A WorkBoard alternative where agent access is not a licence tier

Enterprise reach and rollout muscle. WorkBoard is built for organizations where thousands of people update objectives, and it shows: around twenty native integrations (Slack, Teams, Jira, Azure DevOps, Salesforce, ServiceNow, Snowflake, Tableau and more), and a customer list of household-name enterprises.

A real developer surface, too. Its developer page documents a RESTful API with OAuth 2.0 and instant-token authentication, and application registration from the user dashboard. If you need to pipe objective data into a corporate warehouse, that path exists and is documented.

If your evaluation criteria are "will this survive a 5,000-seat rollout and satisfy our procurement process", WorkBoard is a stronger answer than we are. We are a small product with a public free tier, not an enterprise suite with a customer-success organization behind it.

Both products now assume AI belongs in strategy work. They monetize that assumption in opposite directions.

WorkBoard positions agent access as an add-on above the platform subscription — a WorkBoardAI subscription plus Agent-Tier API licences or an AI credit allocation. That is a coherent enterprise model: the AI is a premium capability, metered like one.

We took the other route because we think the agent is the interface, not a feature of it. The MCP server is the primary way to use hoshin.app. It is open source under MIT, published on npm and in the official Model Context Protocol registry, and it works on the free tier. Your Claude, your Cursor, your consultant's agent — each with its own token, its own rights, every action audited. Nothing meters the agent because nothing needs to: it is how the product is meant to be driven.

The practical consequence at evaluation time: you can point an agent at our public read-only demo matrix and watch it produce a completeness verdict before you talk to anyone, or you can install the sovereign local mode and never send us anything at all.

• WorkBoard: documented REST API; agent access requires a WorkBoardAI subscription + Agent-Tier API licences (their developer page, July 2026).
• hoshin.app: MCP server as the primary interface, MIT-licensed, official registry listing, available on the free tier.
• Both: OAuth 2.0 / 2.1, per-user rights. Ours adds an audit trail on every privileged action and a role cap — a token can never grant more than its owner has.

WorkBoard's pricing page does not display tiers or numbers; it invites you to book a demo (consulted July 2026). Third-party directories quote figures around $9 per user per month, but since the vendor does not publish them we treat those as unverified.

We publish a free tier, and more importantly a public read-only demo: a real scored X-Matrix you can query from your own AI client with a token we publish openly in our llms.txt. No form, no call, no card. That is not generosity — it is the only honest way to sell an agent-native product. If the claim is "your assistant can audit a strategy", the buyer should be able to verify it with their own assistant in five minutes.

WorkBoard models objectives, key results and alignment across a hierarchy. hoshin.app models the X-Matrix: a 3-to-5-year vision, annual objectives, initiatives and KPIs, plus the correlations between adjacent quadrants — and it scores that structure out of 100 across five metrics.

That score is the substantive difference. Coverage, correlation density, orphan elements, complete vision-to-KPI chains, and the balance between leading and lagging indicators. It will tell you that an initiative is connected to nothing, that a vision never reaches a measurement, or that an objective is tracked only by lagging indicators — by name, every week, whether or not anyone feels like hearing it. Dashboards display what you typed; the score renders a verdict on it.

• Coverage 15% · Density 20% · Orphans 20% · Chains 25% · Indicator balance 20%.
• The specification is public and the engine is MIT-licensed — you can audit how your own strategy is judged.
• Exposed as an MCP tool, so the verdict lands in the conversation where decisions are being made, not in a tab nobody opens.

This is the axis on which no enterprise SaaS platform can meet us, and it decides some deals outright. Consulting firms, law firms, family offices and anyone whose business is built on confidential method refuse — on architectural principle, not as a negotiating position — to keep strategy permanently on a third party's servers.

hoshin.app ships a sovereign local mode: the same tools, the same score, against plain JSON files in a directory you chose on your own machine. No account, no network, no telemetry. The files are the export; deletion is `rm`. The one thing that mode refuses is publishing a public share link, because publishing needs a server and standing one up would quietly undo the guarantee.

Stay with WorkBoard if you are deploying OKRs at several thousand seats, if your procurement requires an enterprise vendor with a named success team, if you need its breadth of native connectors into Salesforce, ServiceNow or Snowflake, or if metered agent licences are a model your finance team actively prefers because it makes AI spend legible.

Come to us if the X-Matrix is the object you actually steer with, if you want the agent to be the interface rather than an upsell, if you want to verify the product before speaking to a human, or if your data must stay on your own machine. Those are different products for different bets — ours is that agents will do the maintenance work of strategy deployment, and that the tool should be shaped around that from the first line of code.