We work alongside your implementation team to turn one policy-driven human decision into a governed agent workflow. The engagement is not a demo or a sandbox proof-of-concept. It is a scoped working session where we model one production-relevant decision, run it against real cases using the safest appropriate adoption pattern (typically shadow, gate or selective routing, chosen at scoping), and measure the outcome against criteria we agree together.
If the criteria are met, you convert to a founding-customer contract and we extend to the next decision. If not, clean exit, no ongoing commitment.
Knowledge is already operational. At this stage, the first cohort of design partners shapes which policy packs, defaults and operator workflows we productize next. Founding-partner status is a consequence of the engagement, not its main reason. The main reason is that one decision costs you something today.
What a design partner receives
| What you get | Detail |
|---|---|
| Founding-partner protection | Founding commercial terms remain exclusive to the initial design-partner cohort and are not offered as standard commercial terms after |
| Direct product influence | Your real-world use case directly informs the packs, defaults and product improvements we prioritize during the engagement |
| Founding-customer pricing | Reduced price for the first twelve months of production, locked at signing |
| Named as a founding customer | Logo on the site (optional, opt-in) and a joint case study once the success criteria are met |
| Direct founder access | Weekly working session with the founding team, plus a direct async channel throughout the engagement |
| Pack development | If your decision is not covered by an existing pack, we model it with your domain team. Your proprietary policies and configurations remain yours. Reusable generic patterns may inform future Knowledge packs |
What we ask in return
| What we need | Detail |
|---|---|
| A real, production-relevant decision | One decision that is escalated, mis-decided or slow today, evaluated against real cases |
| A named business champion and a named tech champion | The business champion owns the process and the decision — typically a Business or Operations lead, sometimes a head of AI product when the decision is embedded in a new agent flow. The tech champion is the AI / Automation implementation lead who plumbs Knowledge into the caller. Compliance & Risk and Security & Platform participate when the decision touches their surface, but they are not the primary champions. One working meeting per week for eight weeks |
| Honest feedback | You tell us what breaks, what is confusing, what is missing. We ship fixes weekly during the engagement |
| An eight-week commitment | Long enough for the metrics to accumulate meaningfully. Clean exit at the end if we have not hit the criteria agreed upfront |
| Optional but valued | Willingness to be quoted or referenced once the success criteria are met |
How the engagement runs
1. Scoping conversation. Thirty minutes. We understand which decision hurts most, whether it fits Knowledge's shape, and which adoption pattern is safest for the deployment. Two discovery questions we tend to ask early :
- For a customer-facing journey — what's your current completion rate, and where do most abandons happen?
- For an internal approval flow — how many requests does your team review every month, and what percentage are ultimately approved without requiring real judgment?
Answers to those two shape the success criteria we set for the engagement.
2. Scoped proposal. Within a week, we send a written scope: the decision, the model, the adoption pattern, the success criteria, the timeline, the pricing. You sign or you decline. No drift.
3. Kick-off (week 1). We deploy the agreed Knowledge setup and integrate it with the selected caller, in the adoption mode chosen at scoping.
4. Weekly working session (weeks 2-8). Walk through discrepancies, adjust rules, ship fixes to Knowledge itself if the engagement reveals gaps.
5. Decision at week 8. Did we hit the criteria agreed at scoping? If yes, you convert to the founding-customer contract and we extend to the next decision. If not, clean exit, no ongoing commitment.
Success criteria we agree upfront
Before week 1, we agree on the numbers that would justify converting. Typical criteria:
| Criterion | What it measures |
|---|---|
| Decision agreement | Knowledge and your existing system agree on X% of cases. Discrepancies traced to (a) Knowledge missing a rule, (b) legacy bug, or (c) legitimate ambiguity |
| Manual review reduction | For cases Knowledge marks as complete deterministic verdicts, what % could bypass current manual review? |
| Time to ship a rule | From "compliance asks for a new rule" to "rule live in Knowledge", compared with the same delay in your existing system |
| Collection efficiency | Fields requested, follow-up requests and completion rate on Knowledge's progressive resolution, compared with the current onboarding or intake journey |
| Audit reconstruction time | Time to retrieve the context, applicable rules and policy state behind a historical decision, compared with your current process |
| Enforcement coverage | For engagements where a Policy Enforcement Point is deployed (SDK decorator, MCP interceptor, custom wrapper) : share of automated actions that require a valid signed verdict before executing |
Bring your internal AI team or implementation partner
Knowledge does not build the agent for you. The engagement assumes an implementation team — either your internal AI / automation team, or an implementation partner (SI, agent vendor, boutique consultancy) already engaged on the project. We work alongside them to turn one policy-driven human decision into a governed agent workflow.
If you don't yet have an implementation team lined up, we can introduce partners we have already delivered alongside. That is a discussion, not a marketplace ; the intent is to make sure the first engagement has the right hands, not to route you through a channel.
See Build rule-governed agents if you are the implementation partner.
The engagement in one picture
Your team + SMEs
|
v
formalize the rules
|
v
Knowledge
|
v
historical replay
|
v
shadow deployment
|
v
measure
|
v
production if the numbers land
Why only three
Hands-on engagements with direct founder and product-team involvement. We limit the first cohort to three firms so each partner can materially influence what gets productized, and so we can ship weekly fixes that respond to what the engagements reveal. Fit is determined by the decision and the problem, not the industry label.