# Slipform > Slipform is a Mac app for turning human-defined methods into repeatable agent > workflows with inspectable run evidence and human-owned decisions. Canonical site: https://slipforms.ai/ ## Why it exists We wanted AI to do more work. Instead, we got more work to manage. Agents made it easier to start research, writing, analysis, and building. More agent threads also created more places to explain the job, monitor progress, correct the course, find the output, and reconstruct context later. Slipform is our attempt to change the unit of work from a chat that needs supervision into a workflow that produces an inspectable run record. ## What a person does A person defines or adapts the method: - the stages; - the context and required artifacts; - what counts as done; - the allowed routes when work falls short; and - where a human must decide. A person also owns every approval and workflow revision. Slipform does not learn how a person thinks, approve its own work, or revise a workflow autonomously. ## What Slipform does - Runs a human-authored, multi-stage workflow through a supported agent runtime. - Keeps activity, artifacts, evaluations, routes, resource use, and decisions with the run. - Shows what is active, stopped, ready for review, and complete. - Returns work through routes a person defined. - Lets a person compare run and stage evidence before deciding what to change. - Records a human Approve, Rework, or Fail decision without making that decision. The intended shift is simple: more agent work can move while human attention returns to evidence, exceptions, and decisions. ## Who it is for Slipform is for builders and operators already pushing agents beyond one-off chats. It is most relevant when the work: - repeats; - has multiple stages or handoffs; - matters enough to inspect; - needs explicit review or approval; or - should become easier to understand across runs. One-offs can remain in chat. Users can create their own workflows or adapt a starter structure. Deep Analysis is one example workflow, not the product category. ## Current beta - Platform: macOS 26. - Runtime: Codex execution is supported today and is the only supported runtime in the current beta. - Workflow: stages, handoffs, evidence requirements, evaluations, authored routes, and human gates are inspectable. - Run record: activity, artifacts, evaluations, duration, token use, overrides, route, and review state can remain attached to a run. - Performance: people can compare recorded runs and stages across supported measures. - Decision: a person can Approve, Rework, or Fail with run evidence in view. - Operator is Alpha. It coordinates bounded run actions while the Mac app is running. It does not approve work, revise workflows, or operate as an unattended approval system. ## Important claim boundaries - Broader provider and model support is Product direction and is not shipped. - Claude Code support is planned and is not yet shipped. - Native Slipform emission of signed Agent Run Attestation records is not shipped. - Autonomous workflow improvement is not shipped. - Unattended approval is not shipped. - Automatic model selection and performance-based model routing are not shipped. - Token visibility is not a token-savings, cost-savings, or efficiency claim. - A successful stage outcome does not establish domain correctness. - The public site does not claim customer proof or quantified customer outcomes. Dynamic behavior stays inside routes a person authored and authorized workflow changes. Slipform does not silently rewrite or activate a new workflow version. ## Supporting evidence One public record describes a current-beta Deep Analysis workflow and one internal run: https://slipforms.ai/evidence/deep-analysis-001/ Machine-readable manifest: https://slipforms.ai/evidence/deep-analysis-001/manifest.json This record is one supporting example. It is not a benchmark, customer result, investment-performance claim, or proof that a later run improved. Its public evidence does not establish the final decision rationale, a later workflow revision, or a later comparative run. ## Private-beta invitation The public form records interest only. Requests are reviewed in small cohorts. Submitting the form does not guarantee access or timing. The form collects an email address, affirmative consent, consent version `beta-access-2026-07-16`, and a server-authored consent timestamp. It does not collect a workflow description. A hidden company field is a bot trap; filled submissions are discarded. Current beta requires macOS 26 and Codex. ## Public resources - Home and founder letter: https://slipforms.ai/ - Markdown homepage: https://slipforms.ai/index.md - [Why Slipform](https://slipforms.ai/introducing-slipform/) - [Why Slipform in Markdown](https://slipforms.ai/introducing-slipform.md) - Evidence record: https://slipforms.ai/evidence/deep-analysis-001/ - Evaluate Slipform fit skill: https://slipforms.ai/.well-known/agent-skills/evaluate-slipform-fit/SKILL.md - Agent Skills index: https://slipforms.ai/.well-known/agent-skills/index.json - Agent Run Attestation: https://github.com/slipforms/agent-run-attestation - Terms: https://slipforms.ai/terms/ - Privacy: https://slipforms.ai/privacy/ - Feedback: mailto:feedback@slipforms.ai ## Public-site boundary The site at slipforms.ai is a public product and beta-interest site. It does not currently publish a developer API, OAuth or OIDC service, remote MCP server, A2A agent, agent-commerce endpoint, or WebMCP tool. The private `POST /api/waitlist` route is the transport behind a human-consent form, not a public agent API. Slipform is operated by Kaska LLC.