01 / 04
Before a change
Technical alpha focus
Max finds the systems, data, owners, policies, prior incidents, tests, approvals, and unknowns that matter before implementation begins.
PRIVATE ALPHA · FOUNDING DESIGN PARTNERS
AI can make the change. THUS knows what follows.
THUS is building AI teammates that learn and grow with your business, predict how proposed changes affect your people and systems, and stay with the work through execution, incidents, and verified outcomes.
Starting with Max, a company-aware AI engineer in private technical alpha for engineering and data teams.
Starting read-only. One workflow. Measured against your current process.
CONSEQUENCE MAP · SESSION POLICYIllustrative
Proposed change
Increase session lifetime from 24 hours to 7 days
AFFECTED SYSTEMS
- Auth ServiceDirect impact · High confidenceFact
- Partner APIToken consumer · High confidencePR-418Fact
- PaymentsHistorical coupling · Medium confidenceInference
Payments is a historical inference, not a confirmed dependency.
EVIDENCE + REVIEW
OwnerIdentity Platform
- ADR-12Conflicts with 24-hour policyADR-12
- INC-88Prior long-session incidentINC-88
- SecurityRequired reviewerRequired
- Token revocation testRequired evidenceRequired
STILL UNKNOWN
External partner caching
External partner caching behavior has not been verified.
UnknownMAX RECOMMENDATION
Proceed after Security review and token-revocation evidence.
Payments remains an inferred dependency and should be verified.
Illustrative product view · THUS is currently in technical alpha
THE GAP
A change is never just a change.
The code may be local. The consequences are not. Important context is scattered across repositories, tickets, documents, incidents, dashboards, policies, and the memory of experienced employees. Coding agents can implement quickly, but they rarely understand the organization surrounding the task.
Code
Runtime
Ownership
Policy
History
Outcomes
Evidence trail
Implementation is getting cheaper. Confidence is not.
THE FIRST THUS TEAMMATE
Meet Max. A company-aware AI engineer.
Max works alongside software and data teams and the coding agents they already use. It learns how the company actually operates while helping the team investigate, review, and complete consequential work.
02 / 04
Companion + local coding agent
In development
The Companion recognizes what the engineer is currently doing across the IDE, terminal, ticket, dashboard, or incident view. It retrieves the relevant company knowledge, consequences, constraints, owners, tests, and approved scope from THUS Core, then supplies that context to the local coding agent—whether Cursor, Claude Code, Codex, or another tool. As the agent writes code, Max follows the actual diff, flags hidden impact or meaningful scope expansion, helps the agent correct the implementation, and verifies the result against the original intent.
03 / 04
Expectation-aware incident response
Product vision
Before a change ships, Max records what the team expects to remain true: affected systems, acceptable behavior, risk assumptions, rollback conditions, and required signals. If production later diverges, Max does not merely summarize alerts—it identifies the first expectation that became false, traces which decisions depended on it, recommends a pre-approved recovery path, verifies recovery, and updates the organization’s model with what reality disproved.
Max assists the on-call engineer; it does not replace incident ownership. Recovery remains supervised and subject to the team’s approvals.
04 / 04
After the outcome
Product vision
Max verifies what actually changed and whether the intended result occurred. Expert corrections become durable company knowledge, improving the next prediction.
Proposed change
Increase session lifetime from 24 hours to 7 days
- Auth ServiceFact
- Partner APIFact
- PaymentsInference
- ADR-12Fact
- INC-88Fact
- SecurityFact
- Token revocation testFact
- External partner cachingUnknown
Assurance Case
Builds as work progresses.
- Intent
Session lifetime 24h to 7d - Evidence
ADR-12 · INC-88 · PR-418 - Predicted consequences
Auth, Partner API, Payments - Unknowns
External partner caching - Required reviewers
Security - Actual scope
····· - Incident link
····· - Verification
····· - Outcome
····· - Corrections
·····
Illustrative product view · THUS is currently in technical alpha
A company brain answers questions. THUS stays through the action.
What the person or agent is trying to change, and why.
THUS preserves the reasoning around consequential work across people, agents, and tools. It records what was intended, what evidence existed, what was predicted, what was allowed, what actually happened, and what the organization learned.
Assurance Case
The durable record of intent, evidence, authority, predictions, actual work, corrections, and outcome.
Intent
Increase session lifetime from 24 hours to 7 days.
Select a field to see an illustrative entry. Illustrative product view · THUS is currently in technical alpha
Action Envelope
Product visionThe approved scope, exclusions, tests, reviewers, access, and escalation rules that stay with the work across tools and agents.
Allowed scope
Auth Service session settings; Partner API token validation.
Select a field to see an illustrative entry. Illustrative product view · THUS is currently in technical alpha
Early deployments are advisory and supervised. THUS is not enforcing production access today.
THE SYSTEM UNDER MAX
Organizational memory built from evidence and outcomes—not another chat history.
