 05

   Reading

   A recommendation, its confidence and the numbers behind it. Every claim keeps its receipt. Diagnostics say how far to trust it.

The scenario became a gate

In alpha, the most expensive mistake was a run that answered a question nobody had asked. The drafted scenario is now something you approve before anything executes. You see the question, the cast it will be put to, and the platforms it will run across. Change the question and nothing has been spent.

The Interview came out of alpha

Partners kept doing the same thing with a finished report: picking one voice out of the crowd and arguing with it. That is a feature now. Ask a persona why it said what it said, then ask what would change its mind. It answers in character, and every answer cites the run it came from.

What beta includes

• An onboarding session with the engineer who will be on your account, and a scenario drafted with you rather than for you.
• A run allowance over the beta window, with custom persona libraries built for your market.
• Methodology disclosure on every run: calibration history, persona provenance, confidence-band inputs. Written for the security and audit reviews enterprise buyers actually run.
• A direct line into the roadmap. Alpha set the shape of beta; beta partners set the shape of what comes after.

What has not changed

The commitment underneath is the one the platform opened with. Outputs are post-cutoff. Personas are inspectable. Every claim carries the run it came from, and Diagnostics tell you where the run is weak rather than waiting for you to find out. Glasshouse will give you a confidence of 0.54 and say so. It will not round it up.

Nothing runs until you approve the question. Everything that runs keeps its receipt.

Beta is by application, and the gate is fit rather than seniority: decisions that are unambiguous, time-boxed and recordable. Apply from the contact page and reference "beta". We reply within two working days. How a run works, end to end, is on the Glasshouse page.