Where we work
- AI workflow systems
- Legacy service upgrades
- Operator and executive dashboards
Overview
What this page should tell you quickly
What we set up first
- Clear telemetry and core metrics
- Service contracts for legacy systems
- Safe access into business tools
What clients get
- Better executive visibility
- Faster delivery decisions
- A cleaner path to production
Main Products
The Sherlock product set is the main product layer
These are the named Baker Street products that package the method itself: case shaping, repo intelligence, evals, rescue, and adversarial hardening. The partner tools below support delivery, but this is the core product set clients actually buy into.
221B
Open-source hub
The operating layer for fast software investigation.
221B gives Baker Street investigations a shared operating system: repo intake, evidence grading, reusable templates, and MCP access to the underlying knowledge base.
Primary outcome: Faster orientation in unfamiliar systems with clearer evidence and handoff
- MCP-enabled knowledge base
- Repo registry and intake workflow
- Report and decision templates
Adler
Open-source investigation toolkit
Turn an ambiguous AI opportunity into a scoped next move.
Adler packages the Baker Street investigation sprint into reusable templates, example outputs, and executive-ready decision support for teams that need clarity before build work starts.
Primary outcome: A ranked recommendation and implementation brief in days, not weeks
- Hypothesis backlog
- Executive readout templates
- Worked customer-support case
Mycroft
Open-source eval harness
Create an evidence loop before AI changes hit production.
Mycroft is the evaluation and release-readiness layer for AI workflows, combining offline and model-backed evals with richer keyword, latency, and report outputs.
Primary outcome: Objective pass/fail signals before a workflow ships or relaunches
- Offline and OpenAI modes
- Keyword and latency gates
- Markdown and JSON reports
Watson
Open-source rescue playbook
Recover the AI pilot before it turns into an account problem.
Watson is a practical rescue system for stalled pilots, combining triage, stabilization, relaunch gates, KPI recovery tracking, and production handoff assets.
Primary outcome: A credible decision to relaunch, rescope, or stop with evidence
- Failure-mode triage
- KPI recovery scorecard
- Production handoff templates
Moriarty
Open-source red-team starter kit
Pressure-test the workflow before the workflow embarrasses you.
Moriarty is the red-team and failure-seeking layer in the Sherlock product set, built for prompt injection, policy evasion, tool misuse, and remediation verification work.
Primary outcome: A clearer view of exploit paths, severity, and whether fixes actually worked
- Scenario packs
- Rules-of-engagement checklists
- Worked support-assistant red-team example
Core Stack
The infrastructure and access layers that support the products
These are the foundations we reach for early because they make the Sherlock products easier to instrument, integrate, and govern in real delivery.
PostHog
Our default for analytics and product telemetry, so ignored workflow metrics become visible early.
- Track cycle time, delay, and drop-off
- See how operators actually use the workflow
- Roll out new features with better evidence
OpenAPI
Our standard way to modernize legacy services and make them usable by new products, agents, and internal tools.
- Document older systems properly
- Create stable contracts for integration work
- Reduce friction between teams and tools
MCP
Our preferred pattern for giving AI systems and teams structured access to core business tools.
- Connect workflows to real systems safely
- Set clearer boundaries for agents and reviewers
- Make executive and operator access easier to govern
Delivery Tools
The supporting tools we use to move faster
Once the core stack is clear, these tools help us tighten testing, Python setup, dashboard quality, and product pricing without adding unnecessary complexity.
Blacksmith
Speeds up CI runs so builds, tests, and container changes do not slow delivery down.
- Faster GitHub Actions runs
- Less time lost to queues and caches
- Quicker feedback during active sprints
Astral
Improves Python setup, linting, and environment consistency across development and CI.
- Cleaner Python environment setup
- Faster local development loops
- More reliable backend and eval workflows
TanStack
Powers data-heavy front-end products such as review queues, reporting views, and operator dashboards.
- Stronger query and caching patterns
- Better tables and operational data views
- A more robust base for internal tools
Polar
Useful when a workflow becomes a product and needs pricing, subscriptions, or usage-based billing.
- Supports packaged pricing models
- Helps with usage-based billing setup
- Makes product monetization less bespoke
Next Step
Need to modernize an operational workflow or legacy service?
Bring the workflow, the tooling constraint, and the outcome you need to prove. We can shape the first sprint around analytics, access, and one production-grade slice of the system.