THUS builds a temporal, evidence-backed model of systems, people, ownership, policies, decisions, history, and outcomes. Every important assertion keeps its source, time, confidence, and status. Unknowns remain visible instead of being guessed away.
What exists, how it connects, who owns it, and how it changed.
What an intended action may affect or require.
Scope, authority, approvals, escalation, and evidence.
Max, Companion, pull requests, chat, APIs, and agent integrations.
Predicted versus actual outcomes and expert corrections.
One assertion, fully attributed
Illustrative record
- Claim
- Payments consumes Auth session tokens
- Source
- Release history, 7 related releases
- Time
- Observed across the last two quarters
- Confidence
- Medium
- Status
- Inference · needs verification
Max is who the team works with. THUS Core is what lets Max understand and learn the company.
Start where missing context already costs time.
01
Change assurance
Before a code, infrastructure, schema, pipeline, API, or configuration change, identify affected systems, owners, tests, policies, historical failures, and unresolved unknowns.
What Max returns · illustrative
- Affected systems and owners
- Required tests and reviewers
- Policy and incident history
- Unresolved unknowns
02
Incident investigation
When production breaks, reconstruct the relevant context faster and preserve the investigation so the next responder does not begin from zero.
What Max returns · illustrative
- Alert linked to recent changes
- Likely dependency path
- Prior incidents and runbooks
- A preserved investigation trail
03
AI-agent oversight
Give Cursor, Claude Code, Codex, and future execution agents the organization-specific context and boundaries their local task view is missing—then independently verify the result.
What Max returns · illustrative
- Organization context for the agent
- Boundaries for the task
- Scope-expansion flags
- Independent verification
Designed to connect with
- GitHub
- GitLab
- Jira
- Linear
- Slack
- Teams
- Datadog
- PagerDuty
- Documentation
- CI/CD
- Data catalogs
- Coding agents
- APIs
- MCP
Potential connection points, not current integrations. Which ones exist is agreed per design partner.
THE LONG-TERM COMPANY
Different work. The same organizational problem.
Every sector has consequential work whose real rules are scattered across systems, procedures, people, and history. THUS is building one shared organizational core, beginning with engineering.
Max
Building first
Engineering and data systems
Code, infrastructure, pipelines, schemas, deployments, incidents, and agent-generated changes.
Jack
Future capability
Clinical operations
Authorizations, documentation, billing, claims, compliance, scheduling, and operational workflows.
Roadmap · not available
Queen
Future capability
Marketing operations
Customer data, campaigns, experiments, audiences, offers, partnerships, budgets, and outcomes.
Roadmap · not available
Shared THUS Core
Companies can produce more changes than senior teams can confidently evaluate.
Coding agents are increasing implementation speed and volume. The bottleneck moves to organizational context, review, incident response, evidence, and trust. THUS is being built for that bottleneck.
Implementation capacity
rising quickly
Organizational understanding
fragmented
Review and incident load
compounding
Qualitative illustration. Not measured data.
FOUNDING DESIGN PARTNERS
One workflow. Four to six weeks. Let the evidence decide.
We are working with a small number of engineering and data teams to test whether Max materially improves one recurring, consequential workflow. Engagements begin read-only and in shadow mode alongside the team’s existing tools.
- W1
Map
- W2
Replay
- W3
Shadow
- W4–6
Findings
We look for
- A real recurring change or incident workflow.
- A technical owner who can work with us weekly.
- A small, agreed set of historical artifacts.
- One upcoming live change or incident to shadow when practical.
- Honest feedback and permission to measure the current process against THUS.
We deliver
- An initial map of the relevant systems, owners, and evidence.
- A Max alpha for consequence prediction or investigation.
- A historical replay and live shadow evaluation.
- A measured findings report and continue/revise/stop recommendation.
Explore a design partnership
No rip-and-replace. No production control during the initial evaluation.
Tell us the workflow.
Fill in the form to prepare an email, then review it and press Send in your email app. Prefer to write directly? Email us below.
connect@suryanediyadeth.comBuilt by engineers who met the same problem in different systems.
Surya Nediyadeth
Co-founder and CEO
Builds data models, integrations, automation, CI/CD, and governed workflows across fragmented clinical and administrative systems.
Portfolio
Sanket Shah
Co-founder and CTO
Builds customer-data systems, propensity and lookalike modeling, segmentation, measurement, and data engineering.
LinkedInWe have known each other since 2022 and have studied, worked, and built together. In clinical and marketing systems, we kept finding the same underlying problem: the software contains only part of the truth about how the organization actually works.
Start with the minimum access required.
- 01
Read-only onboarding.
- 02
Least-privilege access.
- 03
Evidence provenance.
- 04
Explicit unknowns and confidence.
- 05
Tenant-isolated memory.
- 06
Human review for consequential work.
- 07
Local screen processing where practical.
- 08
No employee-surveillance positioning.
THUS is designed to evaluate work and consequence—not employee productivity.
These are design principles for the technical alpha. THUS does not currently claim security certifications.
Before the next consequential change, know what follows.
THUS
Know what follows.