HarnessHive

HarnessHive is an AI-native, collaborative, browser-based and versioned design ecosystem for wiring harness engineers. It is a next-generation wiring harness CAD software built to fit inside pre-existing workflows while enabling real-time collaboration in the cloud. Its central idea is a single live document that holds schematic, wiring, bill of materials and a to-scale formboard view, with an AI assistant available at every step to help engineers find parts, place them and connect cavities. The product is aimed at the people who actually build harnesses: startups and machine builders, student teams, and harness shops and freelancers. Traditionally, harness design work is split across separate tools and documents: a schematic in one place, physical routing in another, a spreadsheet for the BOM and cut list, and a separate manufacturing drawing for the build board. Keeping those artefacts consistent is manual work, and a change in one rarely propagates to the others automatically. HarnessHive addresses this by making all four views part of the same live document, so routes, cut lengths, the BOM and a 1:1 formboard all derive from one model. The ecosystem is also designed for integration within pre-existing workflows and for the reality that harnesses are rarely built by one person: they are shared between engineers, handed over between seasons, and sent to build benches for manufacturing. Schematic and wiring views form the two halves of the model. The schematic view is the logical netlist: engineers place components and splices, connect their pins, and put the conductor on the connection itself. The wiring view represents the physical side of the same harness, with connectors, branch points and the routes between them, so a team can decide where the bundle splits and add coverings to a route. Because these are two views of one document rather than two files, editing one and having the others follow removes the manual rework of synchronising logic and physical layout. The BOM and cut list view holds parts and cut lengths that are derived live from the model. When the harness is fixed, the numbers follow, and the workbook can be exported to XLSX. The formboard view is the manufacturing drawing at true 1:1 scale, kept in sync with the wiring, and can be exported as a full-size PDF for the build board. Together these two views mean the manufacturing output is generated from the design rather than rebuilt by hand, and the shop receives a document that matches the model. Collaboration is built into the document rather than bolted on. Everyone edits the same live harness, seeing cursors, drags and changes as they happen. Undo only reverts your own work, so one engineer's correction does not roll back a colleague's progress, and every action lands in a history that anyone can open in read-only mode. Access is managed through Viewer, Editor and Admin roles, and history is organised by day with read-only snapshots, giving teams a record of what changed and when. The AI assistant reads the harness itself. It can be asked to find real parts from trusted supplier sources, add them to your library, place them and connect the cavities. It also reads CSV wire lists you attach, creates parts from supplier pages, and can generate a whole harness as a single proposal. The safety model is explicit: every edit arrives as a proposal, and nothing changes until you apply it. Applied changes then show up in the history alongside manual edits, so AI-assisted work is traceable in the same way as human work. Integrations are designed to sit inside an existing toolchain. Teams can bring wire lists, supplier data and whole harnesses into HarnessHive, and send full-scale drawings, BOMs and a complete JSON back out. Concretely, the formboard exports as a PDF at true 1:1 scale, the BOM and cut list export as XLSX, and the harness JSON supports round-trip use. KiCad and Onshape importers are described as coming soon, which reflects the goal of integration within pre-existing workflows. Parts are treated as real, versioned objects. Connector housings, terminals, conductors, seals and splices live in organisation libraries and local parts, with cavity counts defined on the part definition. Every occurrence in a harness pins the exact version it was built with, so updating a library never silently changes an existing design, and BOM rows follow the model. This matters because harness designs are long-lived: version pinning keeps a released harness reproducible even as the parts library evolves. HarnessHive also ships an MCP server so teams can bring their own agent. Claude, ChatGPT, Cursor or a custom agent can be pointed at the workspace to read harnesses, BOMs, empty cavities and libraries. Every token gets eight read tools, and writes sit behind an explicit opt-in. Access is scoped to a single workspace and a token can be revoked instantly. The site lists Claude, ChatGPT, Gemini, Cursor and Muse as tools it works with, positioning the MCP server as a way to expose harness data to the AI assistants engineers already use. For the teams using it, the outcomes described are practical. Startups and machine builders can ship their first harness without buying a seat of legacy EDA. Student teams can hand the harness over each season with its full history intact. Harness shops and freelancers can send a 1:1 formboard and a clean BOM straight to the build bench. Because the design lives in the browser, there is nothing to install, and the free trial requires no credit card. In practice, a harness workflow in HarnessHive starts with a design: components and splices with connected pins in the schematic, then connectors, branch points and routes in the wiring view, including where the bundle splits and which routes carry coverings. A CSV wire list can be attached for the AI assistant to read, supplier pages can be turned into parts in the library, and the assistant can propose an entire harness that a user reviews before applying. The resulting BOM and cut list are exported to XLSX, and the formboard is exported as a full-size PDF. Alongside the visual editor, an MCP-connected agent can read harnesses, BOMs, empty cavities and libraries for further automation or analysis. Typical users include startups and machine builders, Formula Student and other student teams, and harness shops and freelancers. Onboarding is deliberately lightweight: a 25-minute demo can be booked with the team, and a free trial gives 14 days of access with no credit card and nothing to install. The product runs in the browser, with sign-up handled at app.harnesshive.com. Community is part of the product's development model, with a Discord server plus accounts on X, LinkedIn and Instagram, and a newsletter for new features and blog posts. HarnessHive states that it is a small team building with the people who use it, and sponsors Formula Student teams across the globe. HarnessHive's core promise is simple: one harness, four views, one live document. Schematic, wiring, BOM and cut list, and a true 1:1 formboard all stay in sync, collaboration and history are part of the document, real parts are versioned, and an AI assistant proposes changes that a human applies. For wiring harness engineers who want AI-native, browser-based, collaborative and versioned design without necessarily buying a seat of legacy EDA, HarnessHive is positioned as the ecosystem that ties logic, physical routing, manufacturing data and AI assistance together in one place.