Software products built around models, agents and AI-native workflows.
Showing 1–4 of 4 dossiers
Ada has added code tools that run a restricted Python subset inside agent conversations. They can transform API responses, perform deterministic calculations and call allowlisted domains, while MCP-authored changes can be staged and reviewed before promotion.
Fin’s new Evals and Releases features let teams test agent changes against simulated conversations before publishing, bundle configuration into a release, ramp traffic or A/B test it, and feed failures from live Monitors back into the next iteration.
Legora’s Agent Pro pricing illustrates a concrete AI SaaS shift: base platform economics can remain seat-oriented while high-variable-cost agent work is metered separately. The model is notable for its controls as much as its pricing—and for what it does not disclose publicly.
Reuters says Anthropic’s annual revenue run rate topped $65B by end-July, materially updating the growth picture behind its reported $190B–$200B 2028 forecast. A pre-IPO revolver is also expected to exceed roughly $10B. The figures remain sourced reporting rather than audited public results.
AI SaaS products inherit the usual product and distribution problems, then add model cost, variable quality, provider dependency and new expectations about automation. A compelling demo is only the beginning; retention depends on whether the product fits a repeated job and can deliver it reliably at a workable margin.
This page follows AI-native software businesses and major platform changes that affect them. BTN examines product design, pricing, defensibility, model choice and operational risk without assuming that adding AI creates a moat. Coverage is aimed at builders testing real opportunities: where new capability creates a useful product category, where economics remain awkward and where a conventional workflow with modest AI may be the stronger business.
The beat also watches how incumbent SaaS products bundle model features and how that affects smaller competitors. Distribution and proprietary workflow data can matter more than access to the newest model. Useful analysis separates the capability a provider can copy from the customer understanding a focused product can keep.