Build internally
- You gain
- Maximum product control
- You still own
- Own every connector, normalization rule, freshness check, model gate, approval flow, and source change.
Get startedTRAINING APPS / SCALABLE COACHING
Health and training intelligence
for your app.
Personalized coaching grounded in training plans, workout history, recovery, and health context. Your team keeps its programming, member experience, and control of every meaningful change.
FOR / PRODUCT + ENGINEERING TEAMSEVIDENCE / SYNTHETIC WORKING SAMPLEBOUNDARY / YOUR PRODUCT OWNS THE ACTION
01 / SEE IT IN A PRODUCT
Every capability.
From a natural request to a controlled result.
Choose a capability. Watch a natural request type, see Northstar respond, then review the exact product state before anything changes.
16 CAPABILITY DEMOSCAPABILITIES01 / 16
01 / 16
01 / 06 · Typing
02 / HEALTH + TRAINING + COACHING
Know the training.
Understand the person.
Give every member coaching that connects the intent of their program with what they have done and how they are recovering. Built for product and engineering teams at training businesses.
ONE INTELLIGENCE LAYER FOR YOUR APP
- 01Training intelligence
Understand the plan, completed workouts, and member constraints. Propose specific changes that preserve your programming intent.
- 02Health intelligence
Connect approved recovery and health context with source evidence, freshness, and visible limitations.
- 03Scalable coaching
Answer questions, remember useful context, and prepare reviewed actions inside the experience your members already use.
03 / THE KNOWNSET BOUNDARY
One controlled path.
Four explicit owners.
Your backend connects identity, training context, and approved health sources. Knownset returns insights and coaching proposals; your product handles member confirmation and final actions.
- 01
Source + permission
Approved context
Only context approved for this person and product enters the request, with provenance and freshness available to inspect.
- 02
Knownset engine
Deterministic fact
Code—not model prose—owns normalized values, calculations, and known limitations.
- 03
Approved purpose
Bounded interpretation
A coaching layer can explain supplied facts inside a defined purpose without becoming the source of record.
- 04
Customer + person
Confirmed action
The customer keeps programming and the member-facing review before a meaningful change is applied.
Design rule
A language model may explain a checked fact. It does not silently become the calculator, identity provider, permission service, or final writer.
Inspect the trust boundary04 / BUY OR BUILD
Choose the responsibility
you actually want to own.
Knownset is not the right choice for every team. It is for teams that want the member experience and programming to remain theirs without building the full context-to-action stack.
Aggregator + internal logic
- You gain
- Faster source access
- You still own
- Still own the meaning, evidence contract, decision logic, coaching boundary, and action workflow.
General-purpose model
- You gain
- Flexible language
- You still own
- Still need a deterministic source of fact, permission boundary, evidence, and safe path to action.
Standalone coaching app
- You gain
- A finished experience
- You still own
- Move the member relationship and programming outside the product your team already owns.
Knownset
- You keep
- Your product, programming, workflow, and member relationship.
- Knownset owns
- The infrastructure between consented source context and a governed action boundary.
05 / RESPONSIBILITY, NOT MAGIC
Your product stays
the owner.
Knownset is designed to fit behind a customer experience, not replace it. Each participant keeps a visible, reviewable role.
01Knownset owns
- Health and training context
- Validated workout proposals
- Evidence-backed insights
- Scalable coaching infrastructure
02Your product owns
- Member experience
- Programming policy
- Product-specific permissions
- Confirmation and final write
03The person owns
- Source approval
- Visible context
- Review of proposed changes
- Final decision
06 / START WITH EVIDENCE
Build your first workflow.
Grow with your members.
Explore the Coach and Health integration guide, then choose the support your team needs. Access is configured during paid guided onboarding; production activation follows readiness checks. The demonstrations use synthetic data.