Roadmap — Fractl Agents

The arc

Three horizons.
One thesis.

Where the platform is going, by quarter. Each horizon has a single thesis it's organized around — the work that lands in it ladders up to that bet.

Q2 2026 Active

Make every output trustworthy.

The bet: output isn't real if it can't be defended. This quarter, every saved deliverable carries lineage, sources, and a confidence score — and the eval pipeline closes the regression loop on Play behavior changes.

→Receipts on every deliverable — lineage + sources + confidence rendered in workspace

→Context-rich behavioral judge — eval pipeline upgrade for regression detection

→Deterministic workflow audit — every node round-tripped and verified

→Per-Play measurement layer — time-saved + margin attribution per run

Q3 2026 Scoped

Make the platform yours.

The bet: the catalog is a starting point, not a ceiling. Q3 brings the builder UI to skills and experts, opens the knowledge-wiki layer for compounding, and ships expert-scoped integrations so teams can codify their own methodology.

→Builder UI for skills + experts — full canvas pattern across all four primitives

→Knowledge wiki evolution — linting, incremental compile, compounding

→Expert scoping + integration controls — per-expert tool access + audit log

→Onboarding redesign — first-run picks the right Play from the brief

Q4 2026 Sequenced

Make outputs richer than markdown.

The bet: a saved deliverable should be more than a doc. Q4 explores generative-UI surfaces in chat, deeper research-tool integrations, and the next class of Plays that produce interactive artifacts, not just text.

→Generative-UI tool — in-chat interfaces for shaping outputs visually

→Research-tool integrations — BuzzSumo + 4 more, queued behind budget

→Interactive deliverables — Plays that produce dashboards, not docs

→External commercialization — design partners on the platform

Honest planning · themes are bets, not guarantees · order changes when the work tells us to → Subscribe to the changelog

The arc bends toward more trust, more leverage.

Every horizon makes Plays more defensible, more yours, more useful.

Run your first agent

Three horizons.One thesis.

Make every output trustworthy.

Make the platform yours.

Make outputs richer than markdown.

The arc bends toward more trust, more leverage.

Output is the only thingthat actually matters.

Three horizons.
One thesis.

Output is the only thing
that actually matters.